100% found this document useful (1 vote)
544 views

Web Tech

WebTech
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (1 vote)
544 views

Web Tech

WebTech
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2698

Web Technologies:

Concepts, Methodologies,
Tools, and Applications

Arthur Tatnall
Victoria University, Australia

Volume I

Information science reference


Hershey • New York
Director of Editorial Content: Kristin Klinger
Development Editor Julia Mosemann
Senior Managing Editor: Jamie Snavely
Managing Editor: Michael Brehm
Assistant Managing Editor: Carole Coulson
Typesetters: Michael Brehm, Carole Coulson, Elizabeth Duke, Christopher Hrobak, Jamie Snavely, Sean Woznicki
Cover Design: Lisa Tosheff
Printed at: Yurchak Printing Inc.

Published in the United States of America by


Information Science Reference (an imprint of IGI Global)
701 E. Chocolate Avenue
Hershey PA 17033
Tel: 717-533-8845
Fax: 717-533-8661
E-mail: [email protected]
Web site: http://www.igi-global.com/reference

and in the United Kingdom by


Information Science Reference (an imprint of IGI Global)
3 Henrietta Street
Covent Garden
London WC2E 8LU
Tel: 44 20 7240 0856
Fax: 44 20 7379 0609
Web site: http://www.eurospanbookstore.com

Copyright © 2010 by IGI Global. All rights reserved. No part of this publication may be reproduced, stored or distributed in any form or by
any means, electronic or mechanical, including photocopying, without written permission from the publisher.
Product or company names used in this set are for identification purposes only. Inclusion of the names of the products or companies
does not indicate a claim of ownership by IGI Global of the trademark or registered trademark.

Library of Congress Cataloging-in-Publication Data

Web technologies : concepts, methodologies, tools and applications / Arthur


Tatnall, editor.
p. cm.
Includes bibliographical references and index.
ISBN 978-1-60566-982-3 (hbk.) -- ISBN 978-1-60566-983-0 (ebook) 1. World
Wide Web. 2. Internet. 3. Information technology. I. Tatnall, Arthur.
TK5105.888.W377 2010
004.6'7--dc22
2009037778

British Cataloguing in Publication Data


A Cataloguing in Publication record for this book is available from the British Library.

All work contributed to this book set is original material. The views expressed in this book are those of the authors, but not necessarily of
the publisher.
Editor-in-Chief
Mehdi Khosrow-Pour, DBA
Editor-in-Chief
Contemporary Research in Information Science and Technology, Book Series

Associate Editors
Steve Clarke
University of Hull, UK

Murray E. Jennex
San Diego State University, USA

Annie Becker
Florida Institute of Technology USA

Ari-Veikko Anttiroiko
University of Tampere, Finland

Editorial Advisory Board


Sherif Kamel
American University in Cairo, Egypt

In Lee
Western Illinois University, USA

Jerzy Kisielnicki
Warsaw University, Poland

Keng Siau
University of Nebraska-Lincoln, USA

Amar Gupta
Arizona University, USA

Craig van Slyke


University of Central Florida, USA

John Wang
Montclair State University, USA

Vishanth Weerakkody
Brunel University, UK
Additional Research Collections found in the
“Contemporary Research in Information Science and Technology”
Book Series
Data Mining and Warehousing: Concepts, Methodologies, Tools, and Applications
John Wang, Montclair University, USA • 6-volume set • ISBN 978-1-60566-056-1

Electronic Business: Concepts, Methodologies, Tools, and Applications


In Lee, Western Illinois University • 4-volume set • ISBN 978-1-59904-943-4

Electronic Commerce: Concepts, Methodologies, Tools, and Applications


S. Ann Becker, Florida Institute of Technology, USA • 4-volume set • ISBN 978-1-59904-943-4

Electronic Government: Concepts, Methodologies, Tools, and Applications


Ari-Veikko Anttiroiko, University of Tampere, Finland • 6-volume set • ISBN 978-1-59904-947-2

Knowledge Management: Concepts, Methodologies, Tools, and Applications


Murray E. Jennex, San Diego State University, USA • 6-volume set • ISBN 978-1-59904-933-5

Information Communication Technologies: Concepts, Methodologies, Tools, and Applications


Craig Van Slyke, University of Central Florida, USA • 6-volume set • ISBN 978-1-59904-949-6

Intelligent Information Technologies: Concepts, Methodologies, Tools, and Applications


Vijayan Sugumaran, Oakland University, USA • 4-volume set • ISBN 978-1-59904-941-0

Information Security and Ethics: Concepts, Methodologies, Tools, and Applications


Hamid Nemati, The University of North Carolina at Greensboro, USA • 6-volume set • ISBN 978-1-59904-937-3

Medical Informatics: Concepts, Methodologies, Tools, and Applications


Joseph Tan, Wayne State University, USA • 4-volume set • ISBN 978-1-60566-050-9

Mobile Computing: Concepts, Methodologies, Tools, and Applications


David Taniar, Monash University, Australia • 6-volume set • ISBN 978-1-60566-054-7

Multimedia Technologies: Concepts, Methodologies, Tools, and Applications


Syed Mahbubur Rahman, Minnesota State University, Mankato, USA • 3-volume set • ISBN 978-1-60566-054-7

Virtual Technologies: Concepts, Methodologies, Tools, and Applications


Jerzy Kisielnicki, Warsaw University, Poland • 3-volume set • ISBN 978-1-59904-955-7

Free institution-wide online access with the purchase of a print collection!

Information Science reference


Hershey • New York
Order online at www.igi-global.com or call 717-533-8845 ext.100
Mon–Fri 8:30am–5:00 pm (est) or fax 24 hours a day 717-533-7115
List of Contributors

Abdelmoty, Alia I. \ Cardiff University UK & University of Glamorgan, UK................................... 648


Abramowicz, Witold \ Poznań University of Economics, Poland................................................... 2082
Abu-Samaha, Ala M. \ Amman University, Jordan............................................................................ 344
Al-Dabass, David \ Nottingham Trent University, UK....................................................................... 604
Ali, Radwan \ Kennesaw State University, USA................................................................................. 188
Allendoerfer, Kenneth \ FAA Human Factors Team - Atlantic City, USA....................................... 1312
Alonso, Fernando\ Universidad Politécnica de Madrid, Spain....................................................... 1663
Al-Salem, Lana S. \ SpecTec Ltd & MEP, Greece.............................................................................. 344
Altarawneh, Haroon \ Albalqa’ Applied University, Jordan............................................................ 1696
Amigó, Enrique \ Cabrera, ETSI Informática, UNED, Spain.......................................................... 1994
Amous, Ikram \ MIRACL, Tunisie..................................................................................................... 681
Argyrides, Paris \ University College London, UK............................................................................ 119
Arias, Jose J. Pazos \ University of Vigo, Spain............................................................................... 2460
Bacellar Leal Ferreira, Simone \ Universidade Federal do Estado do Rio de Janeiro, Brazil...... 2423
Bagui, Sikha \ The University of West Florida, USA.......................................................................... 848
Baral, Chitta \ Arizona State University, USA.................................................................................... 355
Barbry, Benjamin \ University of Sciences and Technologies of Lille, France............................... 1159
Barnes, Richard T. \ SunGard Higher Education, USA..................................................................... 219
Baruh, Lemi \ Kadir Has University, Turkey................................................................................... 2341
Basharat, Amna \ National University of Computer and Emerging Sciences, Pakistan................... 571
Beatty, James \ San Diego State University, USA.............................................................................. 277
Ben Djemaa, Raoudha \ MIRACL, Tunisie........................................................................................ 681
Ben Hamadou, Abdelmajid \ MIRACL, Tunisie................................................................................ 681
Benharref, Abdelghani \ Concordia University, Canada................................................................ 1745
Benjamins, Richard \ Intelligent Software Components, S.A., Spain.............................................. 1878
Bergman, Margo \ Northwest Health Services Research &
Development (HSR&D), USA......................................................................................................... 1814
Bernardi, Ansgar \ German Research Center for Artificial
Intelligence (DFKI) GmbH, Kaiserslautern, Germany.................................................................. 2279
Berrueta, Diego \ Fundación CTIC, Spain....................................................................................... 1090
Blanco, Noha Veiguela \ Artificial Neural Networks and Adaptative Systems Group,
Spain & University of Corunna, Spain........................................................................................... 2328
Bochmann, Gregor V. \ University of Ottawa, Canada................................................................... 2099
Boley, Harold \ National Research Council, Canada......................................................................... 623
Boonstra, Albert \ University of Groningen, The Netherlands........................................................ 1298
Booth, Danielle \ Pennsylvania State University, USA....................................................................... 145
Borsje, Jethro \ Erasmus University Rotterdam, The Netherlands.................................................... 503
Brambilla, Marco \ Politecnico di Milano, Italy................................................................................ 541
Brenner, Daniel \ University of Mannheim, Germany...................................................................... 2403
Brown, Susan \ University of Manchester, UK................................................................................. 1141
Bryan, Martin \ CSW Group Ltd., UK............................................................................................... 924
Bryant, Samantha C. \ Philip Morris, USA....................................................................................... 909
Bults, Richard \ University of Twente, The Netherlands & Mobihealth B.V.,, The Netherlands..... 1515
Cáceres, Jesús \ University of Alcalá, Spain....................................................................................... 759
Camprubí, Raquel \ Universitat de Girona, Spain.......................................................................... 1050
Carriço, Luís \ LaSIGE, University of Lisbon, Portugal.......................................................... 522, 1437
Castellano, G. \ University of Bari, Italy.......................................................................................... 2381
Caus, Thorsten \ Georg August University of Göttingen, Germany.................................................... 11
Chang, Cynthia \ Stanford University, KSL, USA............................................................................ 2304
Chang, Mei-Yu \ National Hsinchu University of Education, Taiwan............................................... 379
Charlet, Jean \ STIM, DPA/AP-Hopitaux Paris & Université
Paris 6, France................................................................................................................................. 880
Chen, Hong \ Altran Netherlands B.V., The Netherlands................................................................. 1515
Cheng, Jingwei \ Northeastern University, China................................................................................ 63
Christmann, Stefan \ Georg August University of Göttingen, Germany............................................. 11
Chutimaskul, Wichian \ King Mongkut’s University of Technology Thonburi, Thailand............... 1328
Cicurel, Laurent \ iSOCO, Spain....................................................................................................... 971
Contreras, Jesús \ iSOCO, Spain....................................................................................................... 971
Cooper, Vanessa \ RMIT University, Australia................................................................................. 1069
Corazza, Laura \ Università di Bologna, Italy....................................................................................... 1
Corcho, Oscar \ Universidad Politécnica de Madrid, Spain............................................................ 1878
Cordier, Marie-Odile \ IRISA, France............................................................................................. 1970
Costa, Mário Figueiredo \ Brazilian School of Public and Business
Administration of Getulio Vargas Foundation, Brazil.................................................................... 1724
Costa, Paulo Cesar G. \ George Mason University, USA................................................................ 1852
Cousins, Jay \ CSW Group Ltd., UK................................................................................................... 924
Cuevas, I. Aedo \ Universidad Carlos III de Madrid, Spain............................................................ 1374
Cyr, Dianne \ Simon Fraser University, Canada................................................................................ 245
da Silva, Paulo Pinheiro \ University of Texas at El Paso (UTEP), USA........................................ 2304
da Silveira, Denis Silva \ Programa de Engenharia de Produção - COPPE/UFRJ, Brazil............ 2423
Dague, Philippe \ University Paris-Sud 11, France......................................................................... 1970
Damljanović, Danica \ University of Sheffield, UK.......................................................................... 1027
Daniel, Florian \ University of Trento, Italy....................................................................................... 417
De Angeli, A. \ University of Manchester, UK.................................................................................. 1374
de la Calle, Julián Dorado \ University of A Coruña, Spain........................................................... 2328
de Manuel, Francisco Javier Novóa \ Center of Medical Informatics
and Radiological Diagnosis, Spain & University of Corunna, Spain............................................ 2328
de Oliveira, José Palazzo M. \ Universidade Federal do Rio Grande do Sul (UFRGS), Brazil..... 1388
de Pablos, Patricia Ordóñez \ University of Oviedo, Spain............................................................ 1090
Decker, Stefan \ National University of Ireland, Ireland.................................................................. 2279
DeSchryver, Michael \ Michigan State University, USA.................................................................. 2563
Detlor, Brian \ McMaster University, Canada...................................................................................... 89
Devedžić, Vladan \ University of Belgrade, Serbia.......................................................................... 1027
Devi, M. Indra \ Thiagarajar College of Engineering, India........................................................... 1462
Di Martino, Sergio \ Università di Salerno, Italy & Università degli Studi di Napoli
“Federico II”, Italy........................................................................................................................ 2600
Dietze, Stefan \ Open University, UK............................................................................................... 1834
Ding, Li \ Tetherless World Constellation, Rensselaer Polytechnic Institute (RPI),
and Stanford University, KSL, USA................................................................................................ 2304
Domingue, John \ Open University, UK........................................................................................... 1834
Dssouli, Rachida \ Concordia University, Canada........................................................................... 1745
Ebner, Hannes \ Royal Institute of Technology (KTH), Sweden......................................................... 759
El-Geresy, Baher A. \ Cardiff University, UK & University of Glamorgan, UK............................... 648
Elia, Gianluca \ University of Salento, Italy....................................................................................... 805
El-Khatib, Khalil \ University of Ontario Institute of Technology, Canada.................................... 2099
El-Saddik, Abdulmotaleb \ University of Ottawa, Canada............................................................. 2099
El-Shiekh, Asim \ The Arab Academy for Banking and Financial Sciences, Jordan....................... 1696
Erickson, John \ University of Nebraska - Omaha, USA................................................................. 1786
Facca, Federico M. \ Leopold-Franzens-Universität Innsbruck, Austria.......................................... 541
Falbo, Ricardo de Almeida \ Universidade Federal do Espírito Santo, Brazil................................. 310
Fanelli, A. M. \ University of Bari, Italy........................................................................................... 2381
Fernández, José Antonio Seoane \ Artificial Neural Networks
and Adaptative Systems Group, Spain & University of Corunna, Spain........................................ 2328
Fernández, Sergio \ Fundación CTIC, Spain................................................................................... 1090
Fernández-Cavia, José \ Universitat Pompeu Fabra, Spain........................................................... 1274
Ferreira, Marcos Gurgel do Amaral Leal \ Holden Comunicação Ltda, Brazil........................... 2423
Ferri, Fernando \ IRPPS-CNR, Rome, Italy..................................................................................... 1345
Ferrucci, Filomena \ Università di Salerno, Italy............................................................................ 2600
Fioreze, Tiago \ Federal University of Rio Grande do Sul, Porto Alegre, Brazil............................. 1768
Firat, Aykut \ Northeastern University, USA.................................................................................... 2582
Frasincar, Flavius \ Erasmus University Rotterdam, The Netherlands.............................................. 503
Friedman-Berg, Ferne \ FAA Human Factors Team - Atlantic City, USA....................................... 1312
Frigerio, Chiara \ Università Cattolica del Sacro Cuore, Italy......................................................... 988
Furtado, Vasco \ University of Fortaleza, UNIFOR, Brazil............................................................. 2304
García, Roberto \ Universitat de Lleida, Spain................................................................................ 1003
Garlatti, Serge \ TELECOM, France............................................................................................... 1159
Gašević, Dragan \ Athabasca University, Canada................................................................... 478, 2175
Geleijnse, Gijs \ Philips Research, The Netherlands........................................................................ 2048
Geppert, L. \ Catholic University of Milan, Italy............................................................................... 441
Gibson, Susan \ University of Alberta, Canada................................................................................ 1554
Gil, Rosa \ Universitat de Lleida, Spain............................................................................................ 1003
Gimeno, Juan Manuel \ Universitat de Lleida, Spain..................................................................... 1003
Giurca, Adrian \ Brandenburgische Technische Universität at Cottbus, Germany........................... 478
Glass, Alyssa \ Stanford University, KSL, USA................................................................................. 2304
Glover, Derek \ Keele University, UK................................................................................................. 774
Goasdoué, François \ LRI, CNRS et Université Paris Sud XI, France.............................................. 880
Gong, Tao \ Donghua University, China & Central South University, China.................................. 2152
Gonzales, Dagmar Villarroel \ Agency for Quality in Medicine (AquMed), Germany................... 1994
Gonzalez, Sergio Bellido \ Bankinter, Spain...................................................................................... 971
Grady, Jonathan P. \ University of Pittsburgh, USA.......................................................................... 291
Granville, Lisandro Zambenedetti \ Federal University of Rio Grande do Sul–
Porto Alegre, Brazil........................................................................................................................ 1768
Gravino, Carmine \ Università di Salerno, Italy............................................................................. 2600
Grayson, James \ Augusta State University, USA............................................................................ 1814
Grifoni, Patrizia \ IRPPS-CNR, Rome, Italy.................................................................................... 1345
Grimnes, Gunnar Aastrand \ German Research Center for Artificial
Intelligence (DFKI) GmbH, Kaiserslautern, Germany.................................................................. 2279
Groza, Tudor \ National University of Ireland, Ireland................................................................... 2279
Gugliotta, Alessio \ Open University, UK......................................................................................... 1834
Guizzardi, Giancarlo \ Universidade Federal do Espírito Santo, Brazil.......................................... 310
Guzzo, Tiziana \ IRPPS-CNR, Rome, Italy....................................................................................... 1345
Hadjiefthymiades, Stathes \ National and Kapodistrian University of Athens, Greece................... 456
Hagenhoff, Svenja \ Georg August University of Göttingen, Germany............................................... 11
Handschuh, Siegfried \ National University of Ireland, Ireland..................................................... 2279
Hansmann, Uwe \ IBM Germany Research and Development, Germany......................................... 394
Harison, Elad \ University of Groningen, The Netherlands............................................................. 1298
Harth, Andreas \ National University of Ireland, Ireland................................................................ 2206
Hatala, Marek \ Simon Fraser University, Canada......................................................................... 2175
Hawkey, Kirstie \ University of British Columbia, Canada............................................................. 1951
Head, Milena \ McMaster University, Canada................................................................................... 245
Heinecke, Johannes \ France Telecom R&D, France........................................................................ 880
Hernandez, Victor de Vega \ ONO, Spain....................................................................................... 1708
Hobson, Paola \ Motorola Labs, UK.................................................................................................. 880
Hogan, Aidan \ National University of Ireland, Ireland................................................................... 2206
Holewa, Hamish \ International Program of Psycho-Social Health Research,
Central Queensland University, Australia........................................................................................ 670
Huertas-Roig, Assumpció \ Universitat Rovira i Virgili, Spain....................................................... 1274
Hupfer, Maureen \ McMaster University, Canada.............................................................................. 89
Ivanov, Alex \ Simon Fraser University, Canada............................................................................... 245
Jansen, Bernard J. \ Pennsylvania State University, USA................................................................. 145
Jatowt, Adam \ Kyoto University, Japan.......................................................................................... 2544
Jazayeri, Mehdi \ University of Lugano, Switzerland...................................................................... 2279
Jennex, Murray E. \ San Diego State University, USA...................................................................... 277
Jiang, Yabing \ Fordham University, USA........................................................................................ 2479
Joia, Luiz Antonio \ Brazilian School of Public and Business Administration
of Getulio Vargas Foundation and Rio de Janeiro State University, Brazil................................... 1724
Jones, Christopher B. \ Cardiff University UK, & University of Glamorgan, UK............................ 648
Juan, Javier Piqueres \ Systar, Spain............................................................................................... 1708
Kaczmarek, Monika \ Poznań University of Economics, Poland................................................... 2082
Kamthan, Pankaj \ Concordia University, Canada......................................................................... 2250
Karacapilidis, Nikos I. \ University of Patras, Greece...................................................................... 724
Karali, Isambo \ National and Kapodistrian University of Athens, Greece....................................... 456
Karampiperis, Pythagoras \ National Center of Scientific Research “Demokritos”, Greece........ 1994
Kardan, Ahmad \ Amirkabir University of Technology, Iran.......................................................... 2353
Karkaletsis, Vangelis \ National Center of Scientific Research “Demokritos”, Greece.................. 1994
Karvalics, Laszlo Z. \ University of Szeged, Hungary..................................................................... 2530
Kastrantas, Kostas \ Greek Research & Technology Network (GRNET S.A.), Greece...................... 759
Kaviani, Nima \ University of British Columbia, Canada............................................................... 2175
Kawai, Yukiko \ Kyoto Sangyo University, Japan............................................................................ 2544
Kazepov, Yuri \ University of Urbino “Carlo Bo,” Italy................................................................. 1642
Kelly, Diana K. \ San Diego Miramar College, USA....................................................................... 1578
Kelly, Mike \ ICT Consultant, Canada............................................................................................. 1536
Kennell, Trudy \ ICT Consultant, Canada....................................................................................... 1536
Kim, Yeon-Seok \ Yonsei University, South Korea............................................................................. 706
Kimes, Bryan \ Altria Client Services, Inc., USA............................................................................... 177
Kollmann, Tobias \ University of Duisburg-Essen, Campus Essen, Germany................................ 1799
Konstantas, Dimitri \ University of Geneva, Switzerland................................................................ 1515
Korst, Jan \ Philips Research, The Netherlands............................................................................... 2048
Kumar, Muneesh \ University of Delhi South Campus, India & ESC-PAU, France....................... 1195
Labra, Jose E. \ University of Oviedo, Spain................................................................................... 1090
Labský, Martin \ University of Economics, Prague, Czech Republic............................................. 1994
Lam, King Tin \ The University of Hong Kong, Hong Kong........................................................... 2436
Lappas, Georgios \ Technological Educational Institution of Western Macedonia,
Kastoria Campus, Greece............................................................................................................... 1495
Laskey, Kathryn Blackmond \ George Mason University, USA..................................................... 1852
Lau, B.-Y.-Simon \ Multimedia University, Malaysia...................................................................... 1159
Lawless, W.F. \ Paine College, USA................................................................................................. 1814
Lee, Dongwon \ The Pennsylvania State University, USA.................................................................. 739
Lee, Kyong-Ho \ Yonsei University, South Korea............................................................................... 706
Léger, Alain \ France Telecom R&D, France..................................................................................... 880
LeRouge, Cynthia \ Saint Louis University, USA............................................................................ 1360
Levering, Leonard \ Erasmus University Rotterdam, The Netherlands............................................ 503
Levialdi, S. \ Sapienza University of Rome, Italy.............................................................................. 1374
Li, Maozhen \ Brunel University, UK................................................................................................. 830
Li, Ning \ University of Surrey, UK....................................................................................................... 41
Li, Xin \ University of North Carolina at Pembroke, USA............................................................... 1206
Lichtenstein, Sharman \ Deakin University, Australia.................................................................... 1069
Likothanassis, Spiridon \ Pattern Recognition Laboratory, University of Patras, Greece............... 522
Lingras, Pawan \ Saint Mary’s University, Canada......................................................................... 2034
Lingras, Rucha \ Saint Mary’s University, Canada.......................................................................... 2034
Little, Joseph P. \ Saint Louis University, USA.................................................................................. 229
Lizcano, David\ Universidad Politécnica de Madrid, Spain............................................................ 1663
Loggins, Adam \ Zilliant Inc., USA.................................................................................................... 848
Lopes, Rui \ LaSIGE, University of Lisbon, Portugal.............................................................. 522, 1437
López, Genoveva\ Universidad Politécnica de Madrid, Spain........................................................ 1663
López-Cobo, José-Manuel \ iSOCO, Spain....................................................................................... 971
Losada, Silvestre \ iSOCO, Spain............................................................................................. 971, 1878
Loucky, John Paul \ Seinan Jogakuin University, Japan................................................................. 1411
Lukasiewicz, Thomas \ Oxford University Computing Laboratory, UK......................................... 1852
Lukichev, Sergey \ Brandenburg University of Technology at Cottbus, Germany............................ 478
Luo, Xiangfeng \ Shanghai University, P. R. China......................................................................... 2518
Ma, Jiao \ Saint Louis University, USA............................................................................................. 1360
Ma, Zongmin \ Northeastern University, China................................................................................... 63
Maamar, Zakaria \ Zayed University, UAE..................................................................................... 2134
Madnick, Stuart \ Massachusetts Institute of Technology, USA...................................................... 2582
Malaka, Rainer \ University of Bremen, Germany........................................................................... 2403
Malizia, A. \ Universidad Carlos III de Madrid, Spain.................................................................... 1374
Manouselis, Nikos \ Greek Research & Technology Network (GRNET S.A.), Greece....................... 759
Marchi, Simona \ University “Sapienza” of Rome, Italy................................................................. 1628
Markless, Sharon \ King’s College, London, UK............................................................................. 2115
Marshall-Bradley, Tina \ Paine College, USA................................................................................ 1814
Martin, Fernando Llorente \ ONO, Spain...................................................................................... 1708
Mayer, Miquel Angel \ Medical Association of Barcelona (COMB), Spain.................................... 1994
McBride, Rob \ ICT Consultant, Canada......................................................................................... 1536
McGuinness, Deborah L. \ Tetherless World Constellation, Rensselaer
Polytechnic Institute (RPI), and Stanford University, KSL, USA................................................... 2304
Memmola, Massimo \ Catholic University, Italy............................................................................... 943
Merdes, Matthias \ Heidelberg Mobil International GmbH, Germany........................................... 2403
Mesnage, Cédric \ University of Lugano, Switzerland..................................................................... 2279
Mikroyannidis, Alexander \ University of Manchester, UK................................................................ 78
Milanović, Milan \ University of Belgrade, Serbia............................................................................ 478
Miller, David \ Keele University, UK.................................................................................................. 774
Mills, Steven C. \ The University Center of Southern Oklahoma, USA........................................... 1478
Moessner, Klaus \ University of Surrey, UK......................................................................................... 41
Möller, Knud \ National University of Ireland, Ireland................................................................... 2279
Monclús-Guitart, Ricard \ Universitat Rovira i Virgili, Spain.......................................................... 794
Mostefaoui, Ghita Kouadri \ Oxford University Computing Laboratory, UK................................ 2134
Motteram, Gary \ University of Manchester, UK............................................................................ 1141
Naeve, Ambjorn \ Royal Institute of Technology (KTH), Sweden...................................................... 759
Narendra, Nanjangud C. \ IBM India Research Lab, India............................................................ 2134
Nauerz, Andreas \ IBM Research and Development, Germany......................................................... 404
Neisse, Ricardo \ Federal University of Rio Grande do Sul–Porto Alegre, Brazil.......................... 1768
Nicholas, David \ Ciber, University College London, UK.................................................................. 119
Nixon, Lyndon J.B. \ Freie Universität Berlin, Germany.................................................................. 880
Nunes, Ricardo Rodrigues \ Universidade Federal do Estado do Rio de Janeiro, Brazil.............. 2423
Oguz, Fatih \ Valdosta State University, USA................................................................................... 1127
Oh, Seog-Chan \ General Motors R&D Center, USA........................................................................ 739
Okazaki, Shintaro \ Universidad Autónoma de Madrid, Spain....................................................... 1228
Oliva, Marta \ Universitat de Lleida, Spain..................................................................................... 1003
Ordóñez, Juan Luis Pérez \ Center of Medical Informatics
and Radiological Diagnosis, Spain & University of Corunna, Spain............................................ 2328
Osman, Taha \ Nottingham Trent University, UK............................................................................... 604
Paech, Barbara \ University of Heidelberg, Germany..................................................................... 2403
Page, G. Andrew \ The University of Alaska Anchorage, USA.......................................................... 188
Pai, Shantanu \ Engility Corporation, USA..................................................................................... 1312
Palmer, Matthias \ Royal Institute of Technology (KTH), Sweden.................................................... 759
Papasratorn, Borworn \ King Mongkut’s University of Technology Thonburi, Thailand............... 1328
Papataxiarhis, Vassilis \ National and Kapodistrian University of Athens, Greece.......................... 456
Pappis, Costas P. \ University of Piraeus, Greece.............................................................................. 724
Park, Myung-Woo \ Yonsei University, South Korea......................................................................... 706
Paschke, Adrian \ Freie Universität Berlin, Germany....................................................................... 623
Pegrum, Mark \ University of Western Australia, Australia............................................................ 1105
Pencolé, Yannick \ LAAS-CNRS, France.......................................................................................... 1970
Perdrix, Ferran \ Universitat de Lleida, Spain & Diari Segre Media Group, Spain....................... 1003
Petrucco, Corrado \ University of Padua, Italy............................................................................... 1570
Pham-Nguyen, Cuong \ TELECOM, France................................................................................... 1159
Polgar, Jana \ NextDigital, Australia................................................................................................ 2015
Pöllä, Matti \ Helsinki University of Technology, Finland............................................................... 1994
Polleres, Axel \ National University of Ireland, Ireland................................................................... 2206
Pontelli, Enrico \ New Mexico State University, USA........................................................................ 355
Povalej, Roman \ University of Karlsruhe (TH), Germany................................................................ 167
Prats-Planagumà, Lluís \ Universitat de Girona, Spain.................................................................. 1050
Predoiu, Livia \ University of Mannheim, Germany........................................................................ 1896
Qi, Man \ Canterbury Christ Church University, UK......................................................................... 830
Raghupathi, Viju \ City University of New York, USA..................................................................... 2479
Raghupathi, Wullianallur \ Fordham University, USA................................................................... 2479
Raith, Thomas \ The University of Education Heidelberg, Germany.............................................. 1596
Rajaram, R. \ Thiagarajar College of Engineering, India............................................................... 1462
Ratnasingam, Pauline \ University of Central Missouri, USA.......................................................... 865
Reif, Gerald \ University of Lugano, Switzerland............................................................................ 2279
Reyes, Marcos\ Telefónica I+D, Spain............................................................................................. 1663
Ribarić, Marko \ Mihailo Pupin Institute, Serbia.............................................................................. 478
Rigo, Sandro José \ Universidade Federal do Rio Grande do Sul (UFRGS), Brazil...................... 1388
Rodríguez-Merayo, Araceli \ Universitat Rovira i Virgili, Spain...................................................... 794
Rong, Guang \ Clemson University, USA......................................................................................... 1206
Rotondi, Armando J. \ University of Pittsburgh, USA....................................................................... 291
Ruhi, Umar \ University of Ottawa, Canada........................................................................................ 89
Růžička, Marek \ University of Economics, Prague, Czech Republic............................................. 1994
Sahota, Vijay \ Brunel University, UK................................................................................................ 830
Salem, Mohamed \ University of Wollongong, Dubai, UAE............................................................ 1745
Sanchez-Alonso, Salvador \ University of Alcalá, Spain................................................................... 759
Sareen, Mamta \ University of Delhi, India..................................................................................... 1195
Sauermann, Leo \ German Research Center for Artificial
Intelligence (DFKI) GmbH, Germany............................................................................................ 2279
Schröer, Carsten \ University of Duisburg-Essen, Campus Essen, Germany.................................. 1799
Secundo, Giustina \ University of Salento, Italy................................................................................ 805
Segovia, Roy H. \ San Diego State University, USA........................................................................... 277
Selvakuberan, K. \ Tata Consultancy Services, India...................................................................... 1462
Serhani, Mohamed Adel \ United Arab Emirates University, UAE................................................ 1745
Sfakianakis, Stelios \ ICS-FORTH, Greece.......................................................................................... 22
Sheidaei, Shahin \ Simon Fraser University, Canada........................................................................ 478
Shen, Pei-Di \ Ming Chuan University, Taiwan...................................................................... 1181, 1614
Shi, Lian \ Fundación CTIC, Spain.................................................................................................. 1090
Shin, Fu-Yu \ Chien-Kuo Elementary School, Taiwan....................................................................... 379
Shvaiko, Pavel \ University of Trento, Italy........................................................................................ 880
Siau, Keng \ University of Nebraska - Lincoln, USA........................................................................ 1786
Sigala, Marianna \ University of the Aegean, Greece...................................................................... 1249
Simón-Olmos, M. José \ Universitat Rovira i Virgili, Spain.............................................................. 794
Singh, Nitish \ Saint Louis University, USA....................................................................................... 229
Sintek, Michael \ German Research Center for Artificial
Intelligence (DFKI) GmbH, Kaiserslautern, Germany.................................................................. 2279
Škapa, Radoslav \ Masaryk University Brno, Czech Republic........................................................ 1228
Smart, Philip D. \ Cardiff University UK, & University of Glamorgan, UK..................................... 648
Smith, Ross \ RMIT University, Australia......................................................................................... 1069
Sofge, Donald A. \ Naval Research Laboratory, USA...................................................................... 1814
Son, Tran Cao \ New Mexico State University, USA.......................................................................... 355
Soriano, Javier \ Universidad Politécnica de Madrid, Spain........................................................... 1663
Souza, Vítor Estêvão Silva \ Universidade Federal do Espírito Santo, Brazil................................. 310
Spillan, John E. \ University of North Carolina, USA........................................................................ 229
Spinelli, Gabriella \ Brunel University, UK....................................................................................... 571
Spiro, Rand J. \ Michigan State University, USA............................................................................. 2563
Spring, Michael B. \ University of Pittsburgh, USA........................................................................... 291
Stamatakis, Konstantinos \ National Center of Scientific Research “Demokritos”, ..................... 1994
Stamatopoulos, Panagiotis \ National and Kapodistrian University of Athens, Greece................... 456
Stephens, R. Todd \ AT&T Corporation Collaboration & Online Services, USA............................. 201
Stober, Thomas \ IBM Germany Research and Development, Germany........................................... 394
Stöckmann, Christoph \ University of Duisburg-Essen, Campus Essen, Germany........................ 1799
Streatfield, David \ Information Management Associates, UK........................................................ 2115
Stuckenschmidt, Heiner \ University of Mannheim, Germany........................................................ 1896
Sturm, Matthias \ ICT Consultant, Canada..................................................................................... 1536
Svátek, Vojtěch \ University of Economics, Prague, Czech Republic.............................................. 1994
Taghipour, Nima \ Amirkabir University of Technology, Iran......................................................... 2353
Tanaka, Katsumi \ Kyoto University, Japan.................................................................................... 2544
Tarng, Wernhuar \ National Hsinchu University of Education, Taiwan........................................... 379
Taurino, Cesare \ University of Salento, Italy.................................................................................... 805
Thakker, Dhavalkumar \ Nottingham Trent University, UK............................................................. 604
Thatcher, Jason B. \ Clemson University, USA................................................................................ 1206
Theodoulidis, Babis \ University of Manchester, UK........................................................................... 78
Thompson, Rich \ IBM T.J. Watson Research Center, USA............................................................... 404
Torres-Coronas, Teresa \ Universitat Rovira i Virgili, Spain............................................................ 794
Torrisi, Giovanni \ University of Urbino “Carlo Bo”, Italy............................................................ 1642
Torsello, M. A. \ University of Bari, Italy......................................................................................... 2381
Tsai, Chia-Wen \ Ming Chuan University, Taiwan........................................................................... 1181
Tsetsos, Vassileios \ National and Kapodistrian University of Athens, Greece.................................. 456
Tsoulfas, Giannis T. \ University of Piraeus, Greece......................................................................... 724
Tung, Hui-Lien \ Paine College, USA.............................................................................................. 1814
Tzannis, Alessandra \ Catholic University, Italy............................................................................... 943
Tzovaras, Dimitrios \ Informatics and Telematics Institute, CERTH, Greece................................... 522
Uribe, José Luis Bas \ Bankinter, Spain............................................................................................. 971
van Beijnum, Bert-Jan \ University of Twente, The Netherlands.................................................... 1515
van Elst, Ludger \ German Research Center for Artificial Intelligence (DFKI) GmbH,
Kaiserslautern, Germany................................................................................................................ 2279
Vantroys, Thomas \ University of Sciences and Technologies of Lille, France............................... 1159
Vianna, Ricardo Lemos \ Federal University of Rio Grande do Sul–Porto Alegre, Brazil............ 1768
Vidal-Blasco, M. Arántzazu \ Universitat Rovira i Virgili, Spain..................................................... 794
Vilas, Ana Fernández \ University of Vigo, Spain............................................................................ 2460
Vilas, Julio Fernández \ University of Vigo, Spain.......................................................................... 2460
Votis, Konstantinos \ Pattern Recognition Laboratory, University of Patras
& Informatics and Telematics Institute, CERTH, Greece................................................................. 522
Wac, Katarzyna \ University of Geneva, Switzerland, & University of Twente,
The Netherlands.............................................................................................................................. 1515
Wagner, Gerd \ Brandenburg University of Technology, Germany................................................... 478
Wang, Cho-Li \ The University of Hong Kong, Hong Kong............................................................ 2436
Wang, Hailong \ Northeastern University, China................................................................................ 63
Wangpipatwong, Sivaporn \ Bangkok University, Thailand........................................................... 1328
Weaver, Alfred C. \ University of Virginia, USA.............................................................................. 2498
Weiß, Peter \ University of Karlsruhe (TH), Germany....................................................................... 167
Welicki, Leon \ Microsoft, Canada................................................................................................... 1708
Wives, Leandro Krug \ Universidade Federal do Rio Grande do Sul (UFRGS), Brazil................ 1388
Wood, LTC, Joseph \ US Army, USA............................................................................................... 1814
Wu, Lynn \ Massachusetts Institute of Technology, USA................................................................. 2582
Wu, Zhengping \ University of Bridgeport, USA............................................................................. 2498
Xanthidis, Dimitrios \ Ciber, University College London, DEI College, & NYC College,
UK & Greece.................................................................................................................................... 119
Yan, Li \ Northeastern University, China.............................................................................................. 63
Yan, Yuhong \ Concordia University, Canada.................................................................................. 1970
Yee, George O.M. \ Institute for Information Technology, National Research Council,
Canada............................................................................................................................................ 1929
Yu, Bin \ Level E Limited, UK............................................................................................................. 830
Yu, Jie \ Shanghai University, P. R. China........................................................................................ 2518
Zhdanova, Anna V. \ The Telecommunications Research Center Vienna, Austria............................... 41
Zyskowski, Dominik \ Poznań University of Economics, Poland................................................... 2082
Contents

Volume I
Section I. Fundamental Concepts and Theories

This section serves as the foundation for this exhaustive reference tool by addressing crucial theories
essential to the understanding of Web technologies. Chapters found within these pages provide an
excellent framework in which to position Web technologies within the field of information science and
technology. Individual contributions provide overviews of the mobile Web, semantic Web, and Web 2.0,
while also exploring critical stumbling blocks of this field. Within this introductory section, the reader
can learn and choose from a compendium of expert research on the elemental theories underscoring
the research and application of Web technologies.

Chapter 1.1. ICT and Interculture Opportunities Offered by the Web..................................................... 1


Laura Corazza, Università di Bologna, Italy

Chapter 1.2. Mobile Social Web: Opportunities and Drawbacks.......................................................... 11


Thorsten Caus, Georg August University of Göttingen, Germany
Stefan Christmann, Georg August University of Göttingen, Germany
Svenja Hagenhoff, Georg August University of Göttingen, Germany

Chapter 1.3. Social Semantic Web and Semantic Web Services........................................................... 22


Stelios Sfakianakis, ICS-FORTH, Greece

Chapter 1.4. Semantic Web in Ubiquitous Mobile Communications.................................................... 41


Anna V. Zhdanova, The Telecommunications Research Center Vienna, Austria
Ning Li, University of Surrey, UK
Klaus Moessner, University of Surrey, UK
Chapter 1.5. A Review of Fuzzy Models for the Semantic Web........................................................... 63
Hailong Wang, Northeastern University, China
Zongmin Ma, Northeastern University, China
Li Yan, Northeastern University, China
Jingwei Cheng, Northeastern University, China

Chapter 1.6. Semantic Web Adaptation................................................................................................. 78


Alexander Mikroyannidis, University of Manchester, UK
Babis Theodoulidis, University of Manchester, UK

Chapter 1.7. Tips for Tracking Web Information Seeking Behavior..................................................... 89


Brian Detlor, McMaster University, Canada
Maureen Hupfer, McMaster University, Canada
Umar Ruhi, University of Ottawa, Canada

Chapter 1.8. A Proposed Template for the Evaluation of Web Design Strategies............................... 119
Dimitrios Xanthidis, Ciber, University College London, UK & DEI College,
& NYC College, Greece
David Nicholas, Ciber, University College London, UK
Paris Argyrides, University College London, UK

Chapter 1.9. A Review of Methodologies for Analyzing Websites..................................................... 145


Danielle Booth, Pennsylvania State University, USA
Bernard J. Jansen, Pennsylvania State University, USA

Chapter 1.10. Basics to Develop Web Services for Human Resources............................................... 167
Roman Povalej, University of Karlsruhe (TH), Germany
Peter Weiß, University of Karlsruhe (TH), Germany

Chapter 1.11. Web 2.0 and E-Discovery.............................................................................................. 177


Bryan Kimes, Altria Client Services, Inc., USA

Chapter 1.12. The Power and Promise of Web 2.0 Tools.................................................................... 188
G. Andrew Page, The University of Alaska Anchorage, USA
Radwan Ali, Kennesaw State University, USA

Chapter 1.13. Integrating Web 2.0 Technologies within the Enterprise.............................................. 201
R. Todd Stephens, AT&T Corporation Collaboration & Online Services, USA

Chapter 1.14. Security in a Web 2.0 World.......................................................................................... 219


Richard T. Barnes, SunGard Higher Education, USA

Chapter 1.15. Web Site Localization Practices: Some Insights into the Localization Industry........... 229
Nitish Singh, Saint Louis University, USA
John E. Spillan, University of North Carolina, USA
Joseph P. Little, Saint Louis University, USA
Section II. Development and Design Methodologies

This section provides in-depth coverage of conceptual architectures, frameworks and methodologies
related to the design and implementation of Web technologies. Throughout these contributions, re-
search fundamentals in the discipline are presented and discussed. From broad examinations to specific
discussions on particular frameworks and infrastructures, the research found within this section spans
the discipline while also offering detailed, specific discussions. Basic designs, as well as abstract de-
velopments, are explained within these chapters, and frameworks for designing successful Web sites,
Web-based applications, and Web portals are provided.

Chapter 2.1. Perceptions of Mobile Device Website Design: Culture,


Gender and Age Comparisons............................................................................................................. 245
Dianne Cyr, Simon Fraser University, Canada
Milena Head, McMaster University, Canada
Alex Ivanov, Simon Fraser University, Canada

Chapter 2.2. Paralingual Web Design and Trust in E-Government..................................................... 277


Roy H. Segovia, San Diego State University, USA
Murray E. Jennex, San Diego State University, USA
James Beatty, San Diego State University, USA

Chapter 2.3. Designing Medical Research Web Sites.......................................................................... 291


Jonathan P. Grady, University of Pittsburgh, USA
Michael B. Spring, University of Pittsburgh, USA
Armando J. Rotondi, University of Pittsburgh, USA

Chapter 2.4. Designing Web Information Systems for a Framework-Based Construction................. 310
Vítor Estêvão Silva Souza, Universidade Federal do Espírito Santo, Brazil
Ricardo de Almeida Falbo, Universidade Federal do Espírito Santo, Brazil
Giancarlo Guizzardi, Universidade Federal do Espírito Santo, Brazil

Chapter 2.5. Focused Requirements Engineering Method for Web Application Development.......... 344
Ala M. Abu-Samaha, Amman University, Jordan
Lana S. Al-Salem, SpecTec Ltd & MEP, Greece

Chapter 2.6. A Logic Programming Based Framework for Intelligent Web Service
Composition......................................................................................................................................... 355
Enrico Pontelli, New Mexico State University, USA
Tran Cao Son, New Mexico State University, USA
Chitta Baral, Arizona State University, USA

Chapter 2.7. The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System............. 379
Mei-Yu Chang, National Hsinchu University of Education, Taiwan
Wernhuar Tarng, National Hsinchu University of Education, Taiwan
Fu-Yu Shin, Chien-Kuo Elementary School, Taiwan
Chapter 2.8. WebSphere Portal 6.1: An Agile Development Approach.............................................. 394
Thomas Stober, IBM Germany Research and Development, Germany
Uwe Hansmann, IBM Germany Research and Development, Germany

Chapter 2.9. Adaptation and Recommendation in Modern Web 2.0 Portals....................................... 404
Andreas Nauerz, IBM Research and Development, Germany
Rich Thompson, IBM T.J. Watson Research Center, USA

Chapter 2.10. Context-Aware Applications for the Web: A Model-Driven


Development Approach....................................................................................................................... 417
Florian Daniel, University of Trento, Italy

Chapter 2.11. Different Web Strategies for Different E-Marketplaces................................................ 441


L. Geppert, Catholic University of Milan, Italy

Chapter 2.12. Developing Rule-Based Web Applications: Methodologies and Tools........................ 456
Vassilis Papataxiarhis, National and Kapodistrian University of Athens, Greece
Vassileios Tsetsos, National and Kapodistrian University of Athens, Greece
Isambo Karali, National and Kapodistrian University of Athens, Greece
Panagiotis Stamatopoulos, National and Kapodistrian University of Athens, Greece
Stathes Hadjiefthymiades, National and Kapodistrian University of Athens, Greece

Chapter 2.13. Modeling of Web Services using Reaction Rules......................................................... 478


Marko Ribarić, Mihailo Pupin Institute, Serbia
Shahin Sheidaei, Simon Fraser University, Canada
Milan Milanović, University of Belgrade, Serbia
Dragan Gašević, Athabasca University, Canada
Adrian Giurca, Brandenburgische Technische Universität at Cottbus, Germany
Sergey Lukichev, Brandenburg University of Technology at Cottbus, Germany
Gerd Wagner, Brandenburg University of Technology, Germany

Chapter 2.14. A Semantic Web-Based Approach for Building Personalized News Services.............. 503
Flavius Frasincar, Erasmus University Rotterdam, The Netherlands
Jethro Borsje, Erasmus University Rotterdam, The Netherlands
Leonard Levering, Erasmus University Rotterdam, The Netherlands

Chapter 2.15. A Service Oriented Ontological Framework for the Semantic Validation
of Web Accessibility............................................................................................................................ 522
Rui Lopes, LaSIGE, University of Lisbon, Portugal
Konstantinos Votis, Pattern Recognition Laboratory, University of Patras & Informatics
and Telematics Institute, CERTH, Greece
Luís Carriço, LaSIGE, University of Lisbon, Portugal
Spiridon Likothanassis, Pattern Recognition Laboratory, University of Patras, Greece
Dimitrios Tzovaras, Informatics and Telematics Institute, CERTH, Greece
Chapter 2.16. Building Semantic Web Portals with a Model-Driven Design Approach..................... 541
Marco Brambilla, Politecnico di Milano, Italy
Federico M. Facca, Leopold-Franzens-Universität Innsbruck, Austria

Chapter 2.17. Enabling Distributed Cognitive Collaborations on the Semantic Web......................... 571
Amna Basharat, National University of Computer and Emerging Sciences, Pakistan
Gabriella Spinelli, Brunel University, UK

Chapter 2.18. Utilisation of Case-Based Reasoning for Semantic Web Services Composition.......... 604
Taha Osman, Nottingham Trent University, UK
Dhavalkumar Thakker, Nottingham Trent University, UK
David Al-Dabass, Nottingham Trent University, UK

Volume II
Chapter 2.19. Rule Markup Languages and Semantic Web Rule Languages...................................... 623
Adrian Paschke, Freie Universität Berlin, Germany
Harold Boley, National Research Council, Canada

Chapter 2.20. Semantic Web Rule Languages for Geospatial Ontologies........................................... 648
Philip D. Smart, Cardiff University, UK & University of Glamorgan, UK
Alia I. Abdelmoty, Cardiff University, UK & University of Glamorgan, UK
Baher A. El-Geresy, Cardiff University, UK & University of Glamorgan, UK
Christopher B. Jones, Cardiff University, UK & University of Glamorgan, UK

Section III. Tools and Technologies

This section presents extensive coverage of the technology that informs and impacts Web technologies.
These chapters provide an in-depth analysis of the use and development of innumerable devices and
tools, while also providing insight into new and upcoming technologies, theories, and instruments that
will soon be commonplace. Within these rigorously researched chapters, readers are presented with
examples of the tools that facilitate and support the emergence and advancement of Web technologies.
In addition, the successful implementation and resulting impact of these various tools and technologies
are discussed within this collection of chapters.

Chapter 3.1. New Paradigms: A Collaborative Web Based Research Tool......................................... 670
Hamish Holewa, International Program of Psycho-Social Health Research,
Central Queensland University, Australia

Chapter 3.2. Adaptability and Adaptivity in The Generation of Web Applications............................. 681
Raoudha Ben Djemaa, MIRACL, Tunisie
Ikram Amous, MIRACL, Tunisie
Abdelmajid Ben Hamadou, MIRACL, Tunisie
Chapter 3.3. Migrating Web Services in Mobile and Wireless Environments.................................... 706
Myung-Woo Park, Yonsei University, South Korea
Yeon-Seok Kim, Yonsei University, South Korea
Kyong-Ho Lee, Yonsei University, South Korea

Chapter 3.4. Applying Web-Based Collaborative Decision-Making in Reverse Logistics:


The Case of Mobile Phones................................................................................................................. 724
Giannis T. Tsoulfas, University of Piraeus, Greece
Costas P. Pappis, University of Piraeus, Greece
Nikos I. Karacapilidis, University of Patras, Greece

Chapter 3.5. WSBen: A Web Services Discovery and Composition Benchmark Toolkit................... 739
Seog-Chan Oh, General Motors R&D Center, USA
Dongwon Lee, The Pennsylvania State University, USA

Chapter 3.6. Architecture of the Organic.Edunet Web Portal.............................................................. 759


Nikos Manouselis, Greek Research & Technology Network (GRNET S.A.), Greece
Kostas Kastrantas, Greek Research & Technology Network (GRNET S.A.), Greece
Salvador Sanchez-Alonso, University of Alcalá, Spain
Jesús Cáceres, University of Alcalá, Spain
Hannes Ebner, Royal Institute of Technology (KTH), Sweden
Matthias Palmer, Royal Institute of Technology (KTH), Sweden
Ambjorn Naeve, Royal Institute of Technology (KTH), Sweden

Chapter 3.7. Interactive Whiteboards in the Web 2.0 Classroom........................................................ 774


David Miller, Keele University, UK
Derek Glover, Keele University, UK

Chapter 3.8. Web 2.0 Technologies: Social Software Applied to Higher Education
and Adult Learning.............................................................................................................................. 794
Teresa Torres-Coronas, Universitat Rovira i Virgili, Spain
Ricard Monclús-Guitart, Universitat Rovira i Virgili, Spain
Araceli Rodríguez-Merayo, Universitat Rovira i Virgili, Spain
M. Arántzazu Vidal-Blasco, Universitat Rovira i Virgili, Spain
M. José Simón-Olmos, Universitat Rovira i Virgili, Spain

Chapter 3.9. SWELS: A Semantic Web System Supporting E-Learning............................................ 805


Gianluca Elia, University of Salento, Italy
Giustina Secundo, University of Salento, Italy
Cesare Taurino, University of Salento, Italy

Chapter 3.10. Web Services Discovery with Rough Sets.................................................................... 830


Maozhen Li, Brunel University, UK
Bin Yu, Level E Limited, UK
Vijay Sahota, Brunel University, UK
Man Qi, Canterbury Christ Church University, UK
Chapter 3.11. Generating Join Queries for Large Databases and Web Services................................. 848
Sikha Bagui, The University of West Florida, USA
Adam Loggins, Zilliant Inc., USA

Section IV. Utilization and Application

This section introduces and discusses the utilization and application of Web technologies. These par-
ticular selections highlight, among other topics, the application of semantic Web technologies to e-
tourism, e-banking, and in car repairs as well as the adoption of Web services in digital libraries. Con-
tributions included in this section provide excellent coverage of today’s online environment and insight
into how Web technologies impact the fabric of our present-day global village.

Chapter 4.1. The Role of Web Services: A Balance Scorecard Perspective........................................ 865
Pauline Ratnasingam, University of Central Missouri, USA

Chapter 4.2. Semantic Web Take-Off in a European Industry Perspective......................................... 880


Alain Léger, France Telecom R&D, France
Johannes Heinecke, France Telecom R&D, France
Lyndon J.B. Nixon, Freie Universität Berlin, Germany
Pavel Shvaiko, University of Trento, Italy
Jean Charlet, STIM, DPA/AP-Hopitaux Paris & Université Paris 6, France
Paola Hobson, Motorola Labs, UK
François Goasdoué, LRI, CNRS et Université Paris Sud XI, France

Chapter 4.3. A Strategic Framework for Integrating Web 2.0 into the Marketing Mix....................... 909
Samantha C. Bryant, Philip Morris, USA

Chapter 4.4. Applying Semantic Web Technologies to Car Repairs.................................................... 924


Martin Bryan, CSW Group Ltd., UK
Jay Cousins, CSW Group Ltd., UK

Chapter 4.5. The Web Strategy Development in the Automotive Sector............................................. 943
Massimo Memmola, Catholic University, Italy
Alessandra Tzannis, Catholic University, Italy

Chapter 4.6. Using Semantic Web Services in E-Banking Solutions.................................................. 971


Laurent Cicurel, iSOCO, Spain
José Luis Bas Uribe, Bankinter, Spain
Sergio Bellido Gonzalez, Bankinter, Spain
Jesús Contreras, iSOCO, Spain
José-Manuel López-Cobo, iSOCO, Spain
Silvestre Losada, iSOCO, Spain

Chapter 4.7. Innovating through the Web: The Banking Industry Case.............................................. 988
Chiara Frigerio, Università Cattolica del Sacro Cuore, Italy
Chapter 4.8. Semantic Web for Media Convergence: A Newspaper Case......................................... 1003
Ferran Perdrix, Universitat de Lleida, Spain & Diari Segre Media Group, Spain
Juan Manuel Gimeno, Universitat de Lleida, Spain
Rosa Gil, Universitat de Lleida, Spain
Marta Oliva, Universitat de Lleida, Spain
Roberto García, Universitat de Lleida, Spain

Chapter 4.9. Applying Semantic Web to E-Tourism.......................................................................... 1027


Danica Damljanović, University of Sheffield, UK
Vladan Devedžić, University of Belgrade, Serbia

Chapter 4.10. E-Tourism Image: The Relevance of Networking for Web Sites
Destination Marketing....................................................................................................................... 1050
Lluís Prats-Planagumà, Universitat de Girona, Spain
Raquel Camprubí, Universitat de Girona, Spain

Chapter 4.11. Successful Web-Based IT Support Services: Service Provider Perceptions


of Stakeholder-Oriented Challenges.................................................................................................. 1069
Vanessa Cooper, RMIT University, Australia
Sharman Lichtenstein, Deakin University, Australia
Ross Smith, RMIT University, Australia

Chapter 4.12. Mailing Lists and Social Semantic Web...................................................................... 1090


Sergio Fernández, Fundación CTIC, Spain
Diego Berrueta, Fundación CTIC, Spain
Lian Shi, Fundación CTIC, Spain
Jose E. Labra, University of Oviedo, Spain
Patricia Ordóñez de Pablos, University of Oviedo, Spain

Chapter 4.13. Communicative Networking and Linguistic Mashups on Web 2.0............................. 1105
Mark Pegrum, University of Western Australia, Australia

Chapter 4.14. Adoption of Web Services in Digital Libraries: An Exploratory Study...................... 1127
Fatih Oguz, Valdosta State University, USA

Chapter 4.15. A Context-Based Approach to Web 2.0 and Language Education.............................. 1141
Gary Motteram, University of Manchester, UK
Susan Brown, University of Manchester, UK

Chapter 4.16. An Adaptive and Context-Aware Scenario Model Based


on a Web Service Architecture for Pervasive Learning Systems....................................................... 1159
Cuong Pham-Nguyen, TELECOM, France
Serge Garlatti, TELECOM, France
B.-Y.-Simon Lau, Multimedia University, Malaysia
Benjamin Barbry, University of Sciences and Technologies of Lille, France
Thomas Vantroys, University of Sciences and Technologies of Lille, France
Chapter 4.17. Exploring the Effects of Web-Enabled Self-Regulated Learning
and Online Class Frequency on Students’ Computing Skills in Blended Learning Courses............. 1181
Pei-Di Shen, Ming Chuan University, Taiwan
Chia-Wen Tsai, Ming Chuan University, Taiwan

Section V. Organizational and Social Implications

This section includes a wide range of research pertaining to the social and organizational impact of
Web technologies around the world. Chapters included in this section analyze social marketing, e-
government, Web vendors, and Web tourism. The inquiries and methods presented in this section offer
insight into the implications of Web technologies at both a personal and organizational level, while also
emphasizing potential areas of study within the discipline.

Chapter 5.1. Building Trust in E-Commerce through Web Interface................................................ 1195


Muneesh Kumar, University of Delhi South Campus, India & ESC-PAU, France
Mamta Sareen, University of Delhi, India

Chapter 5.2. Swift Trust in Web Vendors: The Role of Appearance and Functionality..................... 1206
Xin Li, University of North Carolina at Pembroke, USA
Guang Rong, Clemson University, USA
Jason B. Thatcher, Clemson University, USA

Chapter 5.3. Understanding Brand Website Positioning in the New EU Member States:
The Case of the Czech Republic........................................................................................................ 1228
Shintaro Okazaki, Universidad Autónoma de Madrid, Spain
Radoslav Škapa, Masaryk University Brno, Czech Republic

Chapter 5.4. WEB 2.0, Social Marketing Strategies and Distribution Channels for City
Destinations: Enhancing the Participatory Role of Travelers and Exploiting
their Collective Intelligence............................................................................................................... 1249
Marianna Sigala, University of the Aegean, Greece

Chapter 5.5. City Brands and their Communication through Web Sites: Identification
of Problems and Proposals for Improvement..................................................................................... 1274
José Fernández-Cavia, Universitat Pompeu Fabra, Spain
Assumpció Huertas-Roig, Universitat Rovira i Virgili, Spain

Chapter 5.6. Assessing the Performance of Airline Web Sites: The ARTFLY Case.......................... 1298
Elad Harison, University of Groningen, The Netherlands
Albert Boonstra, University of Groningen, The Netherlands

Chapter 5.7. Aviation-Related Expertise and Usability: Implications for the Design
of an FAA E-Government Web Site................................................................................................... 1312
Ferne Friedman-Berg, FAA Human Factors Team - Atlantic City, USA
Kenneth Allendoerfer, FAA Human Factors Team - Atlantic City, USA
Shantanu Pai, Engility Corporation, USA
Volume III
Chapter 5.8. Quality Enhancing the Continued Use of E-Government Web Sites:
Evidence from E-Citizens of Thailand.............................................................................................. 1328
Sivaporn Wangpipatwong, Bangkok University, Thailand
Wichian Chutimaskul, King Mongkut’s University of Technology Thonburi, Thailand
Borworn Papasratorn, King Mongkut’s University of Technology Thonburi, Thailand

Chapter 5.9. Social Aspects of Mobile Technologies on Web Tourism Trend................................... 1345
Fernando Ferri, IRPPS-CNR, Rome, Italy
Patrizia Grifoni, IRPPS-CNR, Rome, Italy
Tiziana Guzzo, IRPPS-CNR, Rome, Italy

Chapter 5.10. Healthcare Quality and Cost Transparency Using Web-Based Tools......................... 1360
Jiao Ma, Saint Louis University, USA
Cynthia LeRouge, Saint Louis University, USA

Chapter 5.11. Exploiting Collaborative Tagging Systems to Unveil the User-Experience


of Web Contents: An Operative Proposal.......................................................................................... 1374
A. Malizia, Universidad Carlos III de Madrid, Spain
A. De Angeli, University of Manchester, UK
S. Levialdi, Sapienza University of Rome, Italy
I. Aedo Cuevas, Universidad Carlos III de Madrid, Spain

Chapter 5.12. Identifying Users Stereotypes for Dynamic Web Pages Customization..................... 1388
Sandro José Rigo, Universidade Federal do Rio Grande do Sul (UFRGS), Brazil
José Palazzo M. de Oliveira, Universidade Federal do Rio Grande do Sul (UFRGS), Brazil
Leandro Krug Wives, Universidade Federal do Rio Grande do Sul (UFRGS), Brazil

Chapter 5.13. Improving Online Readability in a Web 2.0 Context.................................................. 1411


John Paul Loucky, Seinan Jogakuin University, Japan

Chapter 5.14. Querying Web Accessibility Knowledge from Web Graphs....................................... 1437
Rui Lopes, LaSIGE, University of Lisbon, Portugal
Luís Carriço, LaSIGE, University of Lisbon, Portugal

Chapter 5.15. Feature Selection for Web Page Classification............................................................ 1462


K. Selvakuberan, Tata Consultancy Services, India
M. Indra Devi, Thiagarajar College of Engineering, India
R. Rajaram, Thiagarajar College of Engineering, India

Chapter 5.16. Implementing Collaborative Problem-Based Learning with Web 2.0........................ 1478
Steven C. Mills, The University Center of Southern Oklahoma, USA

Chapter 5.17. Machine Learning and Web Mining: Methods and Applications
in Societal Benefit Areas.................................................................................................................... 1495
Georgios Lappas, Technological Educational Institution of Western Macedonia, Kastoria
Campus, Greece
Chapter 5.18. Towards Mobile Web 2.0-Based Business Methods: Collaborative
QoS-Information Sharing for Mobile Service Users......................................................................... 1515
Katarzyna Wac, University of Geneva, Switzerland & University of Twente, The Netherlands
Richard Bults, University of Twente, The Netherlands & Mobihealth B.V., The Netherlands
Bert-Jan van Beijnum, University of Twente, The Netherlands
Hong Chen, Altran Netherlands B.V., The Netherlands
Dimitri Konstantas, University of Geneva, Switzerland

Chapter 5.19. The Pedagogical Implications of Web 2.0................................................................... 1536


Matthias Sturm, ICT Consultant, Canada
Trudy Kennell, ICT Consultant, Canada
Rob McBride, ICT Consultant, Canada
Mike Kelly, ICT Consultant, Canada

Chapter 5.20. Developing Digital Literacy Skills with WebQuests and Web Inquiry Projects......... 1554
Susan Gibson, University of Alberta, Canada

Chapter 5.21. EduOntoWiki Project for Supporting Social, Educational,


and Knowledge Construction Processes with Semantic Web Paradigm............................................ 1570
Corrado Petrucco, University of Padua, Italy

Chapter 5.22. Modeling Best Practices in Web-Based Academic Development............................... 1578


Diana K. Kelly, San Diego Miramar College, USA

Chapter 5.23. The Use of Weblogs in Language Education.............................................................. 1596


Thomas Raith, The University of Education Heidelberg, Germany

Chapter 5.24. The Effects of Web-Enabled Self-Regulated Learning and Problem-Based


Learning with Initiation on Students’ Computing Skills................................................................... 1614
Pei-Di Shen, Ming Chuan University, Taiwan

Chapter 5.25. Hypertinence, Serendipity or Elicitation of Passion for Knowledge?


Some Critical Elements of Online Learning by Using Web 2.0 Resources....................................... 1628
Simona Marchi, University “Sapienza” of Rome, Italy

Chapter 5.26. Blending Virtual Campuses Managing Differences Through Web 2.0
Experiences in Transnational Cooperation Projects.......................................................................... 1642
Yuri Kazepov, University of Urbino “Carlo Bo,” Italy
Giovanni Torrisi, University of Urbino “Carlo Bo,” Italy

Section VI. Managerial Impact

This section presents contemporary coverage of the managerial implications of Web technologies. Par-
ticular contributions address Web software engineering and Web-enabled employee life-cycle process
management. The managerial research provided in this section allows executives, practitioners, and
researchers to gain a better sense of how Web technologies can inform their practices and behavior.
Chapter 6.1. Enterprise 2.0: Collaboration and Knowledge Emergence
as a Business Web Strategy Enabler.................................................................................................. 1663
Javier Soriano, Universidad Politécnica de Madrid, Spain
David Lizcano, Universidad Politécnica de Madrid, Spain
Marcos Reyes, Telefónica I+D, Spain
Fernando Alonso, Universidad Politécnica de Madrid, Spain
Genoveva López, Universidad Politécnica de Madrid, Spain

Chapter 6.2. Web Engineering in Small Jordanian Web Development Firms:


An XP Based Process Model............................................................................................................. 1696
Haroon Altarawneh, Albalqa’ Applied University, Jordan
Asim El-Shiekh, The Arab Academy for Banking and Financial Sciences, Jordan

Chapter 6.3. Employee Life-Cycle Process Management Improvement


with Web-Enabled Workflow Systems............................................................................................... 1708
Leon Welicki, Microsoft, Canada
Javier Piqueres Juan, Systar, Spain
Fernando Llorente Martin, ONO, Spain
Victor de Vega Hernandez, ONO, Spain

Chapter 6.4. Some Key Success Factors in Web-Based Corporate Training in Brazil:
A Multiple Case Study....................................................................................................................... 1724
Luiz Antonio Joia, Brazilian School of Public and Business Administration of Getulio Vargas
Foundation and Rio de Janeiro State University, Brazil
Mário Figueiredo Costa, Brazilian School of Public and Business Administration of Getulio
Vargas Foundation, Brazil

Chapter 6.5. Multi-Tier Framework for Management of Web Services’ Quality.............................. 1745
Abdelghani Benharref, Concordia University, Canada
Mohamed Adel Serhani, United Arab Emirates University, UAE
Mohamed Salem, University of Wollongong, Dubai, UAE
Rachida Dssouli, Concordia University, Canada

Chapter 6.6. On the Management Performance of Networked Environments


Using Web Services Technologies..................................................................................................... 1768
Lisandro Zambenedetti Granville, Federal University of Rio Grande do Sul–Porto Alegre,
Brazil
Ricardo Neisse, Federal University of Rio Grande do Sul–Porto Alegre, Brazil
Ricardo Lemos Vianna, Federal University of Rio Grande do Sul–Porto Alegre, Brazil
Tiago Fioreze, Federal University of Rio Grande do Sul–Porto Alegre, Brazil

Chapter 6.7. Web Services, Service-Oriented Computing, and Service-Oriented Architecture:


Separating Hype from Reality........................................................................................................... 1786
John Erickson, University of Nebraska - Omaha, USA
Keng Siau, University of Nebraska - Lincoln, USA
Chapter 6.8. Diffusion and Oscillation of Telecommunications Services:
The Case of Web 2.0 Platforms......................................................................................................... 1799
Tobias Kollmann, University of Duisburg-Essen, Campus Essen, Germany
Christoph Stöckmann, University of Duisburg-Essen, Campus Essen, Germany
Carsten Schröer, University of Duisburg-Essen, Campus Essen, Germany

Section VII. Critical Issues

This section addresses conceptual and theoretical issues related to the field of Web technologies, which
include issues related to usage, as well as failures and successes in Web implementation. Within these
chapters, the reader is presented with analysis of the most current and relevant conceptual inquires
within this growing field of study. Particular chapters address privacy concerns in Web logging, Web
information extraction, and Web rules. Overall, contributions within this section ask unique, often
theoretical questions related to the study of Web technologies and, more often than not, conclude that
solutions are both numerous and contradictory.

Chapter 7.1. Applying an Organizational Uncertainty Principle: Semantic Web-Based Metrics...... 1814
Joseph Wood, LTC, US Army, USA
James Grayson, Augusta State University, USA
Hui-Lien Tung, Paine College, USA
Margo Bergman, Northwest Health Services Research & Development (HSR&D), USA
Tina Marshall-Bradley, Paine College, USA
W.F. Lawless, Paine College, USA
Donald A. Sofge, Naval Research Laboratory, USA

Chapter 7.2. Bridging the Gap between Mobile Application Contexts


and Web Resources............................................................................................................................ 1834
Stefan Dietze, Open University, UK
Alessio Gugliotta, Open University, UK
John Domingue, Open University, UK

Chapter 7.3. Uncertainty Representation and Reasoning in the Semantic Web................................ 1852
Paulo Cesar G. Costa, George Mason University, USA
Kathryn Blackmond Laskey, George Mason University, USA
Thomas Lukasiewicz, Oxford University Computing Laboratory, UK

Chapter 7.4. Semantic Web-Enabled Protocol Mediation for the Logistics Domain........................ 1878
Oscar Corcho, Universidad Politécnica de Madrid, Spain
Silvestre Losada, Intelligent Software Components, S.A., Spain
Richard Benjamins, Intelligent Software Components, S.A., Spain

Chapter 7.5. Probabilistic Models for the Semantic Web: A Survey................................................. 1896
Livia Predoiu, University of Mannheim, Germany
Heiner Stuckenschmidt, University of Mannheim, Germany
Chapter 7.6. Estimating the Privacy Protection Capability of a Web Service Provider.................... 1929
George O.M. Yee, Institute for Information Technology, National Research Council, Canada

Chapter 7.7. Privacy Concerns for Web Logging Data...................................................................... 1951


Kirstie Hawkey, University of British Columbia, Canada

Chapter 7.8. A Model-Based Approach for Diagnosing Fault in Web Service Processes................. 1970
Yuhong Yan, Concordia University, Canada
Philippe Dague, University Paris-Sud 11, France
Yannick Pencolé, LAAS-CNRS, France
Marie-Odile Cordier, IRISA, France

Volume IV
Chapter 7.9. Management of Medical Website Quality Labels via Web Mining.............................. 1994
Vangelis Karkaletsis, National Center of Scientific Research “Demokritos”, Greece
Konstantinos Stamatakis, National Center of Scientific Research “Demokritos”, Greece
Pythagoras Karampiperis, National Center of Scientific Research “Demokritos”, Greece
Martin Labský, University of Economics, Prague, Czech Republic
Marek Růžička, University of Economics, Prague, Czech Republic
Vojtěch Svátek, University of Economics, Prague, Czech Republic
Enrique Amigó Cabrera, ETSI Informática, UNED, Spain
Matti Pöllä, Helsinki University of Technology, Finland
Miquel Angel Mayer, Medical Association of Barcelona (COMB), Spain
Dagmar Villarroel Gonzales, Agency for Quality in Medicine (AquMed), Germany

Chapter 7.10. User Facing Web Services in Portals.......................................................................... 2015


Jana Polgar, NextDigital, Australia

Chapter 7.11. Hyperlink Structure Inspired by Web Usage............................................................... 2034


Pawan Lingras, Saint Mary’s University, Canada
Rucha Lingras, Saint Mary’s University, Canada

Chapter 7.12. Search Engine-Based Web Information Extraction..................................................... 2048


Gijs Geleijnse, Philips Research, The Netherlands
Jan Korst, Philips Research, The Netherlands

Chapter 7.13. Profiling of Web Services to Measure and Verify


their Non-Functional Properties......................................................................................................... 2082
Witold Abramowicz, Poznań University of Economics, Poland
Monika Kaczmarek, Poznań University of Economics, Poland
Dominik Zyskowski, Poznań University of Economics, Poland
Chapter 7.14. On the Use of Web Services in Content Adaptation................................................... 2099
Khalil El-Khatib, University of Ontario Institute of Technology, Canada
Gregor V. Bochmann, University of Ottawa, Canada
Abdulmotaleb El-Saddik, University of Ottawa, Canada

Chapter 7.15. Reconceptualising Information Literacy for the Web 2.0 Environment?................... 2115
Sharon Markless, King’s College, London, UK
David Streatfield, Information Management Associates, UK

Chapter 7.16. Aspect-Oriented Framework for Web Services (AoF4WS): Introduction


and Two Example Case Studies......................................................................................................... 2134
Ghita Kouadri Mostefaoui, Oxford University Computing Laboratory, UK
Zakaria Maamar, Zayed University, UAE
Nanjangud C. Narendra, IBM India Research Lab, India

Chapter 7.17. A Static Web Immune System and Its Robustness Analysis....................................... 2152
Tao Gong, Donghua University, China & Central South University, China

Chapter 7.18. Mapping Policies to Web Rules: A Case of the KAoS Policy Language.................... 2175
Nima Kaviani, University of British Columbia, Canada
Dragan Gašević, Athabasca University, Canada
Marek Hatala, Simon Fraser University, Canada

Chapter 7.19. Scalable Authoritative OWL Reasoning for the Web.................................................. 2206
Aidan Hogan, National University of Ireland, Ireland
Andreas Harth, National University of Ireland, Ireland
Axel Polleres, National University of Ireland, Ireland

Chapter 7.20. A Framework for Integrating the Social Web Environment


in Pattern Engineering........................................................................................................................ 2250
Pankaj Kamthan, Concordia University, Canada

Section VIII. Emerging Trends

This section highlights research potential within the field of Web technologies while exploring unchart-
ed areas of study for the advancement of the discipline. Chapters within this section highlight emerging
semantic Web applications, Web personalization, and learning on the Web. These contributions, which
conclude this exhaustive, multi-volume set, provide emerging trends and suggestions for future research
within this rapidly expanding discipline.
Chapter 8.1. The Social Semantic Desktop: A New Paradigm Towards Deploying
the Semantic Web on the Desktop..................................................................................................... 2279
Ansgar Bernardi, German Research Center for Artificial Intelligence (DFKI) GmbH,
Kaiserslautern, Germany
Stefan Decker, National University of Ireland, Ireland
Ludger van Elst, German Research Center for Artificial Intelligence (DFKI) GmbH,
Kaiserslautern, Germany
Gunnar Aastrand Grimnes, German Research Center for Artificial Intelligence (DFKI)
GmbH, Kaiserslautern, Germany
Tudor Groza, National University of Ireland, Ireland
Siegfried Handschuh, National University of Ireland, Ireland
Mehdi Jazayeri, University of Lugano, Switzerland
Cédric Mesnage, University of Lugano, Switzerland
Knud Möller, National University of Ireland, Ireland
Gerald Reif, University of Lugano, Switzerland
Michael Sintek, German Research Center for Artificial Intelligence (DFKI) GmbH,
Kaiserslautern, Germany
Leo Sauermann, German Research Center for Artificial Intelligence (DFKI) GmbH,
Germany

Chapter 8.2. Explaining Semantic Web Applications........................................................................ 2304


Deborah L. McGuinness, Tetherless World Constellation, Rensselaer
Polytechnic Institute (RPI), USA & Stanford University, KSL, USA
Vasco Furtado, University of Fortaleza, UNIFOR, Brazil
Paulo Pinheiro da Silva, University of Texas at El Paso (UTEP), USA
Li Ding, Tetherless World Constellation, Rensselaer Polytechnic Institute (RPI), USA and
Stanford University, KSL, USA
Alyssa Glass, Stanford University, KSL, USA
Cynthia Chang, Stanford University, KSL, USA

Chapter 8.3. A New System for the Integration of Medical Imaging Processing
Algorithms into a Web Environment................................................................................................. 2328
José Antonio Seoane Fernández, Artificial Neural Networks and Adaptative Systems Group,
Spain & University of Corunna, Spain
Juan Luis Pérez Ordóñez, Center of Medical Informatics and Radiological Diagnosis, Spain
& University of Corunna, Spain
Noha Veiguela Blanco, Artificial Neural Networks and Adaptative Systems Group, Spain &
University of Corunna, Spain
Francisco Javier Novóa de Manuel, Center of Medical Informatics and Radiological
Diagnosis, Spain & University of Corunna, Spain
Julián Dorado de la Calle, University of A Coruña, Spain

Chapter 8.4. Social Media Marketing: Web X.0 of Opportunities.................................................... 2341


Lemi Baruh, Kadir Has University, Turkey
Chapter 8.5. Web Content Recommendation Methods Based on Reinforcement Learning.............. 2353
Nima Taghipour, Amirkabir University of Technology, Iran
Ahmad Kardan, Amirkabir University of Technology, Iran

Chapter 8.6. On the Use of Soft Computing Techniques for Web Personalization........................... 2381
G. Castellano, University of Bari, Italy
A. M. Fanelli, University of Bari, Italy
M. A. Torsello, University of Bari, Italy

Chapter 8.7. Enhancing the Testability of Web Services................................................................... 2403


Daniel Brenner, University of Mannheim, Germany
Barbara Paech, University of Heidelberg, Germany
Matthias Merdes, Heidelberg Mobil International GmbH, Germany
Rainer Malaka, University of Bremen, Germany

Chapter 8.8. Making the Web Accessible to the Visually Impaired................................................... 2423
Simone Bacellar Leal Ferreira, Universidade Federal do Estado do Rio de Janeiro, Brazil
Denis Silva da Silveira, Programa de Engenharia de Produção - COPPE/UFRJ, Brazil
Marcos Gurgel do Amaral Leal Ferreira, Holden Comunicação Ltda, Brazil
Ricardo Rodrigues Nunes, Universidade Federal do Estado do Rio de Janeiro, Brazil

Chapter 8.9. Web Application Server Clustering with Distributed Java Virtual Machine................. 2436
King Tin Lam, The University of Hong Kong, Hong Kong
Cho-Li Wang, The University of Hong Kong, Hong Kong

Chapter 8.10. Virtual Web Services: Extension Architecture to Alleviate Open Problems
in Web Services Technology.............................................................................................................. 2460
Julio Fernández Vilas, University of Vigo, Spain
Jose J. Pazos Arias, University of Vigo, Spain
Ana Fernández Vilas, University of Vigo, Spain

Chapter 8.11. Web-Based Corporate Governance Information Disclosure:


An Empirical Investigation................................................................................................................ 2479
Yabing Jiang, Fordham University, USA
Viju Raghupathi, City University of New York, USA
Wullianallur Raghupathi, Fordham University, USA

Chapter 8.12. Using Web Service Enhancements to Establish Trust Relationships


with Privacy Protection: (Extended and Invited from ICWS 2006 with id 47)................................. 2498
Zhengping Wu, University of Bridgeport, USA
Alfred C. Weaver, University of Virginia, USA

Chapter 8.13. The Interactive Computing of Web Knowledge Flow: From Web
to Knowledge Web............................................................................................................................. 2518
Xiangfeng Luo, Shanghai University, P. R. China
Jie Yu, Shanghai University, P. R. China
Chapter 8.14. Knowledge Producing Megamachines: The Biggest Web 2.0 Communities
of the Future....................................................................................................................................... 2530
Laszlo Z. Karvalics, University of Szeged, Hungary

Chapter 8.15. Utilizing Past Web for Knowledge Discovery............................................................ 2544


Adam Jatowt, Kyoto University, Japan
Yukiko Kawai, Kyoto Sangyo University, Japan
Katsumi Tanaka, Kyoto University, Japan

Chapter 8.16. New Forms of Deep Learning on the Web: Meeting the Challenge
of Cognitive Load in Conditions of Unfettered Exploration in Online
Multimedia Environments................................................................................................................. 2563
Michael DeSchryver, Michigan State University, USA
Rand J. Spiro, Michigan State University, USA

Chapter 8.17. General Strategy for Querying Web Sources in a Data Federation Environment....... 2582
Aykut Firat, Northeastern University, USA
Lynn Wu, Massachusetts Institute of Technology, USA
Stuart Madnick, Massachusetts Institute of Technology, USA

Chapter 8.18. Empirical Studies for Web Effort Estimation.............................................................. 2600


Sergio Di Martino, Università di Salerno & Università degli Studi di Napoli “Federico II”,
Italy
Filomena Ferrucci, Università di Salerno, Italy
Carmine Gravino, Università di Salerno, Italy
xxxii

Preface

Since its development just two decades ago, the World Wide Web has grown to become the infrastruc-
ture that supports innumerable applications essential to everyday life. It’s not an exaggeration to claim
that if you can think it, you can create a Web page about it. We use Web sites and the information they
contain to create and connect with a seemingly unlimited amount of information. As such, it is important
to understand the tools and technologies that support the continued growth of the Web and contribute
to its role as an increasingly-pervasive aspect of our lives.
With the constant changes in the landscape of Web technologies, it is a challenge for researchers
and experts to take in the volume of innovative advances and up-to-the-moment research in this diverse
field. Information Science Reference is pleased to offer a four-volume reference collection on this
rapidly growing discipline, in order to empower students, researchers, academicians, and practitioners
with a wide-ranging understanding of the most critical areas within this field of study. This collection
provides the most comprehensive, in-depth, and recent coverage of all issues related to the development
of cutting-edge Web technologies, as well as a single reference source on all conceptual, methodological,
technical and managerial issues, and the opportunities, future challenges and emerging trends related to
the development, application, and implications of Web technologies.
This collection entitled, “Web Technologies: Concepts, Methodologies, Tools, and Applications”
is organized in eight (8) distinct sections, providing the most wide-ranging coverage of topics such
as: 1) Fundamental Concepts and Theories; 2) Development and Design Methodologies; 3) Tools and
Technologies; 4) Utilization and Application; 5) Organizational and Social Implications; 6) Manage-
rial Impact; 7) Critical Issues; and 8) Emerging Trends. The following provides a summary of what is
covered in each section of this multi-volume reference collection:
Section 1, Fundamental Concepts and Theories, serves as a foundation for this extensive reference
tool by addressing crucial theories essential to the understanding of Web technologies. Chapters such as
“Tips for Tracking Web Information Seeking Behavior” by Brian Detlor, Maureen Hupfer, and Umar Ruhi
and “A Proposed Template for the Evaluation of Web Design Strategies” by Dimitrios Xanthidis, David
Nicholas, and Paris Argyrides provide analyses of user behavior and Web design. “Mobile Social Web;
Opportunities and Drawbacks,” by Thorsten Caus, Stefan Christmann, and Svenja Hagenhoff presents
an overview of recent trends in mobile Web usage, which is becoming an increasingly important area of
study as more and more people obtain Internet access for their wireless devices. Later selections, such
as “Web 2.0 and E-Discovery” by Bryan Kimes and “The Power and Promise of Web 2.0 Tools” by G.
Andrew Page and Radwan Ali explore the application of Web 2.0 as well as the issues companies must
address as a result. These and several other foundational chapters provide a wealth of expert research
on the elemental concepts and ideas which surround Web design and access.
Section 2, Development and Design Methodologies, presents in-depth coverage of the conceptual
design and architecture of Web sites, services, and systems. “Paralingual Web Design and Trust in E-
xxxiii

Government,” by Roy H. Segovia, Murray E. Jennex, and James Beatty and “Designing Medical Research
Web Sites” by Jonathan Grady, Michael B. Spring, and Armando J. Rotondi discuss context-specific
Web design projects, highlighting the importance of recognizing the specific needs and requirements
of different development initiatives. The latter half of this section introduces concepts that relate to the
development of Semantic Web services. Chapters such as “A Semantic Web-Based Approach for Building
Personalized News Services” by Flavius Frasincar, Jethro Borsje, and Leonard Levering and “Building
Semantic Web Portals with a Model-Driven Design Approach” by Marco Brambilla and Federico M.
Facca offer specific considerations for the creation of Semantic Web services, while later selections such
as “Rule Markup Languages and Semantic Web Rule Languages” by Adrian Paschke and Harold Boley
and “Semantic Web Rule Languages for Geospatial Ontologies” by Philip D. Smart, Alia I. Abdelmoty,
Baher A. El-Geresy, and Christopher B. Jones present more technical considerations relating to the use
and communication of rule languages in the Semantic Web. With 20 contributions from leading inter-
national researchers, this section offers copious developmental approaches and methodologies for Web
services and technologies.
Section 3, Tools and Technologies, presents extensive coverage of the various tools and technolo-
gies used in the development and implementation of Web services and applications. This comprehen-
sive section opens with the chapters “New Paradigms: A Collaborative Web Based Research Tool,”
by Hamish Holewa, and “Adaptability and Adaptivity in The Generation of Web Applications,” by
Raoudha Ben Djemaa, Ikram Amous, and Abdelmajid Ben Hamadou, which describe new tools that
support the development of Web applications and the challenges faced in the management and creation
of new technology. “Migrating Web Services in Mobile and Wireless Environments,” by Myung-Woo
Park, Yeon-Seok Kim, and Kyong-Ho Lee revisits Web use on wireless devices, specifically explor-
ing the mitigation and replication of Web services among mobile devices. Later selections such as
“Web 2.0 Technologies: Social Software Applied to Higher Education and Adult Learning” by Teresa
Torres-Coronas, M. Arántzazu Vidal-Blasco, Ricard Monclús-Guitart, M. José Simón-Olmos, and Ar-
aceli Rodríguez-Merayo and “Interactive Whiteboards in the Web 2.0 Classroom” by David Miller and
Derek Glover provide insight into the use of specific Web tools (namely social software and interactive
whiteboards) in educational settings. In all, this section provides coverage of a variety of Web tools and
technologies under development and in use.
Section 4, Utilization and Application, describes the implementation and use of an assortment of Web
technologies. Including chapters such as “Semantic Web Take-Off in a European Industry Perspective”
by Alain Léger, Jean Charlet, Johannes Heinecke, Paola Hobson, Lyndon J.B. Nixon, François Goas-
doué, and Pavel Shvaiko and “Semantic Web for Media Convergence: A Newspaper Case” by Ferran
Perdrix, Juan Manuel Gimeno, Rosa Gil, Marta Oliva, and Roberto García provide specific insight into
the application of Web tools and technologies in both the professional and private sector. “Mailing Lists
and Social Semantic Web” by Sergio Fernández, Jose E. Labra, Diego Berrueta, Patricia Ordóñez de
Pablos, and Lian Shi describes the use of mailing lists and presents a method for extracting data from
these lists. Later selections, such as “A Context-Based Approach to Web 2.0 and Language Education” by
Gary Motteram and Susan Brown and “Exploring the Effects of Web-Enabled Self-Regulated Learning
and Online Class Frequency on Students’ Computing Skills in Blended Learning Courses” by Pei-Di
Shen and Chia-Wen Tsai suggest approaches and consider the impact of Web-based learning on student
performance. Contributions found in this section provide comprehensive coverage of the practicality
and current use of Web technologies.
Section 5, Organizational and Social Implications, includes chapters discussing the impact of Web
technology on social and organizational practices. Chapters such as “Building Trust in E-Commerce
through Web Interface,” by Muneesh Kumar and Mamta Sareen and and “Swift Trust in Web Vendors:
xxxiv

The Role of Appearance and Functionality,” by Xin Li, Guang Rong, and Jason B. Thatcher discuss the
growth and influence of e-commerce and the important role trust plays in impacting e-marketplaces.
Specific Web implementation and resulting implications of such initiatives are explored in selections
such as “Assessing the Performance of Airline Web Sites: The ARTFLY Case” by Elad Harison and
Albert Boonstra and “Aviation-Related Expertise and Usability: Implications for the Design of an FAA
E-Government Web Site” by Ferne Friedman-Berg, Kenneth Allendoerfer, and Shantanu Pai. This sec-
tion continues with discussions of Web accessibility and customization, concluding with a discussion of
educational implications of Web technology. Overall, these chapters present a detailed investigation of
how Web technology is implemented and how this implementation impacts the individual and society
as a whole.
Section 6, Managerial Impact, presents focused coverage of Web services and technology as it re-
lates to improvements and considerations in the workplace. “Employee Life-Cycle Process Management
Improvement with Web-Enabled Workflow Systems” by Leon Welicki, Javier Piqueres Juan, Fernando
Llorente Martin, and Victor de Vega Hernandez presents a real-world case of constructing a Web-enabled
worklflow for managing employee-life cycle processes, which include hiring and dismissing of employ-
ees. “Web Engineering in Small Jordanian Web Development Firms: An XP Based Process Model” by
Haroon Altarawneh and Asim El-Shiekh describes a model for small Web project development and ex-
plains, from a managerial perspective, how this differs from the more large-scale implementation projects
adopted by larger firms. In all, the chapters in this section offer specific perspectives on how work and
Web technologies interact and inform each other to create more meaningful user experiences.
Section 7, Critical Issues, addresses vital issues related to Web technology, which include privacy
and quality, among other topics. Chapters such as “Privacy Concerns for Web Logging Data” by Kirstie
Hawkey explore the issues that must be considered when collecting user data and offer recommendations
for enhancing privacy. Later selections, such as “Search Engine-Based Web Information Extraction” by
Gijs Geleijnse and Jan Korst, continue the discussion of information gathering and extraction which,
in this chapter, is discussed in terms of approaches to expressing and sharing structured information in
Semantic Web languages. This section continues by asking unique questions about information literacy,
as well as presenting new solutions to questions about the social Web and Web services profiling.
The concluding section of this authoritative reference tool, Emerging Trends, highlights areas for
future research within the field of Web technology, while exploring new avenues for the advancement
of the discipline. Beginning this section is “The Social Semantic Desktop: A New Paradigm Towards
Deploying the Semantic Web on the Desktop” by Ansgar Bernardi, Mehdi Jazayeri, Stefan Decker, Cédric
Mesnage, Ludger van Elst, Knud Möller, Gunnar Aastrand Grimnes, Michael Sintek, Tudor Groza, Leo
Sauermann, and Siegfried Handschuh. This selection presents the Social Semantic Desktop project, ad-
dressing design considerations of a project whose aim is to blur the lines between individual applications
and users’ physical workspace. Trends in marketing are explored in “Social Media Marketing; Web X.0
of Opportunities” by Lemi Baruh with the aim of introducing new techniques for advertisers whose aim
is to reach consumers through social media. These and several other emerging trends and suggestions
for future research can be found within the final section of this exhaustive multi-volume set.
Although the primary organization of the contents in this multi-volume work is based on its eight
sections, offering a progression of coverage of the important concepts, methodologies, technologies,
applications, social issues, and emerging trends, the reader can also identify specific contents by utilizing
the extensive indexing system listed at the end of each volume. Furthermore to ensure that the scholar,
researcher and educator have access to the entire contents of this multi volume set as well as additional
coverage that could not be included in the print version of this publication, the publisher will provide
unlimited multi-user electronic access to the online aggregated database of this collection for the life
xxxv

of the edition, free of charge when a library purchases a print copy. This aggregated database provides
far more contents than what can be included in the print version in addition to continual updates. This
unlimited access, coupled with the continuous updates to the database ensures that the most current
research is accessible to knowledge seekers.
The diverse and comprehensive coverage of Web technologies presented in this four-volume authori-
tative publication will contribute to a better understanding of all topics, research, and discoveries in
this developing, significant field of study. Furthermore, the contributions included in this multi-volume
collection series will be instrumental in the expansion of the body of knowledge in this enormous field,
resulting in a greater understanding of the fundamental concepts and technologies while fueling the
research initiatives in emerging fields. We at Information Science Reference, along with the editor of
this collection, hope that this multi-volume collection will become instrumental in the expansion of the
discipline and will promote the continued growth of all aspects of Web technology.
xxxvi

Web Technologies:
Concepts, Applications, Trends and Research Issues

Arthur Tatnall
Victoria University, Australia

Introduction

Web-based systems and technologies are now used for a vast number of applications, and this chapter aims
to provide an overview of the technologies themselves and also the uses to which the Web is now put, as
well as the social and political impact of this use. A number of important concepts underlie the Web as
well as a good deal of jargon, and some of the main concepts and terms are explained here. Design and
development of web-based systems is an important topic and this is briefly discussed along with some
of the tools and issues involved in this development. It is impossible to do justice to the huge range of
applications of the Web in an incredibly diverse range of areas, but this chapter attempts to do just this
by examining some of the most important applications. Not all aspects of the Web can be considered
to be either worthwhile or healthy for society as a whole and issues like identify theft and the distribu-
tion of pornography, both of which have been made easier by access to the Web, are also discussed. No
technological innovation can be useful, however, until it has been adopted and the factors leading to
adoption of some Web-based systems and not others, and to adoption of some aspects of these systems
by one organisation and other aspects by another organisation are also discussed. Finally this chapter
also tackles the issue of researching the Web; what this involves and what sort of methodologies might
be appropriate in this socio-technical area.

The Web: Technology, Applications and People

Today everyone knows of the World Wide Web and very many people around the world make daily
use of its facilities. It is hard to imagine what it must have been like before the Web became such an
important part of our lives, but it was only in 1989 that Tim Berners-Lee, based at the European Labora-
xxxvii

tory for Particle Physics (CERN1), in looking for a solution to the management and sharing of the large
amounts of scientific information his colleagues created, wrote a proposal for a large online hypertext
database that by 1991 had become what we now call the World Wide Web (Lawrence, Newton, Corbitt,
Braithwaite, & Parker, 2002; Sklar, 2009). Thus the Web began as a means of improving information
sharing and document handling between the research scientists at CERN and throughout the world. It
was designed to allow pages containing hypertext to be stored in a way that allowed other computers
access to these pages. It was probably not until about the mid 1990s, however, that the Web began to
really gain in popularity. At that time, few could have foreseen the multitude of uses it would be put to
by 2010, and the number of people who would make use of it. It is no exaggeration to say that the Web
has now become quite ubiquitous.
The Internet has been around much longer of course, tracing its ancestry back to the ARPANET
(Advanced Research Projects Agency Network), developed to link US Defense Department research-
ers with those in several universities in the USA. It became operational in late 1969 (Lawrence et al.,
2002; Davison, Burgess, & Tatnall, 2008). The first appearance of the term ‘Internet’ was in 1974 as an
abbreviation for ‘Internetworking’ (Network Working Group, 1974), and things developed from there,
with electronic mail soon becoming an important form of communication within the research com-
munity that used this technology. At this time, however, making use of the Internet was not something
that the average person or business could easily do or find much value in, and it was not until the later
advent of the Web that use of the Internet became common, and a general topic of conversation in many
communities.
No one knows exactly the size of the Internet, but an article in New Scientist (Barras, 2009) notes
that back in 2005, Google estimated that the Internet contained 5 million terabytes of data. In July 2008
when the new search engine Cuil.com commenced operation claiming to be the world’s largest search
engine, Google announced that it had registered a trillion unique pages, but in reality the Internet is
probably even bigger than this as some estimates suggest that the pages indexed by Google and Cuil may
represent only a hundredth of the information on the Internet (Barras, 2009). This same article suggests
that 210 billion e-mails were sent every day in 2008.
Today a great deal has been written about the Web, its concepts, its technologies, its design tools,
its applications and the social and political effects that have gone with its growth. The Web is not, of
course, just technology and the socio-technical nature of Web systems is an important consideration.
For the purposes of this chapter, I will consider the study of Web-based systems and Web technology
as consisting of three areas: technology, applications and people. Figure 1 (below) shows these as three
concentric circles (with technology in the middle). Each of these circles then contains a number of enti-
ties related to that circle. (Space does not permit all relevant entities to be shown here.)
This introductory chapter will present an overview of these topics. Of necessity it cannot cover every
topic fully and must be somewhat selective in its coverage with some topics only touched on and others
omitted completely.

Fundamental Concepts

Both the Internet and the World Wide Web come with a large amount of jargon. Some understanding
of this jargon and also of the concepts underlying web technology is necessary to fully appreciate the
complexity and power of the Web. For those readers who may be new to some of these ideas, some
of the main terms and concepts, along with a brief description of their meaning and use, will now be
presented.
xxxviii

Figure 1. Components of web-based systems

• Hypertext allows words (or other objects) in one document to be linked to other documents. It
provides a dynamic means of organising and accessing information where pages of information
are connected together by hypertext links. A Hyperlink (Hypertext Link) can be text, or a picture
that is associated with the location (path and filename) of another document and used to link to
this other document. These documents are called web pages and can contain text, graphics, video
and audio as well as hyperlinks. A website is a collection of related pages stored on a web server.
A web server is made up of computer hardware and software, based on a PC or other larger com-
puter.
• A URL (Uniform Resource Locator) gives the address or location of any specific website. Each
URL defines the path that will transmit the document, the Internet protocol being used, and the
server on which the website is located. Each Internet address is translated into a series of numbers
called an IP address. A domain name is used by an organisational entity to identify its website
and is based on the Domain Name System (DNS) hierarchy.
xxxix

Figure 2. A Communication Model

• A web browser is a software application used to access and view web pages. Web browsers include:
Internet Explorer, Firefox, Safari, Opera and Netscape Navigator.
• Communications Model: Communication involves a sender dispatching some form of message to
a receiver. It occurs within a particular context or setting, and involves the transfer of some form of
information from the sender to the receiver over some type of communications channel (Davison
et al., 2008). The message is coded into an appropriate form by the sender before transmission,
and later decoded by the receiver. Feedback lets the sender know how the message was received.
Noise, or interference, can upset transmission and lead to the message received differing from the
one that was sent.
• Packet switching. Data to be transmitted is broken into discrete packets consisting of groups of
characters, which are then sent independently through whichever path between sender and receiver
is most convenient at the time. The connection is virtual and so the data may follow different
paths. Each packet is labelled electronically with codes to indicate its origin and destination, and
may follow a different path through the network to that of other packets. When they reach their
destination the packets are re-assembled to produce the original message.
• Internet connection protocols: instructions for connecting a computer to the Internet. Two im-
portant protocols are PPP (Point to Point) and SLIP (Serial Line Internet Protocol).
• HTTP (HyperText Transfer Protocol) is the set of rules that locate and move files of all types,
including multimedia files, across the Internet. It does not process the packages of data it transmits
but simply ensures they reach their destination and so defines computer links to the Web. HTTPS
(HyperText Transfer Protocol (Secure)) indicates that HTTP is to be used, but with a different
default port and an additional encryption/ authentication layer between HTTP and TCP.
• Open Systems Interconnection (OSI). As networks operate in somewhat different ways, there
was a need to define standards for data communication to provide a means for computers from
different manufacturers to communicate with each other. Data sent over a network must arrive at
its destination in a timely, correct, and recognisable format and to facilitate this, the OSI model
consists of seven layers, each of which is selected to perform a well-defined function. While OSI
is an international standard it is really more important as a concept and TCP/IP is the standard that
is actually used.
• TCP/IP (Transmission Control Protocol, Internet Protocol). These are two protocols that support
the basic operation of the Internet and include rules that computers on a network use to establish
and break connections. TCP/IP controls the assembly of a message into small packets before trans-
mission, controls the reassembly of packets once they reach their destination, has rules for routing
individual data packages from their source to their destination and allows users to send and receive
messages, find information, exchange data and download software.
xl

• HTML (HyperText Markup Language) is used to describe how a web browser should display a
text file retrieved from a web server. HTML defines how a page is displayed and is a derivative of
SGML – Standard Generalised Markup Language which is a standard system (an ISO standard)
used to specify document structure. HTML allows the developer of a web page to define hyperlinks
between that document and any others that are required.
• XML (Extendible Markup Language) defines the page’s actual content and also interprets the
meaning of the data. It defines which data are displayed, whereas HTML only defines how a page
is displayed. Elements of structured information include: content (pictures, graphics, text etc) and
the role the content plays in the document; where the content is located in the document influences
the meaning given to it.
• XHTML (Extensible Hypertext Markup Language) is a newer version of HTML, based on XML
and designed to support newer devices.
• Java is a programming language used to create many web-based applications. Much of Java’s
syntax derived from C++ and its compiled version can run on any computer architecture in a Java
virtual machine.
• An intranet is a private (internal) network accessible only by selected individuals within an organi-
sation. It is based on web technology, but access is restricted so as to exclude the outside world. An
intranet uses web browsers and hypertext links in the same way as the World Wide Web, the only
real differences being where the web pages are located, and who can access them. (Applications
of intranets are discussed further in a later section.)
• An extranet is a private network (based on web technologies) that links selected parts of a com-
pany’s intranet with its customers, suppliers, or other authorised business partners. (Applications
of extranets are discussed in a later section.)
• EDI (Electronic Document Exchange) is the exchange of data or ‘business documents’ directly
between computer systems in business partner companies using a standardised data format. It has
been used for more than 30 years predominantly for purchasing and inventory. This direct computer-
to-computer form of EDI has, to a large extent, now been superseded by web-based purchasing
systems.
• RFID (Radio Frequency Identification) is a technique for storing and retrieving data remotely using
devices called RFID tags that can be attached to, or inserted into a product. An RFID tag incorpo-
rates a silicon chip and antennae that can broadcast a unique identification code when prompted
by a reader device.
• Bluetooth is an open standard for short-range wireless communication (up to 10 metres) between
digital devices. It comprises hardware, software and interoperability standards, protocols and
requirements and is implemented through a small, low-cost radio chip that can be incorporated
in mobile phones, PDAs, pocket PCs, computers, printers and similar devices. (It was initially
developed by Ericsson and is named after Harald Bluetooth, a Viking and the King of Denmark
between 940 and 981.)
• The IEEE 802.11 Protocols. The Standards Association of the Institute of Electrical and Electronic
Engineers (IEEE) has developed a series of standards for wireless local area network (WLAN)
devices operating in close proximity (up to 100m) under the generic title IEEE 802.11. There are
a number of variants of this protocol, such as IEEE 802.11a, IEEE 802.11b and IEEE 802.11g
(Wave Repo2001; O’Hara and & Petrick, 2005).
• PDA (Personal Digital Assistant). Sometimes now called a Smart Phone, these devices offer many
of the communication and information characteristics of a laptop computer and mobile phone in-
corporated into a single device. Often a PDA includes a mobile (cell) phone along with Windows
xli

Mobile operating system with mobile versions of Word, Excel and Outlook, Bluetooth and IEEE
802.11 connectivity and Internet access.
• Personal Area Networks (PAN) are ad hoc networks of personal digital devices such as laptop
computers, personal digital assistants (PDA) and mobile phones able to transfer data and linked to
other personal digital devices by Bluetooth, or some other wireless technology.
• The Semantic Web offers a common framework for data to be shared and reused across applica-
tion, enterprise, and community boundaries. The World Wide Web Consortium (W3C), who are
developing the Semantic Web, note that it is concerned with common formats for the integration and
combination of data from diverse sources and the language needed to show how this data relates to
the real world (http://www.w3.org/2001/sw/). It is an extension of Web in which the semantics of
information and services on the web are defined. This then makes it possible to understand requests
of people and machines who want to use web content.
• Web 2.0 websites build on the interactive facilities of the Web and so allow users to do more than
just retrieve information. This has led to the development of web-based communities and social-
networking sites.
• Internet service provider (ISP). This is a company that offers its customers access to the Inter-
net
• Web hosting service provider. These organisations provide space on their server for an organisa-
tion or individual’s web page. They also offer connectivity of their servers to the Internet.
• Social networking sites (such as Facebook, MySpace, Twitter and LinkedIn) aim to build online
communities of people who share interests. (Social networking is discussed more fully in a later
section.)
• A Wiki is a website that allows the easy creation and editing of a number of interlinked collaborative
pages. The Wikipedia encyclopedia is one of the best-known of these wikis (http://en.wikipedia.
org/wiki/Wiki).
• A Blog (short for weblog) is a website set up by an individual to describe events and materials of
interest to them and relevant to their life.

Development and Design Methodologies and Tools

Web development and design has much in common with the design of other information systems and a
starting point should be in process of analysing and creating information systems (Tatnall, Davey, Bur-
gess, Davison & Wenn, 2002). Creating an information system and getting it working can be considered
to comprise three separate aspects:

• Systems analysis: the description of a logical model of the present or proposed system in non-
technical terms, independent of the actual physical implementation.
• Systems design: the business of converting this logical model into a useable system.
• Systems implementation: involves the construction of the physical system to the level of hardware
acquisition, programming and change management.
xlii

Web Development Issues

As many books describe the development of information systems I will say no more about this here. Web
development, apart from following the logical process of information systems development, involves a
number of other special considerations and I will briefly discuss some of these now (Sklar, 2009).

• Connection speed considerations. Will all your users have high speed broadband connections? If
some are likely to be still using modems then the amount of data to be downloaded needs to be
limited or else it will take too long to load and the potential user will give up.
• Different screen resolutions. Not all users will have the latest high resolution monitors and there
perhaps need to be different version of the site for different monitor resolutions.
• Different operating systems (– Windows, Mac, Linux etc) and different web browsers may make
your site appear differently. This needs to be taken into consideration and portability of design is
a worthwhile goal.
• Design with the user in mind. This seems obvious, but is not always done. Consider the path taken
by the eye in looking around the web page. Consider use of language appropriate to the intended
user. Apart from whether the language is English, French, Spanish, German or Chinese, there is the
issue of complexity – will the intended user understand? Another issue here is ensuring that users
can find their way around inside the website hierarchy. The basic point here is that the website
must be easily usable.
• As different cultures around the world see and understand things somewhat differently, creating
different pages for different countries, as do many commercial websites, is worthwhile both from
the language to be used and because different countries many need slightly different content.
• Accessibility is important. Consideration should be given to people with disabilities such as poor
eyesight, colour blindness, poor hearing etc.
• Programming issues have much in common with those in information systems, except that the
appropriate languages are often different.
• File and data structures – again, approaching this issue has much in common with similar issues
in building an information system.
• Building a website needs graphical design capabilities as well as abilities in information systems
design.
• Security. It is important that appropriate security is built into any website, relevant to the use that
will be made of this site.
• Privacy of users of the website should be respected by not collecting unnecessary personal data
and being careful not to pass on any data that is collected.
• Trust will be built up with website users if the site shows that it deserves such trust.

Web Development Tools

A multitude of different tools and commercial products exist to help with website creation which can
be considered at two levels: design of the underlying code, data and file structures and design of the
appearance of the screen itself. One level needs an information systems professional, the other someone
with graphic design training and abilities. To create the code there are programming languages such
as Java, C++ and Visual Basic and to create or edit graphics there are packages such as Photoshop or
Dreamweaver.
xliii

One web development product is Microsoft Visual Studio with Visual Web Developer which includes
a number of relevant different programming languages such as C++ and Visual Basic, and other tools
useful for website creation. At another level, Microsoft Office Front Page and Expression Web allow
website creation without the need for coding.

Web-Based Applications

Since the start of its rapid growth in popularity in the mid 1990s, a huge number of web-based applica-
tions have been developed. It is, of course, impossible to describe even a small fraction of these but I
will discuss just a few of the most interesting in this section.

E-Business (E-Commerce)

E-Business or electronic business (sometimes called e-Commerce) can be defined as the use of electronic
means on the Web to conduct the business of buying and selling using websites with shopping carts
and e-mail. This can involve many facets of a firm’s operation being conducted electronically between
the business and its partners, suppliers and customers (Davison et al., 2008). This is a very broad topic
containing many sub-areas and represents a major use of the Web.

E-Commerce Business Models

There are several commonly described models that relate the electronic commerce aspects of a business
to other entities of importance:

• Business-to-Business (B-B), for transactions conducted between businesses.


• Business-to-Consumer (B-C), for consumer shopping on the Web.
• Business-to-Government (B-G) is one of several other models used to connect to other entities.
• Business-to-Employee (B-E). Business operations within an organisation (such as B-E) often
make use of the organisation’s intranet.

E-Marketplace

A traditional approach to advertising using well known advertising media such as radio, newspapers and
television is know as the ‘Mass Media Model’. The key to this approach is that marketing campaigns
were typically saturation campaigns that were designed for exposure to a large group of people, with
the hope that a number of them would then purchase the product (Davison et al., 2008). On the other
hand a ‘Hypermedia Computer-Mediated Environment‘ represents a many-to-many communication
model for an environment such as the Web where interaction between businesses can communicate with
customers via a website or automated e-mail system without necessarily conversing with them directly
(Hoffmad & Novak, 1996).
At the business-to-consumer level it is possible to purchase almost anything on the Web. One of the
first major businesses to make use of the web was Amazon.com, launched in 1995, but since that time
many other businesses have looked carefully at this business model and made use of it. The website of
the UK government organisation Businesslink.gov.uk suggests that there are many different types of e-
xliv

Figure 3. E-Business and Web Technologies

marketplace. They note that these are based on a range of different business models and can be broadly
divided into categories as follows:

• Independent e-marketplace – usually a B-B online platform operated by a third party that is
open to buyers or sellers in a particular industry. To participate, some form of payment is usu-
ally required. It allows the business to access requests for quotations or advertisements in their
relevant industry sector.
• Buyer-oriented e-marketplace – normally run by a consortium of buyers to establish an efficient
purchasing environment. This can help the buyers lower their administrative costs and obtain
better prices from suppliers. Suppliers can use this sort of website to place their catalogues for
relevant customers.
• Supplier-oriented e-marketplace – set up and operated by a group of suppliers seeking to
create an efficient sales channel to a large number of buyers. This often provides buyers with
information about suppliers, particularly those they may not be familiar with.
• Vertical e-marketplaces provide online access to businesses of a particular industry sector.
These include the chemical, construction, automotive and textiles industries. The advantage
for suppliers in using a vertical e-marketplace is that this can increase operating efficiency and
help to decrease supply chain costs, inventories and cycle time.
• Horizontal e-marketplaces are able to connect buyers and sellers across different industries
or regions.

Online auctions where people and businesses can buy and sell a variety of goods and services are
another form of e-Marketplace and eBay is a popular example of such a site. On eBay most sales take
place through a set-time auction format, but other approaches are also used.
Many large retail stores also make use of the Web to both advertise and also sell their products. Before
the advent of the Web telephone shopping was quite popular and allowed a number of businesses that
were too small to afford their own retail premises to do business from a private home or other premises.
The e-marketplace of the Web has meant a decrease in shopping of this sort but this has been replaced
xlv

by a huge increase in the use of the Web by these small businesses. For a small business, the costs of
having an online operation can be much lower than setting up a shop and many new ‘virtual enterprises’
are now operating in this area. Paying for purchase or goods or services on the Web can be facilitated
by systems such as PayPal (and Paymate in Australia) that can act as an intermediary providing secure
financial transactions to collect money from the buyer and pass this on to the supplier.

Travel Industry Information and Booking Services

In times past it was necessary to go to a travel agent for booking of hotels and flights, particularly if
they involved international travel. The Web has changed all this with booking and information services
readily available for many hotels and all but the smallest airlines. These services could be considered
under the heading of e-Marketplace, but as they involve more than just the purchase of services and also
provide information I have put them into a category of their own.

E-Banking

Daniel (1999) describes Internet banking as the provision of banking services to customers through
Internet technology. Karjaluoto, Mattila and Pento (2002b) indicate that this involves the provision of
banking services such as accessing accounts, transferring funds between accounts, and offering an online
financial service. Use of web technology in the banking industry can thus be defined as the accessibility

Figure 4. IEEE Computer Society Digital Library Portal


xlvi

and exchange of banking services and/or banking communication technologies on websites available
through the World Wide Web by utilising Internet technology (Al-Hajri, 2005).
Karjaluoto et al. (2002a) then identify two significant benefits that may be gained from Internet
banking:

• Benefits for the banks include a reduction in transactional costs along with the ability to reach
customers anywhere.
• Benefits for customers include being able to take full advantage of various banking services
available online.

Although e-Banking is now common in the developed world, this is not so much the case in develop-
ing countries. A recent exploratory study (Al-Hajri, 2005) of the uptake of Internet banking in Oman (a
developing country) found that in Oman, the bank managers’ perceptions of four issues: relative advan-
tage, organisational performance, customer/organisational relationship and ease of use jointly provided
a broader understanding of Internet technology adoption in the banking industry.

Digital Libraries

A digital library could be described as: “an information system that supports the access, retrieval, selec-
tion, organisation and management of a focused collection of multimedia objects” (Goh, Theng, Lim,
Zhang, Chang, & Chatterjea, 2007). A digital library offers access to collections of selected and organised
digital resources. These resources include books, journal articles and photographs (McCarthy, 2007)
and a digital library’s main advantage over physical libraries lies in their ease of access. Many cultural
and historical associations as well as professional societies such as the IEEE Computer Society (http://
www2.computer.org/portal/web/csdl) and Association for Computing Machinery (ACM) (http://portal.
acm.org/portal.cfm) and the Project Management Institute (http://www.pmi.org/Resources/Pages/) have
their own digital libraries.

Virtual Machines and Locations

Colossus was, arguably, the world’s first electronic digital computer and was used in Bletchley Park
in the United Kingdom to break German military codes towards the end of World War II. A simulation
of using Colossus to break the German Tunny code (Sale, 2006) is available from the Web at: http://
www.codesandciphers.org.uk/anoraks /lorenz/tools/index.htm This is just one of the examples of virtual
machines or operations, another is a simulation of VisiCalc, the first electronic spreadsheet which can
be downloaded from: http://www.bricklin.com/history/vcexecutable.htm
The Web also offers access to a number of virtual tours of interesting locations. One such virtual
tour is of the Italian city, Pompeii: http://www.italyguides.it/us/napoli/ancient_roman_city/virtual_tour_
of_pompeii.htm. Another example is a tour of the Louvre Museum in Paris. This is available at: http://
www.louvre.fr/llv/musee/visite_virtuelle.jsp?bmLocale=en
Each of these examples could have been accessed from files on a purchased CD, but being able to
access them on the Web is yet another reason for the Web’s popularity.
xlvii

Figure 5. VisiCalc executable on a PC

Web Search Engines

The ability to obtain information has in the past involved going to a library and searching for this in
a number of books. While that is still a worthwhile activity, it takes a good deal of time and search
engines like Google, Cuil and Bing offer a much faster alternative. The convenience of using a Web
search engine to find information or the answer to a question is such that people talk about ‘Googling’
the question and the verb ‘to Google’ has entered our vocabulary. There are also related specialist search
engines such as Scholar Google, Google Images and Google Video. Not a lot needs to be said about
search engines as they are so well known, except to remark on their power and on the huge impact they
have had on our lives.

Web Encyclopaedias

An encyclopaedia in book form was (and still is) an extremely expensive purchase that many families
felt obliged to make to help their children with their school work. While the book version is still use-
ful and worth having, the online encyclopaedia has made accessing information ever so much easier.
Products such as Wikipedia (http://en.wikipedia.org) have made a fundamental difference to accessing
information by putting this in everyone’s reach. Other Web encyclopaedia products include Encarta
(http://encarta.msn.com/), Britannica online (http://www.britannica.com/) and Webopedia (http://www.
webopedia.com/).
xlviii

Web Portals

Web portals are now ubiquitous and researching their use in organisations and by individuals is important
(Tatnall, 2007b; Tatnall, 2009d). To illustrate the wide range of sometimes quite specific applications
now being filled by portals, the following list of topic categories (Tatnall, 2005) is taken from articles
by the large number of academics and practitioners from around the world who contributed to the En-
cyclopaedia of Portal Technology and Applications (Tatnall, 2007a).

Portal Concepts, Design and Technology

As one might expect, portal concepts are an area of particular interest with topics such as: what is a
portal? benefits and limitations of portals, comparing portals and web pages, evolution of portals, factors
affecting the adoption of portals using activity theory, information visualisation, the ubiquitous portal,
and portals of the mind (Tatnall, 2009d).
Research on portal design and technology also features prominently with topics such as: collabora-
tive real-time information services via portals, digital interactive channel systems and portals, designing
spatio-temporal portals for continuously changing network nodes, dynamic taxonomies and intelligent
user-centric access to complex portal information, factors affecting portal design, developing semantic
portals, an evolutionary approach to developing on-line learning portals in low bandwidth communities,
the role of ontologies in portal design, Java portals and Java portlet specification API, large-scale ASP
replication of database-driven portals, WSRP specification and alignment with JSR 168 portlet specifica-
tion and user-centric knowledge representation for personalisation of Web portals.

Portal Implementation

As well as the technology itself, there needs to be some consideration of issues involved in the organisa-
tion implementing its portal application (Tatnall, 2009b). Whether the portal is based on a commercial
product or if it is programmed from scratch is itself an important implementation issue. Research issues
involved in the implementation of portals in specific organisations are another point of interest with topics
like the following: evaluation of Web portals, portal quality issues, economical aspects when deploying
enterprise portals, e-management portals and organisational behaviour, user acceptance affecting the
adoption of enterprise portals, enabling technology and functionalities of shopping portals. Articles dealing
with implementation issues often involve a case study of the organisation implementing the portal and
discuss the goals and intentions of using a portal, how the technology was chosen, the implementation
approach used and the problems incurred during the implementation. Not all implementation issues, of
course, are technical ones as implementations all involve people.

Portal Uses and Applications

The largest area of research interest is in how portals are applied and used, and most of this research
refers to quite specific applications such as: the Bizewest portal, the Bluegem portal, the European quality
observatory portal, the future of portals in e-science, hosting portals on an e-marketplace, how corporate
portals support innovation, how the Internet is modifying the news industry, industry portals for small
businesses, portals for business intelligence, strategic planning portals, study of a wine industry portal,
supplier portals in the automotive industry, supply chain management and portal technology, portal
economics and business models, portals for integrated competence management, cultivating memories
xlix

through the Beijing Olympics (2008) Advertainment portal, portals for workflow and business process
management, project management Web portals, provision of product support through enterprise portals,
employee self-service portals, a generic model of an enterprise portal, portal technologies and execu-
tive information systems implementation, the role of portals in consumer search behaviour and product
customisation, guided product selection and comparison of e-commerce portals, business challenges of
online banking portals, Web museums, Web portals as an exemplar for tourist destinations and a Web
portal for the remote monitoring of nuclear power plants (Tatnall, 2009d). Within this area of applica-
tions it is possible to identify a number of major topics of interest. These are:

• Education Portals. Specific example include: academic management portals, large scale inte-
grated academic portals, mobile education portals, artificial intelligence and education portals,
high school portals, primary school portals, corporate e-learning portals, Weblogs, knowledge
portals in education, and subject teaching portals.
• Health and Medical Portals examples include: empowerment and health portals, bioinformatics
portals, biotechnology portals, nursing knowledge portals, network-centric healthcare and the
entry point into the network and genomic and epidemiologic medical data portals.
• Community Portals. Topics researched in this area included: how to promote community
portals, a community geographic domain names portal, designing a portal and community-
community generator, local community Web portals and small businesses and the paradox of
social portals.
• E-Government Portals. This is an area related to Community Portals and there was much
research interest in government portals around the world. There were topics such as: portals in
the public sector, e-government portals, e-value creation in a government Web portal in South
Africa, government portals as a gateway for enhancing electronic governance in Singapore,
interoperability integrating e-government portals, modelling public administration portals, ser-
vice quality in the case of e-government portals, and state portals as a framework to standardise
e-government services.
• National Portals. There is also important research into portals related to national issues: Afri-
can Web portals, business module differentiation and a study of the top three Chinese portals,
cross-cultural dimensions of national Web portals, growth of e-portals in Dubai, how portals
help Chinese enterprises operate successfully in global markets, impacts and revenue models
from Brazilian portals, Web museums and a case study of the French population.
• Personal and Mobile Portals is an area of growing interest as mobile technology continues to
mature. Examples of this area are: accessible personalised portals, mobile portal technologies
and business models, mobile portals as innovations, mobile portals for knowledge management,
the MP3 player as a mobile digital music collection portal, widgets as personalised mini-portals,
wireless local communities in mobile commerce and portals supporting a mobile learning envi-
ronment.
• Knowledge Management, Libraries and Professional Societies. Knowledge Management,
especially relating to libraries and professional societies is another area which attracts a number
of researchers. They were interested in topics such as: designing portals for knowledge work,
mobile portals for knowledge management, knowledge servers, the portal as information broker,
portal strategy for managing organisational knowledge, a prototype portal for use as a knowledge
management tool to identify knowledge assets in an organisation, library portals and an evolving
information legacy, open access to scholarly publications and Web portals, the IFIP portal, and
the portal features of major digital libraries.
l

Intranets

The term Enterprise Information Portal (or sometimes ‘corporate portal’) can be applied to the gate-
ways to corporate intranets that are used to manage the knowledge within an organisation (Davison et
al., 2008). These are often designed for business-to-employee (B-E) processes that offer employees a
means to access and share data and information within the enterprise. An intranet offers a number of
useful applications:

• Access to company documents such as minutes of meetings, forms, policies, procedures, phone
lists, information documents, documentation and archived information.
• Software access and download. Access to common software applications (on a central server),
software downloads, software updates (including anti-virus software).
• Services such as information and communications technology (ICT) technical support and di-
agnosis, training materials and on-line training, human resource data (e.g. leave entitlements),
search facilities to search the site for a specific item.
• Communication – broadcast company information or policy, bulletin boards, chat, electronic
mail, newsgroups, on-line meetings, video conferencing.
• Consistent front-end to software applications – single point of access to other systems, con-
sistent interface to other systems, complexity of databases and applications is hidden, front-ends
to legacy systems.
• Links to outside systems – the Internet, the company extranet (and possibly to business partner’s
systems).

Extranets

Extranets have a number of uses including replacement of direct computer-to-computer Electronic Data
Interchange (EDI). These extranets can use Internet technology to perform the same transactions, although
sometimes rather more slowly. Extranets can also be used to assist with Supply Chain Management. A
supply chain links manufacturers, retailers, customers and suppliers and involves the co-ordination of
order generation, order taking, order fulfilment, order distribution and information sharing. Access by
business partners to parts of the organisation’s intranet can be used to facilitate this process.
An extranet can also be used to facilitate ‘value added marketing’ as the organisation’s Web site
will have basic information of interest to its customers. There may, however, be situations where it
is useful to allow a customer access to its core systems to obtain, for example, account information.
Another example is that a computer supplier may provide (password) extranet access to its own library
for important customers.
Before the Internet, freight delivery companies like DHL (www.dhl.com.au) needed to spend a lot
of time on the phone telling customers the location of their deliveries. It then occurred to someone that
it would be better to let their customers see this for themselves by offering (limited) access to their own
internal systems via an extranet.

Communication

The best know form of communication offered by the Web is, of course, e-mail. This is so well know to
most people that I will say no more about it here except that it was available on the Internet before the
Web came into existence, but was then much more difficult to use.
li

Figure 6. DHL’s parcel tracking extranet

Another form of Web enabled communication is Skype (http://www.skype.com/intl/en/) which is a


software application that enables its users to use the Internet to make almost free phone calls. Skype uses
a proprietary form of Voice over Internet Protocol (VoIP) that allows the call to make use of the Internet
rather than the normal phone system. Skype also offers other features including video conferencing and
instant messaging.
Chat rooms are another form of communication and are offered by a number of companies. These
include: Yahoo (http://messenger.yahoo.com/features/chatrooms), Match.com (http://www.match.com.
au), Google Talk (http://www.google.com/talk/) and Gmail (http://mail.google.com/mail/help/chat.html).
Chat simulates face-to-face communication in that the topic can be anything that the chatters want it to
be and people can come and go at will. One big difference, however, is that Web chat is anonymous in
many cases. In this respect it is a form of communication that is particularly related to the Web.

Group Decision Support Systems

Applications like Google Groups (http://groups.google.com/) and Yahoo Groups (http://groups.yahoo.


com/) offer the possibility for groups of people to easily communicate with each other by e-mail. While
this does not directly offer decision support, it does achieve part of this in allowing a group of people
to communicate easily and discuss the topic under consideration.
A number of researchers have investigated Web-based group decision support systems (GDSS).
Abdelhakim and Shirmohammadi (2007), for example, discuss a Web-based GDSS that can be used to
select and evaluate educational multimedia and Ma, Quan and Zhiping (2009) describe a Web-based
GDSS for Research and Development project outcome assessment in government funding agencies. The
company DSS Resources (http://dssresources.com/) offers more product advice in this area.
lii

The Social Web

The Web has also opened up a wide range of possibilities for people to socialise remotely in various
ways. Whether socialising via the Web is a good thing or not rather depends on your point of view and
I will come back to discuss this later.
Facebook (http://www.facebook.com/) is perhaps the most commonly known social networking
Web site. It is designed so that people can join a network organised by their school, workplace, region
or other similar entity and then interact with other people in this same network. They can then create
their own personal profile, notify other people about themselves and send and receive messages from
these people.
MySpace (http://www.myspace.com/) is another social networking Web site that offers similar
facilities to Facebook.
LinkedIn (http://www.linkedin.com/) is a business-oriented social networking Web site that is mainly
used for professional, rather than personal, networking.
Twitter (http://twitter.com/) is another social networking Web site that is used to send and receive
short (140 character) text messages called tweets. These are displayed on their author’s profile page and
sent to other subscribers (know as followers). Twitter users can send and receive their tweets on their
mobile phones using SMS (Short Message Service).
Another more specific example is Academia.edu (http://www.academia.edu/) which is designed, as
its name suggest, to link academics and researchers.
YouTube (http://www.youtube.com/) is a video sharing Web site where users can upload and share
videos they have made. These can be viewed on a PC in MPEG-4 format.

e-Learning

Electronic learning (e-Learning) involves support of learning using Web technology as the medium for
teaching. In some cases, especially where the students are at a location that is remote to the teaching
institution, it is used to replace face-to-face teaching entirely. In other cases it is used to supplement
classroom learning. This form of teaching is used in universities and some schools, but also for training
in many large companies, and particularly in companies whose operations span the globe. The largest
provider of e-Learning applications is Blackboard Inc (http://www.blackboard.com/) whose products
include WebCT and Blackboard. These products typically offer the facility for discussion boards, e-mail
systems, live chat, Web pages and downloading content including documents. An alternative open source
e-learning package is available from Moodle (http://moodle.org/).
For the student, one big advantage of e-learning is that they do not need to attend the teaching institution
in person but can undertake their studies from home or their place of employment. When combined with
mobile computing the possibilities for undertaking learning at any time and any location are great.

Mobile Computing and the Web

While mobile computing relates not only to the Web but also to other aspects of computing, it is particu-
larly relevant to discuss this here as it offers the possibility of viewing Web content at any location. Few
people would now question the idea that computers and communications devices can be small enough
and mobile enough to use anywhere, but this has not always been the case. The first computers were huge
machines filling whole rooms and weighed several tonnes and telephones were anything but portable
until quite recently. Laptop and net computers, Personal Digital Assistants (PDA) and smart phones
liii

are now so small as to be extremely portable, but it is not just the hardware here that is significant but
also the protocols for short-range radio communications that they require. There are two main existing
standards for short-range radio communications (Davison et al., 2008):

• IEEE 802.11 (WiFi) is a set of standards for wireless local area networks, developed by IEEE,
that operates in the 5 GHz and 2.4 GHz public spectrum bands over a 50-100 metre radius at a
maximum data transfer rate of up to about 50 Mbps.
• Bluetooth offers a 10 metre range with a peak data transfer speed of 1-3 Mbps. It is possible to
operate at least ten Bluetooth devices within a 10 metre radius. Bluetooth provide a means of link-
ing digital devices into Personal Area Networks (PAN)

Each of these protocols offers the possibility for a portable device to link to other portable devices
or to access the Internet at almost any location. In city areas a number of retail premises (including
McDonalds) offer free Internet WiFi ‘hot spots’ where customers (and those nearby) can gain access to
the Internet for no cost using an IEEE 802.11 protocol in their laptop. Other locations, including most
hotels, also offer wireless Internet access but at a cost.
The important point here is that these mobile devices, in conjunction with a WiFi protocol, make it
possible to access the Web from almost any populated location.

Web-Based Gambling

While not permitted in some parts of the world, in others online gambling is perhaps being used at times
to circumvent local gambling laws or at least to make it much easier for gamblers to indulge in this ac-
tivity. Sites such as Web Gamble (http://www.webgamble.org/) and Wager Web (http://www.wagerweb.
com/) are just a few of many sites that offer this facility. The Casino Gambling Web site (http://www.
casinogamblingweb.com/) provides news reports and information about online gambling. For those with
a gambling problem Gamblers Anonymous also has its own Web site (http://www.gamblersanonymous.
org/).

Organisational, Managerial, Social and Political Impact of


Web Technologies

Organisational and Managerial Impacts

Are organisations different to manage when they make extensive use of the Web? The answer is in most
ways that they are not, but there are some issues of which a manager must take account. I will mention
just a few. The first is the ease of access to information by both employees and customers. An organisa-
tion can make good use of the ease with which it and its employees can gain information, but sitting an
employee in front of a computer to search can also be a very time wasting exercise and it is important
that proper goals and time limits are set for this type of activity. That customers can also easily obtain
information about competitors and other products can be a problem which must be recognised by a busi-
ness. Steps need to be taken to become aware of this problem and to seek means to overcome it
At a different level another managerial issue is in the use of an organisational intranet. It is quite easy
to set up an intranet that can be used to store all sorts of company documents such as: policies, forms,
procedures, minutes of meetings, internal phone lists and so on. While in the past it was necessary to
liv

distribute these documents in paper form they can now be put on the intranet and made easily available
to all staff. There will now be no excuse for someone claiming ignorance of a company procedure or
policy because they could not get hold of a copy of it. The intranet can also be used to broadcast mes-
sages from the CEO or other senior managers to all employees. If not abused this can be a very useful
facility. Updates and patches for computer software can also be delivered in this way, making it possible
for the organisation to keep much more up to date.
It is sometimes claimed that if some information, such as a catalogue or price list, is on the organisa-
tion’s Web site then it must be up to date. This is, of course, absolute rubbish and a better claim is that
if it is on the organisation’s Web site then there is the possibility that it is always kept up to date. This
possibility will only eventuate however, if it is someone’s job to keep information on the company Web
site up to date.
Another exciting management possibility with use of the Web is the creation of a virtual enterprise.
Such an organisation does not need a shop front to interact with its customers but can order and supply
its products remotely. Managing such an enterprise, where some employees operate from one location
and others operate from different locations is quite different to running a business located under one
roof and needs quite a different management style.

Social and Political Impacts

In common with the introduction of most new technologies, the advent and growth of the Web comes
with the dual possibilities of good or ill ey and & Tatnall, 2007). Along with freedom in the transmission
of knowledge comes a loss of control for authorities and the possibility of chaos. The introduction of
Guttenberg’s printing press in circa 1455 allowed mass production of the Bible and eventually changed
the nature of Christianity in Europe. For many years printing was the technology used to distribute in-
formation. The came the telephone, and this technology allowed people to keep in contact remotely and
to distribute news and information. The Web furthered this democratisation of information and allowed
people to spread news and to keep in touch without the need to go through any official or governmental
channels. Censorship and control of the Internet in some countries is another hot issue that is unlikely
to go away. George (2005) suggests that the relationship between new technological media and political
factors is far too dynamic and interdependent to be reduced to simple causal statements. He suggests
that the less democratic the society, the more attractive the Internet looks as an emancipatory medium,
but the more likely radical Internet use will be blocked or punished.
The role of the Internet in mobilising and assisting various protest movements is an interesting topic.
Stohl and Ganesh (2009) note the importance of the Internet for spreading information, organising and
constructing networks and forming individual and group identities. They also suggest that it can even
act as a form of protest itself. An article in the Wall Street Journal (Qiang, 2005) discussed the role
of the Internet in organising protests in China in 2005. The article noted that “the most fervently pro-
nationalist segment of society overlaps demographically in China with those who spend the most time
on the Internet, since both are primarily composed of young, educated, urban males” and goes on to
describe how recent protests were almost exclusively organised using the Web.
The Internet & Democracy Blog (http://blogs.law.harvard.edu/idblog/2009/04/07/moldovan-youth-
organize-protests-with-twitter/) describes how Moldovan youth organised protests with the aid of Twitter.
It describes how Twitter not only helped rally protesters but also provided a real glimpse of what was
happening on the ground. Debate also has been progressing on the role of technology in Thailand’s recent
protests (2009) http://neteffect.foreignpolicy.com/posts/2009/04/17/technologys_dubious_role_in_thai-
lands_protests.
lv

How use of the Web affects democracy and political activities is an interesting and evolving study.
One example is that in recent elections in Australia both the Prime Minister (at the time) and the then
Leader of the Opposition made considerable use of YouTube to spread their message. Similar use of
technology also occurred in other countries and no doubt we will see an increase in the use of Web tech-
nologies of all types in the political process. A detailed study in Korea undertaken by Woo-Young (2005)
found that citizen e-participation was characterised by: convenient access to detailed information, free
expression and exchange of opinions, online activism led by politicised agenda and active formation of
cyber groups. Woo-Young notes that Internet political portals are not connected with existing political
power and capital, and that they facilitate communication between citizens.
Hutchins and Lester (2006), however, suggest that the use of the Web by environmental groups is an
example of mob rule. The existence of a Web portal does not indicate the number of people subscribing
to the philosophy of its owners. The Web site allows coordination of efforts, such as protests, in a way
that may make the actions of a group look more important than the size of the group might support.
Francis Bacon is credited with the expression ‘Knowledge is Power’ and the growth in use of Web
technologies has brought this to the fore with the creation of new classes of information rich and in-
formation poor depending on their access to this technology. Access to huge amount of information is
now possible, much than anyone could have dreamed of a few years ago. With access to the Web many
people now also have access to encyclopaedias, news, research results and much more. Those without
access to the Web miss out.

Undesirable Impacts

There are also many undesirable impacts of the growth of the Web. Many people would see the increase
in Internet gambling as one of these and most would see as a bad thing the Web’s role in making much
easier the dissemination of pornography. Theft and fraud resulting from improper Internet financial
transactions are a major problem, and identity theft has been made easier by the increase in general use
of the Web.
Another problem in general use of the Web is the increased possibility of misinterpretation of infor-
mation. Medical Web sites abound on the Internet with almost every major disease being represented by
at least a support group portal. These Web sites offer everything from emotional support and possible
treatment advice to contacts within the medical community (Davey & Tatnall, 2007), but this is not
always seen as useful and Theodosiou and Green (2003) identify five important problems with patients
using medical portals to satisfy their needs:

• Potentially dangerous drugs and other substances may be bought by individuals for themselves or
their children.
• Individuals can spend a lot of money on products or diagnostic procedures that have no scientific
backing or benefit.
• The information may be more negative than the reality of the situation.
• Individuals may abandon treatment programmes of proven efficacy to pursue less-mainstream
approaches.

Information that had previously been sold by its originators soon began to be dispersed without charge
via the Web and the owners of the software, music and videos being freely distributed invented the term
Internet piracy. Yar (2005) indicates the size of the possible loss to copyright owners in terms of billions
of dollars. Plagiarism is another evil that has been made much easier and more common by use of the
lvi

Web. When it is so easy for a student to copy a passage from a journal article found in a digital library
and paste it into their essay it is little wonder that plagiarism has become a major problem in universities
and other educational institutions.
Finally, there is the question of whether spending large periods of time in front of a computer should
be seen as a non-human activity that discourages normal social interaction. While the use of social
networking sites offers many possibilities for social interaction of a sort, unless this is followed up by
face-to-face contact of the normal type there could be something unhealthy about this use of the Web.

The Socio-Technical Nature of Web Systems

Sometimes those of us who design information systems of any type get lost in the elegance and intrinsic
worth of our design efforts and forget that for an information system to have any value, it must first be
adopted and put into use. There is an old Anglo Saxon saying that ‘You can lead a horse to water, but
you can’t make it drink’, and something similar can be said about the adoption and use of information
systems. Because of their human and non-human aspects, Web-based information systems should be
considered as socio-technical entities, and it is most important that both of these aspects are kept in
mind throughout the process of systems design and implementation. This is especially true of Web-based
information systems.
Building any type of information system is a difficult task, partly due to the problems in ascertaining
the requirements of the intended users, but also because of the complexity of the large number of human-
machine interactions (Tatnall & Davey, 2005) that such systems typically involve. This complexity is
reflected in the difficulty of building these systems to operate free from error and to perform as intended
(Tatnall, 2009a), and building Web based systems is no less complex.
The Information Systems (IS) discipline is primarily concerned with the ways that people build and use
computer-based systems to produce useful information. It thus always has to deal with issues involving
both people and machines, and with the multitude of human and non-human entities that comprise an
information system (Tatnall, 2003). The discipline of Information Systems is thus neither merely a technical
discipline nor a social one, but one that is truly socio-technical. Systems developers and researchers all
face the problem of how to handle the many complexities due to interconnected combinations of people
along with computers, peripherals, procedures, operating systems, programming languages, software,
broadband connections, switching technology, data, databases and many other inanimate objects. They
need to consider how all these elements relate to humans and human organisations, and how humans
relate to them (Longenecker, Feinstein, Couger, Davis & Gorgone, 1994; Tatnall, 2009a).

Adoption of Web Technology: Innovation

Just because a new Web technology exists it cannot automatically be assumed that organisations or in-
dividuals will want to adopt or to use it. A Web technology will only be adopted if potential users make
a decision to do so and the adoption of a technological innovation, such as a Web technology, occurs for
a variety of different reasons. Thus the first step in researching the use of a new Web technology by an
organisation (or an individual) is to investigate why it was adopted, and so consider the Web technology
as a technological innovation. This can be done by examining the adoption of the new Web technology
through the lens of innovation theory.
lvii

It is important at this stage to distinguish between invention and innovation. Invention refers to the
construction of new artefacts or the discovery of new ideas, while innovation involves making use of
these artefacts or ideas in commercial or organisational practice (Maguire, Kazlauskas & Weir, 1994).
Invention does not necessarily invoke innovation and it does not follow that invention is necessary
and sufficient for innovation to occur (Tatnall, 2009b). Clearly the Web technology can be seen as an
invention, but the point here is that it will not be used unless it is adopted, and that means looking at it
also as a technological innovation. Of course, the application of innovation theory to the adoption of a
new Web technology assumes that the potential adopter has some choice in deciding whether or not to
make the adoption. In the case of an organisation or individual considering the adoption and use of a
Web technology, however, it is difficult to see any reason why they would not have a large measure of
choice in this adoption decision. This makes the application of adoption theory quite appropriate when
considering the use of Web technology.
A number of approaches exist to modelling how technological innovation takes place, including the
Theory of Reasoned Action (Ajzen & Fishbein, 1980), the Theory of Planned Behavior (Ajzen, 1991),
the Technology Acceptance Model (Davis, 1986), Diffusion of Innovations (Rogers, 1995; Rogers, 2003)
and Innovation Translation (Callon, 1986b; Latour, 1996). In particular, both the Diffusion of Innova-
tions and the Technology Acceptance Model (TAM) are very well known and widely used approaches
to theorising technological innovation.

Researching Web Technology and its Applications

This multi-volume book series contains a large number of research articles on a variety of topics relating
to Web technologies, and in this section I will look at how research in Web technology can be categorised
(Tatnall, 2009b) and how it is being undertaken. While some of articles on Web technology are fairly
descriptive, perhaps examining a commercial product in some depth, others involve detailed analytical
research. It appears to me that most of the articles outlining research on Web technology could be clas-
sified into three main areas which could, of course, be further subdivided:

• Web technology: research on the technical design of computer hardware and software for Web
applications.
• Web implementation: research issues involved in the implementation of Web sites in specific
organisations.
• Web applications: research into the many uses of the Web, and the social and political consequences
of this use.

Web technology itself has been researched by many scholars, especially those involved in the design
and implementation of Web-based systems. This is a wide field and offers many possibilities for fairly
technical research articles dealing with various aspects of these systems. Commercial Web technology
products and vendors each have an important role to play and their investigation and evaluation provides
a profitable avenue of research. Quality issues and standards as well as measurement of effectiveness
could also be validly considered. Another important consideration here is whether certain implementa-
tion factors are more likely to lead to successful adoption of Web technology than others. I will take
this point further shortly.
As well as the technology itself, there needs to be some consideration of issues involved in the organi-
sation implementing its Web-based application. Whether the Web technology is based on a commercial
lviii

product (such as Front Page) or if it is programmed from scratch (in a programming language such as
Java or Visual Basic) is itself an important implementation issue. Articles dealing with implementation
issues often involve a case study of the organisation implementing the Web-based application and discuss
the goals and intentions of using this: how the technology was chosen, the implementation approach
used and the problems incurred during the implementation. Not all implementation issues, of course,
are technical ones as implementations all involve people.
No doubt the largest area, however, consists of articles that investigate Web-based systems uses and
applications, of which there are many. A major user of Web technology around the world is govern-
ments and the public sector and most government departments now have their own Web sites. Social
and community-based Web sites are also common. At the personal level, research is often conducted
into topics including Weblogs, widgets and MP3 players. Medical, health and bio-informatics Web sites
form another significant group of applications as do the business and industrial sectors. Much research
investigates organisational and management issues regarding Web technology use, human resources
and enterprise information portals, Web sites for small to medium enterprises and more specific topics
including shopping, the automotive industry and wine industry Web sites. The economics of setting up
and using these Web sites is also discussed, as are issues of strategic planning, user acceptance, security
and the law. Applications in areas such as business intelligence, artificial intelligence, semantic portals,
intelligent agents and mobile technology could also be included here.
The approach used in Web-based systems research depends, of course, on the type of research being
undertaken and will most likely differ depending on whether this is related to Web technology itself,
implementation or applications. Should a purely descriptive article be considered as research? It is not
the place of this chapter to discuss this, but the sort of article I will discuss involves at least some degree
of analysis. Articles considering specific Web technology products and vendors will most likely look at
some form of analysis of the benefits of this feature or that design approach and may be quite general
in nature. As mentioned previously though, apart from articles discussing Web technology itself, most
research will in some way relate to the use of this Web technology by some particular organisation and
so will probably involve some form of case study.
One research approach that attempts to take proper account of both the social and technical aspects
of Web-based information systems is that of Soft Systems Methodology (SSM), a variant on action re-
search developed by Peter Checkland and his colleagues (Checkland & Scholes, 1991) from Lancaster
University. SSM attempts to give due recognition to both the human and technological aspects of a
system and acknowledges both human and non-human aspects of these systems, which it considers as
entirely separate types of entity.
Another research approach is actor-network theory (ANT) that also considers both social and tech-
nical aspects of an information system. ANT, however, considers that it is not possible to distinguish
between these two types of entities and so attempts to consider and handle both human and non-human
entities in the same way.
Actor-network theory proposes that everything we do involves hybrid entities (Latour 1993) contain-
ing both human and non-human elements. ANT was developed to analyse situations where separation
of these elements is difficult (Callon 1999; Tatnall and Gilding 1999). One could question, for instance,
which parts of an information system are just inanimate objects and which are the result of human in-
teractions. If we consider a Web portal, for example, it is difficult to differentiate its technical aspects
from the influence exerted by the socio-cultural background of the development team (Cusumano &
Selby, 1997; Sahay, 1997). What seems, on the surface, to be social is partly technical, and what may
appear to be only technical is partly social.
lix

ANT handles this difficulty in determining the social-technical divide by refusing to allow the pos-
sibility that either purely technical or purely social relations can exist. It instead offers the notion of
heterogeneity to describe situations such as the development of a website or other web-based system that
involves computer hardware, computer software, communications hardware, communications software,
telecommunications authorities, cables, portal hosting sites, Internet service providers, data, human
analysts, human programmers, human users and so on. The use of heterogeneous entities that can be
partly social and partly technical (Bijker, Hughes & Pinch, 1987) then avoids questions of: ‘is it social?’
or ‘is it technical?’ as missing the point, which should be: “is this association stronger or weaker than
that one?” (Latour, 1988, p. 27). ANT considers both social and technical determinism to be flawed and
proposes instead a socio-technical account (Latour, 1986; Law & Callon, 1988) in which neither social
nor technical positions are privileged.

“ANT was developed to analyse situations in which it is difficult to separate humans and non-humans,
and in which the actors have variable forms and competencies.” (Callon, 1999, p:183)

In order to address the need to treat both human and non-human actors fairly and in the same way,
ANT was designed upon three fundamental principles (Callon, 1986b):

• Agnosticism: analytical impartiality is demanded towards all the actors involved in the project
under consideration, regardless of whether they are human or non-human.
• Generalised symmetry: explains the conflicting viewpoints of different actors in the same terms
by use of an abstract and neutral vocabulary that works the same way for human and non-human
actors. Neither the social nor the technical elements in these ‘heterogeneous networks’ (Law, 1987)
are then given any special explanatory status.
• Free association: requires the abandonment of all distinctions between the technological and the
social (Callon, 1986b; Singleton & Michael, 1993).

Actor-network theory thus attempts impartiality towards all actors in consideration, whether human
or non-human, and makes no distinction in approach between them. Callon (1986b, p. 200) puts it this
way: “The rule which we must respect is not to change registers when we move from the technical to
the social aspects of the problem studied.” Callon (1987) further proposes that entities gain strength
by gathering a ‘mass of silent others’ into a network to give them greater strength and credibility. This
network then becomes durable partly due to the durability of the bonds that hold it together.
In ANT, an actor is any human or non-human entity that is able to make its presence individually felt
by the other actors (Law, 1987). It is made up only of its interactions with these other actors (de Vries,
1995), and Law (1992) notes that an actor thus consists of an association of heterogeneous elements
constituting a network. It should not be considered as a ‘point object’ but rather as an association of
heterogeneous elements themselves constituting a network. Each actor can thus also be considered to
constitute a simplified network (Law, 1992). In this sense, an actor can in many ways also be considered
as a black-box, and when the lid of the box is opened it will be seen to constitute a whole network of
other, perhaps complex, associations (Callon, 1986a). In many cases, details of what constitutes an ac-
tor – details of its network – are a complication we can avoid having to deal with all the time. We can
usually consider this entity just as an actor, but when doing this it must be remembered that behind each
actor there hide other actors that it has, more of less effectively, drawn together (Callon, 1987).
After development, as technological innovations are often not adopted in their entirety or in exactly
the form that their proponents suggested, ANT makes use of a theory of Innovation Translation (Latour,
lx

1986; Law & Callon, 1988; Latour, 1996) which suggests that before adoption, an innovation is first
‘translated into a form which is more appropriate for use by the potential adopter. Callon et al. (1983)
propose that translation involves all the strategies through which an actor identifies other actors and
reorganises them in its own way.
The main advice on method suggested by the proponents of actor-network theory is to “follow the
actors” (Latour, 1996) and let them set the framework and limits of the study themselves. The process
followed in an ANT analysis is thus far from linear, and not just simply a matter of collecting data and
then analysing it.
I will argue that two things make actor-network theory very appropriate for researching and discussing
web-based systems and for framing all but the type of research that only describes the technology itself.
First is its treatment of both human and non-human actors and of the interactions between them. There
is no doubt, in my view, that both human and non-human actors including the hardware, software and
communications technologies themselves, jointly effect the outcome of implementing any web-based
system. To consider one and ignore the other would give a very one sided view of what is going on.
For example, suppose that a particular technology is too expensive or too difficult to use. There is then
little likelihood of it being adopted and used properly. On the other hand suppose that this technology
does not fit with the way that a particular organisation does its business. This is not necessarily due to
the nature of the technology alone any more than it is due alone to the human factors of the way that
business is done. It is more likely due to the interaction of each of these factors. In another situation,
what one person or organisation considers the perfect technological solution will not necessarily suit
another. The reasons it might not suit them could be complex and have more to do with cultural or social
issues than technical issues. A consideration of both human and technological aspects of these systems
then is essential.
Another advantage of using ANT is in its treatment of the adoption of innovations. Most other ap-
proaches to innovation adoption consider only whether some technology is adopted or not and do not
give put much store on partial adoption. When investigating this area it is often the case that one person
or company will adopt certain features of Web technology and not others that it could have been expected
to adopt. There will then often be another similar company that adopts other different features of the
same technology. One must then question the reason for this, and actor-network theory’s approach using
Innovation Translation has the ability, I would suggest, to explain this well. ANT suggests that before
adoption, a technological innovation must first be ‘translated’ into a form appropriate for adoption by
the organisation concerned. It so offers the means for explaining partial adoptions. In my own research
I have found actor-network theory a very useful research methodology in working with socio-technical
systems involving Web technology (Tatnall, 2009c).

Issues and Trends: Conclusion

It is always dangerous to attempt any predictions into the future of technology. Thomas Watson (Senior),
then President of IBM, is alleged to have stated in 1943: “I think there is a world market for maybe five
computers”. Whether Watson made this statement of not, and there is little evidence that he actually
did, it highlights the danger in making future predictions related to information and communications
technologies. I am, however, I think fairly safe in saying that for the future of the Web three things seem
fairly clear.
lxi

1. The Web will continue to grow in size as move users (particularly in developing countries) move
to adopt web technologies and more information is put onto the Internet. No doubt this will put
considerable pressure on the infrastructure, but with new technologies this problem should be
overcome.
2. The speed of access to the Web will continue to increase as governments around the world put
in new higher speed broadband infrastructure. This will mean that websites can contain even
more multi-media material.
3. More people will access the Web with the aid of mobile computing technologies, meaning that
the Web will become more accessible everywhere.

Questions that do not have a clear answer are whether access to most websites will continue to be
free of cost and whether governments around the world will keep their hands off the Web or want to
exercise a greater degree of control as some governments now do. Given that the Web has been around
for only about 15 years it will be interesting to see what a book about the Web in 15 years time will
say – perhaps there will then be no need for such a book as the Web will have become so well know
that there will be no need for one.

References

Abdelhakim, M. N. A., & Shirmohammadi, S. (2007). A Web-Based Group Decision Support System
for the Selection and Evaluation of Educational Multimedia. International Multimedia Conference
Augsburg, Germany ACM.
Ajzen, I. (1991). The Theory of planned behavior. Organizational Behavior and Human Decision Pro-
cesses, 50(2), 179-211.
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. London,
Prentice-Hall, Englewood Cliffs.
Al-Hajri, S. (2005). Internet technology adoption in the banking industry. Business. PhD Thesis, Victoria
University, Melbourne. .
Barras, C. (2009). Internet wonders: How big is the Net? New Scientist, 202, 30.
Bijker, W. E., Hughes, T. P., & Pinch, T. J. (Eds.) (1987). The social construction of technological sys-
tems: new directions in the sociology and history of technology. Cambridge, MA: MIT Press.
Callon, M. (1986a). The sociology of an actor-network: The case of the electric vehicle. In M. Callon,
J. Law, & A. Rip (Eds.), Mapping the dynamics of science and technology (pp. 19-34).
Callon, M. (1986b). Some elements of a sociology of translation: Domestication of the scallops and the
fishermen of St Brieuc Bay. In J. Law (Ed.), Power, action & belief. A new sociology of knowledge? (pp.
196-229). London: Routledge & Kegan Paul.
Callon, M. (1987). Society in the making: The study of technology as a tool for sociological analysis.
In W. E. Bijker, T. P Hughes, & T. P Pinch (Eds.), The social construction of technological systems (pp.
85-103). Cambridge, MA: The MIT Press..
lxii

Callon, M. (1999). Actor-network theory - The market test. In J. Law & J. Hassard (Eds.), Actor network
theory and after (pp. 181-195). Oxford: Blackwell Publishers..
Callon, M., Courtial, J. P., Turner, W. A., & Bauin, S. (1983). From translations to problematic networks:
An introduction to co-word analysis. Social Science Information, 22(2), 191-235.
Checkland, P., & Scholes, J. (1991). Soft systems methodology in action. Chichester: Wiley.
Cusumano, M. A., & Selby, R. W. (1997). How Microsoft builds software. Communications of the ACM,
40(6), 53-61.
Daniel, E. (1999). Provision of electronic banking in the UK and the Republic of Ireland. International
Journal of Bank Marketing, 17(2), 72-82.
Davey, B. and Tatnall, A. (2007). The paradox of social portals. In A. Tatnall (Ed.), Encyclopedia of
Portal Technology and Applications (pp. 689-693). Hershey, PA, Information Science Reference.
Davis, F. D. (1986). A technology acceptance model for empirically testing new end-user information
systems: Theory and results. Unpublished doctoral thesis, MIT, Boston.
Davison, A., Burgess, S., & Tatnall, A. (2008). Internet technologies and business. Melbourne: Data
Publishing.
de Vries, G. (1995). Should we send Collins and Latour to Dayton, Ohio? EASST Review, 14(4).
George, C. (2005). The internet’s political impact and the penetration/participation paradox in Malaysia
and Singapore. Media, Culture & Society, 27(6), 903-920.
Goh, D. H.-L., Theng, Y.-L., Lim, E.-P., Zhang, J., Chang, C. H., & Chatterjea, K. (2007). Learning
geography with the G-Portal Digital Library. In A. Tatnall (Ed.), Encyclopaedia of Portal Technology
and Applications (pp. 547-553). Hershey, PA, Information Science Reference. Hoffman, D. L., & No-
vak, T. P. (1996). A new marketing paradigm for electronic commerce. Retrieved 2009, from http://
sloanwordpress.ucr.edu/1996/10/17/working-paper-hoffman-and-novak-february-1996
Hutchins, B., & Lester, L. (2006). Environmental protest and tap-dancing with the media in the informa-
tion age. Media, Culture & Society, 28(3), 433-451.
Karjaluoto, H., Mattila, M., & Pento, T. (2002a). Electronic banking in Finland: Consumer beliefs and
reactions to a new delivery channel. Journal of Financial Services Marketing, 6(4), 346-361.
Karjaluoto, H., Mattila, M., & Pento, T. (2002b). Factors underlying attitude formation towards online
banking in Finland. International Journal of Bank Marketing, 20(6), 261-272.
Latour, B. (1986). The Powers of association. In J. Law (Ed.), Power, action and belief. A new sociology
of knowledge? Sociological Review monograph 32 (pp. 264-280). London: Routledge & Kegan Paul.
Latour, B. (1988). The Prince for machines as well as for machinations. In B. Elliott (Ed.), Technology
and social process (pp. 20-43). Edinburgh: Edinburgh University Press.
Latour, B. (1993). We have never been modern. Hemel Hempstead, Harvester Wheatsheaf.
Latour, B. (1996). Aramis or the love of technology. Cambridge, MA: Harvard University Press.
lxiii

Law, J. (1987). Technology and heterogeneous engineering: The case of Portuguese expansion. In W. E.
Bijker, T. P. Hughes, & T. J. Pinch (Eds.), The social construction of technological systems: New direc-
tions in the sociology and history of technology (pp. 111-134). Cambridge, MA: MIT Press.
Law, J. (1992). Notes on the theory of the actor-network: Ordering, strategy and heterogeneity. Systems
Practice, 5(4), 379-393.
Law, J., & Callon, M. (1988). Engineering and sociology in a military aircraft project: A network analysis
of technological change. Social Problems, 35(3), 284-297.
Lawrence, E., Newton, S., Corbitt, B., Braithwaite, R., & Parker, C. (2002). Technology of Internet
business. Sydney: John Wiley & Sons.
Longenecker, H. E. J., Feinstein, D. L., Couger, J. D., Davis, G. G., & Gorgone, J. T. (1994). Information
Systems ‘95: A Summary of the Collaborative IS Curriculum Specification of the Joint DPMA, ACM,
AIS Task Force. Journal of Information Systems Education, Winter 1994-95, 174-186.
Ma, J., Zhang, Q., & Fan, Z. (2009). Web-based group decision support for R&D project outcome as-
sessment in government funding agencies. Retrieved 2009, from http://www.is.cityu.edu.hk/Research/
WorkingPapers/.
Maguire, C., Kazlauskas, E. J., & Weir, A. D. (1994). Information services for innovative organizations.
San Diego, CA: Academic Press.
McCarthy, C. (2007). Portal features of major digital libraries. In A. Tatnall (Ed.), Encyclopaedia of
Portal Technology and Applications (pp. 724-736). Hershey, PA: Information Science Reference.
Network Working Group. (1974). Specification of Internet transmission control program. Retrieved
2009, from http://tools.ietf.org/html/rfc675
O’Hara, B., & Petrick, A. (2005). 802.11 Handbook: A Designers Companion. Piscataway, NJ: IEEE
Standards Association.
Qiang, X. (2005). China’s first Internet-organized protests. The Wall Street Journal. New York.
Rogers, E. M. (1995). Diffusion of innovations (4th ed.). New York: The Free Press.
Rogers, E. M. (2003). Diffusion of innovations (5th ed.). New York: The Free Press.
Sahay, S. (1997). Implementation of information technology: A space-time perspective. Organization
Studies, 18(2), 229-260.
Sale, T. (2006). Codes and ciphers in the Second World War. Retrieved July 2006, from http://www.
codesandciphers.org.uk
Singleton, V., & Michael, M. (1993). Actor-networks and ambivalence: General practitioners in the UK
cervical screening programme. Social Studies of Science, 23(2), 227-264.
Sklar, J. (2009). Principles of Web design. Boston, MA: Course Technology, Cengage Learning.
Stohl, C. and Ganesh, S. (2009). Understanding relationships between the Internet and protest strategies
and frames in the antiglobalisation movement. Retrieved 2009, from http://www.allacademic.com/meta/
p172140_index.html
lxiv

Tatnall, A. (2003). Actor-network theory as a socio-technical approach to information systems research.


In S. Clarke, E. Coakes, M. G. Hunter, & A. Wenn (Eds.), Socio-technical and human cognition elements
of information systems (pp. 266-283). Hershey, PA: Information Science Publishing.
Tatnall, A. (2005). Web portals: From the general to the specific. 6th International Working for E-Business
(We-B) Conference, Melbourne, Victoria University.
Tatnall, A., Ed. (2007a). Encyclopedia of portal technology and applications. Hershey, PA: Information
Science Reference.
Tatnall, A. (2007b). The ubiquitous portal. In A. Tatnall (Ed.), Encyclopaedia of portal technology and
applications (pp. 1040-1044). Hershey, PA: Information Science Reference.
Tatnall, A. (2009a). Actor-network theory applied to information systems research. In M. Khosrow-
Pour (Ed.), Encyclopedia of information science and technology (2nd ed.) (pp. 31-35). Hershey, PA: IGI
Global.
Tatnall, A. (2009b). Gateways to portals research. International Journal of Web Portals, 1(1), 1-15.
Tatnall, A. (2009c). Information systems, technology adoption and innovation translation. International
Journal of Actor-Network Theory and Technological Innovation, 1(1), 59-74.
Tatnall, A. (2009d). Web portal research issues. In M. Khosrow-Pour (Ed.), Encyclopedia of information
science and technology (2nd ed.) (pp. 3642-3647). Hershey, PA: IGI Global.
Tatnall, A. and Davey, B. (2005). A new spider on the Web: Modelling the adoption of Web-based
Training. In P. Nicholson, J. B. Thompson, M. Ruohonen, & J. Multisilta, (Eds.), E-training practices
for professional organizations. Assinippi Park, MA: Kluwer Academic Publishers. Tatnall, A., Davey,
B., Burgess, S., Davison, A., & Wenn, A. (2002). Management information systems - Concepts, issues,
tools and applications. Melbourne: Data Publishing.
Tatnall, A., & Gilding, A. (1999). Actor-network theory and information systems research. 10th Austral-
asian Conference on Information Systems (ACIS), Wellington, Victoria University of Wellington.
Theodosiou, L., & Green, J. (2003). Emerging challenges in using health information from the Internet.
Advances in Psychiatric Treatment, 9, 387-396.
Wave Report. (2001). IEEE 802.11 Standard Tutorial. Retrieved May 2004, from http://www.wave-
report.com/tutorials/ieee80211.htm.
Woo-Young, C. (2005). Online civic participation, and political empowerment: Online media and public
opinion formation in Korea. Media, Culture & Society, 27(6), 925-935.
Yar, M. (2005). The global ‘epidemic’ of movie ‘piracy’: Crimewave or social construction? Media,
Culture & Society, 27(5), 677-696.

Endnote
1
Conseil Eroeen pour la Recherche Nucleaire (CERN) – an international scientific organisation
based in Geneva, Switzerland.
lxv

About the Editor

Arthur Tatnall (BSc, BEd, DipCompSc, MA, PhD, FACS) is an associate professor in the Graduate
School of Business at Victoria University in Melbourne (Australia). He holds bachelor's degrees in sci-
ence and education, a graduate diploma in computer science, and a research MA in which he explored
the origins of business computing education in Australian universities. His PhD involved a study in cur-
riculum innovation in which he investigated the manner in which Visual Basic entered the curriculum
of an Australian university. He is a member of three IFIP working groups (WG3.4, WG3.7 and WK9.7)
and is also a fellow of the Australian Computer Society. His research interests include technological in-
novation, information technology in educational management, information systems curriculum, project
management, electronic commerce, and Web portals. He has written several books relating to information
systems and has published numerous book chapters, journal articles and conference papers. He recently
edited Encyclopedia of Portal Technology and Applications for IGI Global.
Section I
Fundamental Concepts
and Theories

This section serves as the foundation for this exhaustive reference tool by addressing crucial theories
essential to the understanding of Web technologies. Chapters found within these pages provide an
excellent framework in which to position Web technologies within the field of information science and
technology. Individual contributions provide overviews of the mobile Web, semantic Web, and Web 2.0,
while also exploring critical stumbling blocks of this field. Within this introductory section, the reader
can learn and choose from a compendium of expert research on the elemental theories underscoring
the research and application of Web technologies.
1

Chapter 1.1
ICT and Interculture
Opportunities Offered
by the Web
Laura Corazza
Università di Bologna, Italy

Background objective to strive for; a situation in which the people


of different cultures begin to open up to reciprocal
In Italy, as in other European countries such as relationships, thereby bringing about the possibil-
Germany and France, the words “multicultural” ity of integration, characterised by a political and
and “intercultural” have distinct meanings1. In some cultural pluralism which aims at reciprocal respect
ways the passage the one from the other indicates and appreciation between the respective cultures”2.
the evolution of a concept that, when examined in With Multiculturalism, people accept to live with
detail, forces us to re-examine educational political, those from other cultures with tolerance but with-
and cultural choices in a society. out promoting forms of exchange and democratic
“In Italy, especially in the field of pedagogy, co-existence. The form of social organization that
the term multicultural indicates a situation where derives from this is that of the “melting pot” which
cultures co-exist side by side but as yet, have not encourages the development of ghettos or “Little
found a means of comparison and relating to each Italy” and “China towns”. However the Intercultural
other . <…>. The term intercultural, which origi- approach is reciprocal. It is for those who accept
nated in France and then spread to other European and listen, those who are tolerant, those who are
countries, on another hand, describes a situation not afraid of “contamination” but constantly seek
of interaction between different cultures, often de- to mediate between different points of view and
scribing a prospect, a possible point of arrival or an backgrounds.

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
ICT and Interculture Opportunities Offered by the Web

Among various other factors, it is globaliza- well as obtain it, they can teach as well as learn
tion, encouraged by new communication and and participate not only as an individual, but as
information technologies, that has contributed to part of a group. From the moment a person be-
the transformation of contemporary world-wide comes aware of this new communicative reality
society into a multicultural society. These tech- and has the possibility of taking part in it, he or
nologies have made it possible to communicate she possesses new rights and also responsibili-
easily and at low cost with every corner of the ties. The rights include the opportunity to access
planet. We can observe events which are taking the Web and its contents independently of limits
place all over the world and take part in collective (economic, time, and movement) and diversity
cultural processes. The Internet is made up of in- (cultural, social, cognitive, or physical). To this we
terconnected nodes without a centre. It encourages can add the right to learning. This must take into
the growth of new trans-national communities consideration, not only the acquisition of techni-
on various levels, ranging from the political and cal and procedural knowledge but also cognitive,
professional to the scientific. It produced some emotive and social competences which are needed
elements of standardization, such as the use of a in order to participate with full awareness in a
single language, English and has lead to uncom- form of communication which is also building
mon social contacts on a worldwide level. At the knowledge. There also the ever present problem
same time however, these new communication of standardized positions, so the individual must
technologies can also be viewed as a further be aware of personal expectations and needs, and
cause of the divide between those who can ac- adopt a critical, constructive attitude and be able
cess information and participate in the creation to question, make hypotheses, choices and checks.
of knowledge and those who cannot. Moreover, it is necessary to do this in a social
setting which requires emotional control and an
ability to collaborate, mediate and negotiate.
Focus The communicative environment is that of the
Web where the passage between the exchange
The concept of freedom of access to information of information and the building of knowledge is
is an integral part of the philosophy and history potentially fast, thanks to the network that allows
of the Internet. It is also guaranteed by the char- for an exchange rather than a one-way flow of
acteristics of computer mediated communication information. This environment is one of demo-
and tools such as e-mail, forums, mailing lists, cratic co-existence based on mutual respect. It
blogs and portals. CMC has changed the concept appreciates individuality and cultural and social
of communication itself. Leaving behind the one- diversity. At the same time the right to access
way logic of communicator- receiver (typical of is also accompanied by certain responsibilities
traditional mass media) the situation has become towards others. Anyone who takes part in this
one where individuals are nodes in a network, part communicative process becomes an adherent
of an interconnected environment. The individual of the philosophy of Internet. As a direct conse-
has the active role of social participant in com- quence of its genesis and evolution, it appears to
munication rather than being a passive consumer be more closely associated with the technology
(as in the case of television). of freedom rather than that of control, and with
In addition to this, there are a number of new democratic relationships rather than ideological
opportunities available for the user. It is now imperialism.
possible to influence the information circuit There are various types of free information on
significantly. People can provide information as the web today. They can be placed into two catego-

2
ICT and Interculture Opportunities Offered by the Web

ries: new forms of news agencies and new forms with each other in a novel way on the Internet
of publishing. Both fit into to the intercultural and since they can be more than mere passive users
democratic view which requires free information, and isolated consumers like today’s television
originating “from below” that is, from all the links viewers. Communication is not limited to “one to
in the network, combined with fast transmission one” exchange. In cyber culture it is reciprocal,
speeds and a high level of diffusion. Examples of interactive and communitarian. Moreover it is
the first category are blogs and more organised universal as anyone can be an active communica-
agencies such as Misna, an international press tor. The expansion of interconnections displays
agency run by congregational missionaries in the fact that there is one general humanity. This
Africa, Asia, Latin America and Oceania. Though is what Levy means by “universal”: the idea of
possessing few resources it can boast many suc- universality in the Enlightenment sense of the
cesses and has earned “recognition after only a word, where our goal is the unity of human kind
brief period of activity including the award of the and the affirmation of universal principles such as
‘St Vincent 2002’ journalism prize”3. the rights of Man. All human beings can come into
The archives of scientific literature (papers, reciprocal contact virtually and become conscious
journal articles, addresses made at conferences) collectively of their existence. The more we add
are another important example of freely available links, the more diverse and heterogenic material
information on the Net. Offered in an electronic is circulated on the net. Humanity is becoming
format at no cost, by the very universities that aware of itself (universality) but this awareness
finance the research and believe in the principle of has no one meaning. It does not pass through
free access to information, they are potentially use- one central point and is not bound by any set of
ful to everyone from college students in America laws unlike science where, for example, universal
to the doctor in a hospital in Tanzania gravitation is the same everywhere. What is in-
Thus we can state the Internet’s contribution teresting in cyber-culture is the bringing together
to interculturalism has manifested itself in three of all differences and heterogeneity
ways: the creation of a cyber-culture; in a new Internet as a location for new knowledge which
concept of knowledge; and in the participation is open, complex, multidisciplined, individual,
“from below” in the building of learning: and collective at the same time.
Internet as a place of communicative interac- Morin (2000) has highlighted the problem of
tion and universal culture. the inadequacy of knowledge (which is divided
For Lévy, what is interesting about cyber- by the boundaries between disciplines while the
culture is the coming together of all differences reality of the world is increasingly global and
and heterogeneity. He believes in the ability of interconnected) and the challenge of complexity.
cyberspace to bring out the best of human intel- Complexity is a method and a form of knowledge
ligence in every person through communicative which requires a dialogical approach. Gregory
relationships which create a collective intelligence Bateson also spent his life demonstrating the
and a cyber-culture. In this sense we can assert interdependence between elements and the
that cyber-culture is both a transculture and an interconnections between different worlds and
inteculture, since it is universal without being disciplines. He theorises about the ecology of
oppressively uniform. ideas, an ecosystem in which there is a plurality
If we take cyberspace to be the location, of levels of application of ideas, using a system-
then communication is the means that permits atic approach which, from the point of view of
the creation of this collective intelligence and learning means giving importance to contexts,
cyber-culture. Computer users can communicate relationships and functions.

3
ICT and Interculture Opportunities Offered by the Web

So the new idea of knowledge, does not only information they have found on the Web. It may
refer to concepts that are to be transmitted but, be incomplete or taken from dubious sites, but
above all, to the itineraries and the network of they begin their treatment in a more knowledge-
individual and collective experiences which are able and participative way. Even though there are
in a state of perpetual change. limits and the information is certainly not enough
This is the challenge that the individual can to provide specialist knowledge, it is enough to
take up, not withstanding the risks of cognitive modify the power/knowledge rapport in the doc-
standardisation on one hand, and fragmentation tor/patient relationship.
or individualism on the other. Morin’s “tête bien The potential for forming opinions, and thereby
faite” requires the full use of intelligence in the expressing judgements and making demands
organisation of ideas, in the search for links, routes aided by the Web, is becoming more apparent in
and new itineraries, knowing full well that it is the political development of citizens. People can
not possible to attain totality but that we must put participate in public debates as they once did in
into practice collective and individual cognitive the classical agora. They can inform themselves
practices at the same time because they exploit and take up positions through blogs, messaging
the power of connectivity. “If a document is put systems and online journalism.
on the World Wide Web you are doing two things If we say that technology is not neutral and
at the same time: firstly, you are increasing the the choice of technology produces profound ef-
amount of information that is available, but sec- fects, the most evident effect of the Internet is the
ondly, you are doing another thing with the links transition from vertical to horizontal communi-
between your document and the others: you offer cation. The first has an undeniably authoritarian
the surfer who finds your document your point of nature, even if the intentions of those using it
view. So you are not merely offering information are democratic, since it requires a silent, passive
but a point of view in a collection of information. viewing. The second however, permits active
The World Wide Web is not only an enormous mass participation, and while this alone is not enough
of information, it is the expression of thousands to confer unquestionably democratic credentials,
of different points of view. It should be viewed it certainly transforms the quality of communica-
from this aspect” (Levy, 1995). There is space for tion. In general, the passage from one-way com-
every diversity and every point of view on the Web munication to interactivity does not automatically
which doesn’t have to become individualism, if guarantee a growth in democracy. On the contrary
you allow yourself to become enchanted by hyper- it can increase the use of extorted consensus to
textual logic, by the links and maps and if you legitimise solutions taken without the effective
allow yourself to feel part of a whole, a collective participation of citizens.
intelligence and a universal culture. The key question is thus the following: Can the
Internet as a place for forming an opinion, Internet be used to manipulate consensus? Is there a
expressing a critical judgement, and participating social use of new technologies? Up to now citizens
in the building of knowledge. have been subjected exclusively to vertical infor-
A surprising example of the impact of the mation offered by television, which creates passiv-
Internet on the power of the individual and thus ity in viewer. The condition of citizens using the
of the group, is the increase in public access to web, benefiting from horizontal communication
health information. This is not just another step which blurs the distinction between producers and
in the spread of knowledge, but a political change consumers and the supremacy of the former over
which introduces a new balance of power. More the later, is totally different. Everyone becomes
and more patients arrive at their doctor’s with a supplier of information, and as such, an active

4
ICT and Interculture Opportunities Offered by the Web

player in communication. The most marginalized Current Limits and


of Internet users can have word power equal to Future Developments
that of a government, especially when the news
they wish to communicate is information that Technology of Freedom or
others want to suppress at any cost. Censorship Technology of Control?
becomes much more difficult, but with one condi-
tion: that there is a real mass internet information Whether we consider technology to be neutral, or
literacy program and a true universal service. something that can structure the user’s activities,
Without this there is the risk that an information we cannot avoid reflecting on the responsibilities
apartheid emerges. Moreover, there is also the of those who use it and the need for user-education.
danger of hyper-information, where citizens are Whilst there is a great potential for helping human-
unable to pick out the important information and ity, we face a number of new questions. A critical
check its validity due the potentially unlimited analysis can reveal both positive and negative
quantity of information available. It is for this aspects, opportunities and risks, and a possible
reason too that the role of education in permitting direction to take in order to achieve a democratic
a social use of new technologies is increasing. It use for the civic education of the citizen. The key
has an important part to play in training active, question is that posed by Stefano Rodotà: Are we
knowledgeable citizens, to guarentee an adequate using the technology of freedom or the technology
command of the tools necessary for developing of control? Can the use of technology be free from
the competences required by society. rules? Is it necessary to reflect on which direction
According to Rifkin, cultural production is the we should be striving for?
first level where the economic life of a country We need to free ourselves from some vices of
is played out. In consequence of this, there is a form and clichés that linger, such as technological
strong move to control such production in the arrogance (and the idea that technology does not
direction of a standardised predetermined models need rules), the optimism of the market (which
(Rifkin, 2000). in reality favours economic interests over social
But if it is true that the Internet is disliked equality) and the political-ideological simplifica-
by governments because it cannot be censored tion which sees technology as a cure for all evils
or controlled easily, it can also become a tool in (in reality technology is not neutral but requires
the democratic education of citizens, if the move careful use)
towards liberty prevails over that towards stan-
dardisation. A conscious citizen is a person who The Digital Divide
participates in the building of a free, universal
society which is constantly developing and chang- Regarding access to the new communication
ing. The process of building knowledge that can technologies and the internet in particular, there
come about through the use of the network of are no equal opportunities between the techno-
ideas and information available on the Internet logically advanced rich countries and countries
produces social progress. This is what Dewey which are technologically behind. In a message
calls “social efficiency”: everything that makes of 2002, the secretary general of the United Na-
an experience valid for the community, as well as tions Kofi Annan underlined the importance of
oneself; everything that makes it communicable communication technologies and exhorted poor
and useful in the demolition of social barriers. The countries on the African continent to unite to create
same efficiency that he considers to be the final a digital revolution which had become indispens-
aim of education in a democratic society. able. In collaboration with the ITU International

5
ICT and Interculture Opportunities Offered by the Web

telecommunications union the UN organised the consider the technical changes in the exchange of
first World Summit on the Information Society, information and the spread of knowledge, but also
WSIS with the aim of building a common vision for the political, social and educational implica-
of the information society and adopt a plan of tions. In substance we have to start to conceive
action to bring it about4. and use new democratic environments for the
The first principle of the Geneva declaration production of culture, new forms of diffusion
states: which offer efficient tools for communication
between individuals, disciplines and political and
We, the representatives of the peoples of the social points of view. To use a more traditional
world, assembled in Geneva from 10-12 Decem- pedagogical language, what emerges from the
ber 2003 for the first phase of the World Summit analysis of the spontaneous (and in some cases
on the Information Society, declare our common experimental) use of the internet is the necessity
desire and commitment to build a people-centred, to use the great potential for communicative and
inclusive and development-oriented Information didactic renewal which is beginning to manifest
Society, where everyone can create, access, utilize itself, in order to produce a culture that overcomes
and share information and knowledge, enabling individualism and is oriented towards forms of
individuals, communities and peoples to achieve collaboration which widen the opportunities
their full potential in promoting their sustainable for democratic participation in the information/
development and improving their quality of life, knowledge process. By moving in this direc-
premised on the purposes and principles of the tion perhaps we can cement the relationships
Charter of the United Nations and respecting between individuals, groups, states, and political
fully and upholding the Universal Declaration organisations providing a glimpse of the added
of Human Rights. value of cooperation, reciprocity, peace, and thus
interculture.
It is not enough to be connected to resolve the
fundamental problems of underdevelopment and
to ensure that the Information Society becomes References
a vehicle for democracy, justice, equality and
respect for individuals and their personal and Castells, M. (2002a). La nascita della società in
social development. Beyond the mere physical rete. Milano: Università Bocconi.
availability of a computer, other factors such as Castells, M. (2002b). Galassia Internet. Milano:
economic/social resources and levels of literacy, Feltrinelli.
influence the successful use technology. It is the
development of the necessary competences and Castells, M. (2002c). Rivoluzione tecnologica
a fruitful, informed use of the internet that is the e identità, intervista a “Mediamente”, Rai Edu-
challenge for the current Information Society and cational, URL: http://www.mediamente.rai.it/
lifelong learning in particular. articoli/20020530b.asp.
Corte, M. (2006). Comunicazione e giornalismo
interculturale. Pedagogia e ruolo dei mass media
Conclusion in una società pluralistica. Padova, Cedam.

In conclusion Internet is a wonderful opportunity Dewey, J. (1954). Il mio credo pedagogico. Anto-
(comparable with the impact on culture of the logia di scritti sull’educazione, trad. it. a cura di
invention of the printing press) not only when we Lamberto Borghi, Firenze, La Nuova Italia.

6
ICT and Interculture Opportunities Offered by the Web

Dewey, J. (1967). Scuola e società, Firenze, La Rodotà, S. (2001). Tecnologia e democrazia, inter-
Nuova Italia. The school and the society. Chicago: vista a: “Enciclopedia multimediale delle scienze
The University of Chicago Press. filosofiche. Il Grillo”, prodotto da Rai Educational,
puntata del 08/01/2001, URL: http://www.emsf.
Dewey, J. (1992). Democrazia e educazione,
rai.it/grillo/trasmissioni.asp?d=742
Firenze, La Nuova Italia. Democracy and educa-
tion: An introduction to the philosophy of educa- Rodotà, S. (2004). Tecnopolitica (2d ed.), Roma,
tion. New York: Macmillan. Laterza.
Genovese, A. (2003). Per una pedagogia in- Rodotà, S. (2006). Sviluppo telematico e democra-
terculturale. Dalla stereotipia dei pregiudizi zia, intervista a:”MediaMente”, Roma, 13 gen-
all’impegno dell’incontro, Bologna, BUP. naio, URL: http://www.mediamente.rai.it/home/
bibliote/intervis/r/rodota02.htm#link001.
Harnad, S. (2003). For whom the gate tolls? In
D. Law & J. Andrews (Eds.), Digital libraries: Web sites Misna,www.misna.org
Policy planning and practice. Ashgate.
Willinsky, J. (2003). Democracy and education:
Lévy, P. (1995a). L’intelligenza collettiva, interv- The missing link may be ours. Harvard Educa-
ista a “Mediamente”, Parigi - European IT Forum, tional Review, 72(3), 40.
4 settembre, URL: http://www.mediamente.rai.it/
Willinsky, J. (2003). The nine flavours of open
HOME/bibliote/intervis/l/levy.htm
access scholarly publishing. Journal of Postgradu-
Lévy, P. (1996). L’intelligenza collettiva. Per ate Medicine, 49(3), 263–267.
un’antropologia del cyberspazio. Milano, Fel-
Willinsky, J. (2005). Scientific research in a
trinelli.
democratic culture: Or what’s a social science
Lévy, P. (1999). Cybercultura. Gli usi sociali delle for? Teachers College Record, 107(1), 38–51.
nuove tecnologie. Milano, Feltrinelli. doi:10.1111/j.1467-9620.2005.00455.x
Misna, (Missionary International Service News
Agency) provides daily news from for and about
the South of the World. Key Terms
Morin, E. (1985). Le vie della complessità, in Cyberculture: For Lévy, what is interesting
G. Bocchi e M. Ceruti (a cura di), La sfida della about cyberculture is the coming together all
complessità, Milano, Feltrinelli. differences and heterogeneity. He believes in
Morin, E. (2000). La testa ben fatta. Riforma the ability of cyberspace to bring out the best of
dell’insegnamento e riforma del pensiero, Milano, human intelligence in every person through com-
Cortina. municative relationships which create a collective
intelligence and a cyberculture. In this sense we
Open Archives Iniziative. http://www.openar- can assert that cyberculture is both a trans-culture
chives.org and an interculture, since it is universal without
Pinto Minerva, F. (2002). L’intercultura, Roma- being oppressively uniform.
Bari, Laterza.
Democratic Education: Morin has highlight-
ed the problem of the inadequacy of knowledge

7
ICT and Interculture Opportunities Offered by the Web

(which is divided by the boundaries between disci- People can provide information as well as obtain
plines while the reality of the world is increasingly it, they can teach as well as learn and participate
global and interconnected) and the challenge of not only as an individual but as part of a group.
complexity. Complexity is a method and a form of From the moment a person becomes aware of this
knowledge which requires a dialogical approach. new communicative reality and has the possibility
Gregory Bateson also spent his life demonstrating of taking part in it, he or she possesses new rights
the interdependence between elements and the and also responsibilities. The rights include the
interconnections between different worlds and opportunity to access the web and its contents in-
disciplines. So the new idea of knowledge, does dependently of limits (economic, time, and move-
not only refer to concepts that are to be transmitted ment) and diversity (cultural, social, cognitive or
but, above all, to the itineraries and the network physical). To this we can add the right to learning.
of individual and collective experiences which This must take into consideration, not only the
are in a state of perpetual change. The process of acquisition of technical and procedural knowledge
building knowledge that can come about through but also cognitive, emotive and social competences
the use of the network of ideas and information which are needed in order to participate with full
available on the internet produces social progress. awareness in a form of communication which is
This is what Dewey calls “social efficiency”: also building knowledge.
everything that makes an experience valid for
the community, as well as oneself; everything Intercultural Education: In Italy, as in other
that makes it communicable and useful in the European countries such as Germany and France,
demolition of social barriers. The same efficiency the words multicultural and intercultural have
that he considers to be the final aim of education distinct meanings. In Italy, especially in the field
in a democratic society. of pedagogy, the term multicultural indicates a
situation where cultures co-exist side by side but
Democratic Information: The concept of as yet, have not found a means of comparison
freedom of access to information is an integral and relating to each other. The term intercultural,
part of the philosophy and history of the inter- which originated in France and then spread to other
net. It is also guaranteed by the characteristics European countries, on the other hand, describes a
of computer mediated communication and tools situation of interaction between different cultures,
such as e-mail, forums, mailing lists, blogs and often describing a prospect, a possible point of
portals. CMC has changed the concept of commu- arrival or an objective to strive for; a situation in
nication itself. Leaving behind the one-way logic which the people of different cultures begin to open
of communicator-receiver (typical of traditional up to reciprocal relationships, thereby bringing
mass media) the situation has become one where about the possibility of integration, characterised
individuals are nodes in a network, part of an in- by a political and cultural pluralism which aims
terconnected environment. The individual has the at reciprocal respect and appreciation between
active role of social participant in communication the respective cultures.
rather than being a passive consumer (as in the
case of television).

Information Society: It is now possible to


influence the information circuit significantly.

8
ICT and Interculture Opportunities Offered by the Web

Endnotes The only exception is intercultural com-


munication.
1
In North American and the majority of 2
Genovese A. (2003). Per una pedagogia in-
English speaking culture however the term terculturale. Dalla stereotipia dei pregiudizi
multicultural is most widely used (especially all’impegno dell’incontro, Bologna, BUP, p.
in education). This can be observed when 181.
consulting ERIC, the bibliography data base 3
http://www.emi.it. Description of the agency
of the U.S. Department of Education which MISNA, under the heading of “cards,
specializes in topics relating to pedagogy profiles,” where they are actually exposed
and didactics and representative of North the various associations that work with
American scientific literature. In its thesau- MISNA.
rus intercultural education is only considered 4
http://www.itu.int/wsis/
as a synonym for multicultural education.

WEB SITES

The Open Archives Initiative develops and promotes interoperability standards that aim to facilitate the
efficient dissemination of content. The Open Archives Initiative has its roots in an effort to enhance ac-
cess to e-print archives as a means of increasing the availability of scholarly communication. Continued
support of this work remains a cornerstone of the Open Archives program. The fundamental technological
framework and standards that are developing to support this work are, however, independent of the both
the type of content offered and the economic mechanisms surrounding that content, and promise to have
much broader relevance in opening up access to a range of digital materials. As a result, the Open Archives
Initiative is currently an organization and an effort explicitly in transition, and is committed to exploring
and enabling this new and broader range of applications. As we gain greater knowledge of the scope of
applicability of the underlying technology and standards being developed, and begin to understand the
structure and culture of the various adopter communities, we expect that we will have to make continued
evolutionary changes to both the mission and organization of the Open Archives Initiative.

Public Library of Science,http://www.plos.org/

PLoS, a nonprofit organization of scientists and physicians committed to making the world’s scientific
and medical literature a freely available public resource. As of 2006 it publishes PLoS Biology, PLoS
Medicine, PLoS Computational Biology, PLoS Genetics and PLoS Pathogens. PLoS ONE is a new
journal to be launched soon.

Our goals are (1) to open the doors to the world’s library of scientific knowledge by giving any scientist,
physician, patient, or student—anywhere in the world—unlimited access to the latest scientific research.
(2) Facilitate research, informed medical practice, and education by making it possible to freely search
the full text of every published article to locate specific ideas, methods experimental results, and obser-
vations. (3) Enable scientists, librarians, publishers, and entrepreneurs to develop innovative ways to
explore and use the world’s treasury of scientific ideas and discoveries.

9
ICT and Interculture Opportunities Offered by the Web

PubMed Central (PMC), http://www.pubmedcentral.nih.gov/

PubMed is a free search engine offering access to the MEDLINE database of citations and abstracts of
biomedical research articles. It is offered by the United States National Library of Medicine as part of
the Entrez information retrieval system. MEDLINE covers over 4,800 journals published in the United
States and more than 70 other countries primarily from 1966 to the present.

Ricerche di Pedagogia e Didattica,http://rpd.cib.unibo.it

Freely accessible journal of the university of Bologna on pedagogical and didactic topics

Unione Europea, http://europa.eu/index_it.htm

Portal: “Recommendations of the European Parliament and Council regarding key competences for
lifelong learning”, 11/20/2005.

This work was previously published in Encyclopedia of Information Communication Technology, edited by A. Cartelli; M.
Palma, pp. 357-364, copyright 2009 by Information Science Reference (an imprint of IGI Global).

10
11

Chapter 1.2
Mobile Social Web:
Opportunities and Drawbacks
Thorsten Caus
Georg August University of Göttingen, Germany

Stefan Christmann
Georg August University of Göttingen, Germany

Svenja Hagenhoff
Georg August University of Göttingen, Germany

Abstract Introduction

As mobile Internet usage continues to grow, the Until recently, the Internet was a domain restricted
phenomenon of accessing online communities to stationary computers, but nowadays it can also be
through mobile devices draws researchers’ atten- accessed through mobile devices equipped with web
tion. Statistics show that close to 60 percent of all browsing capabilities. Now it is not only possible
mobile Internet traffic worldwide is related to the to surf the web using wireless access and mobile
use of mobile social networks. In this chapter, the devices, but there is also a growing number of
mobile social Web is defined, categories of mobile mobile Internet applications and services. Increas-
communities explained, and success factors and ingly, mobile social networking applications have
drawbacks discussed from the technical, social, been made available to a large number of mobile
and economic perspectives. Challenges, including phone users.
low transmission rates, changes in usage patterns, Internet users have accepted online communi-
search for new revenue sources, as well as the need ties and internalized the concept of the Social Web
for development of original mobile Web content and also referred to as Web 2.0 (Koesch, Magdanz, &
applications are addressed. The technical require- Stadler, 2007). Private as well as business users have
ments for the mobile use of online communities become familiar with various online communities
are identified. The chapter closes with a summary (Patrzek, 2007; von Tetzchner, 2008). On the one
of potential economic and social prospects of the hand, mobile social networks are becoming more
emerging mobile social Web. widespread because of the increasing dissemination
of new wireless communication technologies (Heng,
2006, p. 5). On the other hand, a large number of
DOI: 10.4018/978-1-60566-368-5.ch002 devices are designed to implement new communi-

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Mobile Social Web

cations technologies, for example, the Universal have dealt with the Social Web. Various attempts
Mobile Telecommunications System (UMTS) in to provide a definition have resulted in three dif-
Europe (Heng, 2006, p. 1). ferent approaches: technical, social, and economic.
Studies conducted by Opera Software, an The technical approach focuses on the Internet
Internet software and services company, demon- as a medium or platform for a community. The
strate that 40 percent of all mobile Internet traffic sociological point of view stresses the forming
worldwide is related to the use of online com- and functioning of communities, whereas the
munities. In some countries the share is as high economic perspective examines potential gains
as 60 percent: for example, in the United States, and intended profits (Hummel, 2005, p. 8-11).
South Africa, and Indonesia (von Tetzchner, 2008). These three perspectives have led to a variety
Research into the various ways of using the Social of definitions of online communities with differing
Web in a mobile context is now of paramount points of emphasis. A detailed overview of com-
importance. In this chapter, mobile social web is mon definitions is given by Fremuth and Tasch
defined, categories of mobile online communities (2002), Hummel (2005) and Markus (2002). In
and their success factors explained, and selected identifying an online community one perspec-
opportunities and drawbacks of the mobile online tive emphasizes that it is formed by a group of
communities discussed from a technical, social, people, while another stresses its web platform.
and economic perspectives. The definition used in this chapter combines both
approaches, for an online community is seen as a
social group that interacts through a web platform
Background over an extended period of time.
An online community can be characterized by
The Social Web can be viewed as a concept and a four elements (Gebert & von Rosenstiel, 1992, p.
platform that utilizes social software (e.g., forums, 122-123; Hamman, 2000, p. 225):
wikis, blogs, etc.) to fulfill or support some of the
important human needs, such as: self-realization, • group of people with shared objectives
acceptance, social connectedness, and safety (e.g., interests, goals)
(Maslow, 1943, p. 372-383). The purpose of the • interaction over an extended period of
Social Web is to support human communication time
and facilitate social contact. The Social Web • closeness due to bonds and relationships
encompasses numerous Internet applications, • shared space for interactions governed
such as social networking sites, massively mul- by certain rules (for example, role
tiplayer online role-playing games, photo and definitions).
video sharing, online stores and auction houses,
virtual worlds, and wiki collaborations. The most Without shared objectives there would be no
popular and widespread actualizations are online interaction and relationship and, subsequently, no
communities (e.g., MySpace, Facebook, StudiVZ community at all (Markus, 2002, p. 36). Interac-
or XING). The term “Social Web” is often used tions within the community are seen as topic-
in everyday language as well as in scholarly oriented communication as well as the execution
literature as a synonym for “virtual” and “online of actions (Kim, 2000, p. 5). Both can take place
communities” (Hummel, 2005, p. 5), although independently of time and location (Winkler &
these terms do not differ greatly (Fremuth & Mandl, 2004, p. 14). The process of founding
Tasch, 2002, pp. 5-6). and maintaining such online communities usually
In the past years many academic disciplines takes place on the Internet (Eigner & Nausner,

12
Mobile Social Web

2003, p. 58). The second defining characteristic of interruption caused by the movement of the
is the web platform, which can be seen as an user from one location to another. The devices
Internet communication system which acts as an employed can be either wireless or wired (Kurose
intermediary. It enables and facilitates meetings, & Ross, 2005, p. 504). Suitable types of devices
the maintenance of the community, and its inter- are mobile phones, smart phones (Michelsen &
action with other people (Reichwald, Fremuth, Schaale, 2002, p. 51) and personal digital assis-
& Ney, 2002, p. 8). tants, as these can always stay switched on and do
There are different ways of categorizing online not need to be booted. Although mobility does not
communities (Brunold, Merz, & Wagner, 2000, p. necessarily require wireless connections to these
30-37; Fremuth & Tasch, 2002, p. 21; Hummel, devices, wireless mobile networks are used most
2005, p. 46). A reasonable approach is to catego- commonly (Gerum, Sjurts, & Stieglitz, 2003, p.
rize them according to similarities, for they play a 145). European standards, for instance, include
major role in online communities. Therefore, an the Global System for Mobile Communication
online community can be geographic (bound to (GSM) extensions, Enhanced Data Rates for GSM
an area), demographic (classification according Evolution (EDGE) standard (Herzig, 2001, p. 399),
to nationality, age, gender), or based on shared General Radio Packet Service (GPRS) protocol
interests or activities (Kim, 2000, p. 5). (Stader, 2001, p. 37), and the 3rd generation
Online communities can be viewed as social Universal Mobile Telecommunications System
systems. Relationships and interactions can only (UMTS) technology (Kurose & Ross, 2005,
develop once a web platform has been established, pp. 534-535). The mobile social web involves
which makes it difficult to start a community (Leit- opportunities as well as drawbacks, as will be
ner, 2003, p. 36). The network-effect character discussed below.
of online communities shows this very clearly.
There will only be accelerated growth once a Opportunities of the
critical amount of relationships and interactions Mobile Social Web
between users has been achieved. This is due to
the fact that users do not benefit before this point The features of mobile communities generally
is reached (Reichwald, Fremuth, & Ney, 2002, correspond to those of traditional online com-
p. 8). Even though this development is difficult munities. They are enhanced by a new way of
to predict, an operator is able to influence the accessing the community-web-platform through
development of an online community by making mobile, wireless devices. On closer examination,
it more attractive (Reichwald, Fremuth, & Ney,
2002, p. 9-10). The success factors that have been
Table 1. Success factors on online communities
identified are listed in Table 1.
(adopted from Koch, Groh, & Hillebrand, 2002;
Leitner, 2003)

THE MOBILE SOCIAL Web     Corporations’ point of


    Users’ point of view
view

The mobile use of online communities can be re- • Advantages of usage, for • Personal network and personal
example in the form of prob- characteristics of entrepreneur-
ferred to as the mobile social web. In this context, lem solving or entertainment
• ial team
• Product or service
mobility is understood as the unrestricted transfer Simple compilation of contri- idea in business model
• Avail-
butions
• Easy technical access, able resources and capabilities
•
of text, voice or data independent of user’s physi- usage and adequate stability
• Marketing strategy with viral
cal location (Kurose & Ross, 2005, pp. 536-538). Equality, credibility and trust
• emphasis
• Potential for speedy
Non-commercial orientation marketing
Therefore, mobile online communities are free

13
Mobile Social Web

these communities do not only seem to benefit device then serves as a membership card or an
from mobile access, but also from additional po- individual entrance ticket (Reichwald, Fremuth,
tentials resulting from mobility and localization & Ney, 2002, p. 7). Identifying the user can be
possibilities. Furthermore the question arises, automated, making it more reliable. It is also more
whether this newly opened potential encompasses authentic and results in more confidence within
not only technical and social aspects but economic the community (Hummel, 2005, p. 72).
aspects as well (see Figure 1).
Social Aspects
Technical Aspects
A local context can be defined by identifying a
Mobile wireless devices facilitate ubiquitous user’s current whereabouts (Koch, Groh, & Hille-
access to online communities (Koch, Groh, & brand, 2002, p. 3). A user’s whereabouts could be
Hillebrand, 2002, p. 2). The user is able to gain presented to other users through geomapping or
access anywhere and any time, while being textually. Furthermore, an awareness service could
liberated from the world of stationary Internet be employed in addition to existing information
and permanently installed devices (Reichwald, services, such as “buddy lists” (Tasch & Brakel,
Fremuth & Ney, 2002, p. 6). Furthermore, users 2004, p. 7). Contacts and friends in the vicinity
benefit from being reachable at all times by being of the user can be shown on the mobile device.
armed with mobile wireless devices (Wiedmann, Mobile online communities could thus improve
Buckler, & Buxel, 2000, p. 86). Therefore, the interactions between community members and
users can stay online and be accessible continu- extend social ties. For example, people with
ously and without interruption. No time is lost: similar interests can get together spontaneously.
for example, by turning the devices on or logging Contexts can be used to find out about the acces-
in. Thus, interaction becomes more spontaneous sibility of mobile community users (Groh, 2003, p.
and expressive. 9). Depending on the user’s current whereabouts
Mobile wireless devices make it possible to a service can provide information about if and
identify the user via his PIN and SIM card, both how a person may be contacted and the user can
of which are requirements of mobile communica- decide how to do this. The contact information
tions (Tasch & Brakel, 2004, p. 4). This explicit can simply rely on such area aspects as availability
possibility of identification can be used to identify of UMTS. Personal preferences can also be used
the user in an online community. The user’s mobile for this purpose.
Another potential feature is to filter content
according to current contexts (Groh, 2003, p. 9).
Figure 1. Overview of the mobile social web A personal information service can select infor-
potentials mation about places of interest in the vicinity of
the user’s current location and report it to the user
(context specific presentation of information).
Moreover, context specific capturing of content
is conceivable (Groh, 2003, p. 8). Metadata (e.g.,
location data or location names) and information
and news services form the basis for this idea.
For example, an entry reviewing the quality of
food and drinks in a local café could be generated
automatically by entering the name of the café or

14
Mobile Social Web

its location. et al., 2002, pp. 24-25). Users pay either


Personal meetings can be organized more easily for the usage of each single service or for
as the awareness of a person’s location increases, using services during a certain period of
matching the users’ interests is extended, and time. Both approaches have proved unsuc-
the management of accessibility becomes more cessful in the case of non-mobile online
efficient. Stronger relationships between persons communities (Reichwald et al., 2002, pp.
will typically result when they are geographically 23-24). It may be expected that the advan-
close and have personal encounters (Larsen, Urry, tages of mobile communities and their im-
& Axhausen, 2006, pp. 12-13). These aspects are proved attractiveness are great enough to
not available in stationary online communities generate an increased willingness to pay
because they depend on time and location. Face- (Reichwald, Fremuth, & Ney, 2002, p. 12).
to-face communication can lead to an improved Moreover, users are generally accustomed
quality of relationships because interactions be- to paying for mobile services.
come closer and more intense. All of the above • Advertising and sponsoring: Even now,
will result in the strengthening of a community non-mobile online communities allow the
(Schneider, 2003, p. 99). generation of advertisements tailored for
special target groups by analyzing user
Economic Aspects profiles. In comparison to mass advertise-
ments, this approach involves less waste-
Mobile communities allow an opportunity for fulness (Schubert & Ginsburg, 2000, pp.
spontaneous, affective, and meaningful com- 51-52). For example, depending on the
munity activities, which may result in closer user’s current location, advertisements can
relationships between community members than be provided. Companies willing to spon-
in the case of traditional online communities. Data sor certain activities are another potential
input and output can be accomplished more easily source of revenue (Reichwald et al., 2002,
through context specific services (Diekmann et p. 20).
al., 2006). Therefore, community mobile operators • Data transmission fees: Fees for using
assume that users will be more willing to pay for telecommunication (TC) services have
participation in mobile communities than in sta- to be paid to the TC companies. The op-
tionary online communities (Reichwald, Fremuth, erator of a mobile online community could
& Ney, 2002, p. 12). The following list provides negotiate a share of the revenues. For this
details of primary and secondary revenue sources revenue sharing purpose, cooperation be-
for mobile operators (Reichwald, Fremuth, & tween the TC company and the community
Ney, 2002). operator is needed, and accounting models
Primary revenue sources arise from operating have to be developed.
a mobile online community (Reichwald, Fremuth,
& Ney, 2002, p. 11). Three possibilities can be The users themselves can become the potential
identified: generators of possibilities to earn money for the
community. Their content contributions can be
• To levy usage fees: Users pay for the us- used for market research as well as ideas about
age of mobile information and interactivity new services and products. Documented behavior,
services and for community content. The usage patterns, and preferences can be used to filter
fees depend on the usage of the services properties of target groups. The main potential for
or they are usage-independent (Reichwald community operators has to be made up by selling

15
Mobile Social Web

the gathered knowledge as consulting services or problem for mobile device users. Established
by using it for themselves (Reichwald, Fremuth, GSM services, such as Short Message Service
& Ney 2002, p. 12). Mobile communities can (SMS), provide an adequate transmission rate;
facilitate activities at a higher level of utility and however, there are only limited possibilities for
quality, as data are potentially more differentiated data presentation. Consequently, it makes more
and reliable compared to stationary activities. sense to opt for 3rd generation technologies like
GPRS and UMTS, especially as the increasing
Drawbacks of the Mobile Social Web dissemination of these technologies renders this
alternative increasingly feasible.
Mobile online communities have drawbacks, as An appropriate localization technology must
summarized in Figure 2. be integrated in order to realize enhanced mo-
bile services and provide new functionalities of
Technical Aspects mobile online communities, such as contextual
community content, context-oriented information
The utility of a service is a success factor as well collection, or efficient connectivity management.
as a challenge. It has to be decided which services The localization of end devices connected to cel-
should be accessible in a mobile community lular networks via their radio cell is currently still
to make it attractive (Yom, 2002, p. 177). Not expensive. However, free and precise localization
only information and interaction services have using Global Positioning System (GPS) is not yet
to be selected, content needs to be chosen, too. feasible, as the vast majority of users do not have
Restricted possibilities of data presentation and GPS-compatible end devices.
device handling as well as limited capacities of The possibility of unambiguous identification
mobile devices affect the utility. poses another problem. In case a mobile device
Other factors influencing usability are data gets lost, unauthorized persons may pretend to be
transmission costs, low data transmission rates, the original user, as mobile devices usually have
device handling problems, and inconvenient data only a very low degree of security (Reisinger,
input and output possibilities (Bliemel & Fassott, 2007). This is especially a problem when users
2002, p. 14). The fees of mobile network operators believe in the trustworthiness of the identification
are mostly based on data volume. Therefore, a using mobile end devices.
community user will have to pay fees depending
on the intensity of his or her usage. Social Aspects
Low data transmission rates are often a
The electronic capturing and processing of the user
context as well as the opening of communities for
Figure 2. Overview of the drawbacks of the mobile mobile usage can lead to negative consequences.
social web Publishing the actual location of a user means an
intrusion into his or her privacy and a limitation
to intimacy. Being spied upon undiscovered could
be the result of using contextual services. Parents
could use these services to locate their children.
Partners in a relationship could use it to track
each other. The risks involved may lead to a fear
of misuse or limited intimacy, resulting in the
merely conditional use of contextual services. In

16
Mobile Social Web

stationary online communities, we observe that access (Reichwald et al., 2002, p. 27). Fees
little emphasis is put on intimacy - users publish for the use of services or content would in-
a multitude of personal data (Gross & Acquisti, crease these costs.
2005, p. 4-8). Hence, it remains open whether the • Advertising and sponsoring: Advertising
added value of contextual services or the fear of in online communities was often frowned
misuse and limited intimacy will prevail. upon in the past (Leitner, 2003, pp. 41-42).
At present mobile radio services, including Even now, although an increased number
SMS or telephony, are preferably used for contact- of advertising banners are placed, the ac-
ing persons one already knows (Tasch & Brakel, ceptance of advertising does not seem to
2004, p. 4). Stationary online communities are be self-evident. Moreover, advertising
commonly used to create new relationships: i.e. messages on mobile devices are still fairly
contacting persons hitherto unknown to the user uncommon. It is unclear whether a satis-
(Fremuth & Tasch, 2002, p. 24). The projection factory advertising effect can be achieved
of mobile radio usage patterns onto stationary by mobile advertising in online communi-
online communities could lead to the change ties. It is assumed that the perception du-
from a preferably theme-oriented usage towards ration of advertisement tends to drop with
a person-oriented or communication-oriented mobile usage compared to the stationary
usage of mobile online communities (Reichwald Internet (Heinonen & Stransdvik, 2007, p.
et al., 2002, p. 13). Mobile community activities 610; Reichwald et al., 2002, p. 22). Even
would then be realized preferentially with already when advertisements are noticed, little
known persons. Such trends could jeopardize the space for advertising messages is available
establishment of theme-orientated mobile online due to the small displays of mobile devices
communities. (Michelsen & Schaale, 2002, p. 20).
• Mobile data transmission fees: As yet
Economic Aspects volume-billing models for the use of the
mobile Internet are widespread; this could
A commercial design of mobile online communi- be a restriction to the time-consuming use
ties has to identify primary and secondary revenue of a mobile online community (Reichwald
sources for their operators. Commercial intentions et al., 2002, p. 28). This can lead to lower
should be declared and openly communicated to data volumes, resulting in fewer payments
the community (Leitner, 2003, pp. 43-44). This is to mobile radio operators. As a conse-
the only way to grant trust and authenticity from quence, these operators would be less will-
the very beginning. Using primary revenue sources ing to forward payments to a community
involves the following challenges: operator. It remains to be seen how this
revenue potential develops through bill-
• Collection of usage fees: So far, the sta- ing models for mobile Internet usage and
tionary Internet largely provides free con- cooperation.
tent and services (Reichwald, Fremuth, &
Ney, 2002, p. 11). Internet users are ac- There are also new challenges for the sec-
customed to free services. There is a pos- ondary revenue sources: Using a mobile online
sibility that this factor will have a negative community as an instrument for market research
impact on the willingness to pay for mobile can be profitable for an operator, but it seems
services. Users already pay for mobile data reasonable that the users need to know that this
services such as SMS or mobile Internet is drafted on the basis of content and the analysis

17
Mobile Social Web

of their usage habits (Leitner, 2003, p. 43-44). significance of the mobile Internet goes further
This could especially affect success factors of than providing communities with an additional
an online community like equality, credibility access channel. Besides mobile access and instant
and trust. Sharing part of the revenues with users connectivity, unambiguous identification and
could be a solution. contextual services can also be realized.
The mobile web platform gains efficiency
and facilitates not only flexible, spontaneous,
FUTURE TRENDS and emotional interactions, but also credible and
intensive ones. Reckoning with these possibili-
A growing number of Internet services make ties, a community mobile operator can take an
their applications available to mobile users. This economic perspective and identify revenue sources
can be attributed to the increasing proliferation that make the commercialization of mobile com-
of mobile broadband Internet access (especially munities feasible.
UMTS in Europe). Three variants of mobile However, the specific technical and social
implementation are available. Users can browse characteristics of mobile communities can cast
profile pages and photo albums via WAP and doubt upon their prospective potentials. Com-
add new images and texts by means of SMS and mercialization appears less promising as the
Multimedia Message Service (MMS). The ShoZu willingness of users to pay fees is rather low.
service,1 in contrast, performs as an integration Mobile advertising poses additional problems
platform and allows mobile users to upload and despite its context-relatedness. Moreover, the
download content at multiple online communities realization of secondary revenue sources can lead
with a single message. One ConnectTM provided to problems of trust.
by Yahoo is another illustration which integrates The assumption that the social significance of
social communities (e.g., MySpace, Dopplr, Fa- the mobile Internet will go beyond providing an
cebook or Last.fm), instant messaging services additional access channel to online communities
(e.g., Yahoo! Messenger, MSN Messenger), and has so far only partially been confirmed. Many
the ordinary communication channels of mobile challenges to mobile communities remain to be
phones. Finally, as the example of the COSMOS resolved in the future for the economic perspective
project2 shows, a comprehensive use of mobile to gain relevance and the commercial interpreta-
technical and social opportunities appears to be no tion to become a success.
so distant future—the COSMOS project integrates
contextual services into the existing mobile social
web. Users are informed about the geographical References
distance between them and may send messages
to contacts within their own vicinity. The above- Bliemel, F., & Fassott, G. (2002). Kundenfokus
mentioned trends are expected to advance in the im mobile commerce: Anforderungen der kunden
future; yet, the possible dominance of one of them und anforderungen an die kunden. In G. Silberer, J.
cannot be predicted. Wohlfahrt, & T. Wilhelm (Eds.), Mobile commerce
- grundlagen, geschäftsmodelle, erfolgsfaktoren
(pp. 3-23). Wiesbaden, Germany: Gabler.
Conclusion

The analysis of the potentials of mobile online


communities leads to the conclusion that the social

18
Mobile Social Web

Brunold, J., Merz, H., & Wagner, J. (2000). www. Hamman, R. B. (2000). Computernetze als
cybercommunities.de: Virtual communities: verbindendes element von gemeinschaftsnetzen.
Strategie, umsetzung, erfolgsfaktoren. Landsberg, In U. Thiedeke (Ed.), Virtuelle gruppen. Charak-
Germany: Verlag Moderne Industrie. teristika und problemdimensionen (pp. 221-243).
Wiesbaden, Germany: VS Verlag für Sozialwis-
Diekmann, T., Kaspar, C., Seidenfaden, L., &
senschaften.
Hagenhoff, S. (2006). Kontextbewusste infor-
mationsdienste auf grundlage von information Heinonen, K., & Strandvik, T. (2007). Consumer
beacons. Retrieved June 23, 2008, from http:// responsiveness to mobile marketing. Interna-
www.gi-mms.de/mms2006/kurzbeitraege/diek- tional Journal of Mobile Communications, 5(6),
mann.pdf 603–617. doi:10.1504/IJMC.2007.014177
Eigner, C., & Nausner, P. (2003). Willkommen, Heng, S. (2006). Entgegen vielen erwartungen!
‘social learning’! In C. Eigner, H. Leitner, P. Naus- Breitbandige mobilfunktechnologie UMTS ist
ner, & U. Schneider (Eds.), Online-communities, realität. Retrieved June 5, 2008, from http://
weblogs und die soziale rückeroberung des netzes www.dbresearch.com/PROD/DBR_INTER-
(pp. 52-94). Graz, Austria: Nausner & Nausner. NET_DEPROD/PROD0000000000198071.pdf
Fremuth, N., & Tasch, A. (2002). Virtuelle und Herzig, M. (2001). Basistechnologien und stan-
mobile communities - begriffsklärungen und imp- dards des mobile business. Wirtschaftsinformatik,
likationen für geschäftsmodelle (Arbeitsberichte 43(4), 397–404.
des Lehrstuhls für Allgemeine und Industrielle
Hummel, J. (2005). Online-gemeinschaften
Betriebswirtschaftslehre, No. 35). München:
als geschäftsmodell - eine analyse aus sozio-
Lehrstuhl für Allgemeine und Industrielle Betrie-
ökonomischer perspektive. Wiesbaden, Germany:
bswirtschaftslehre.
Deutscher Universitätsverlag.
Gebert, D., & Von Rosenstiel, L. (1992). Organisa-
Kim, A. J. (2000). Community building on the
tionspsychologie: Person und organisation. Köln,
Web: Secret strategies for successful online com-
Germany: Kohlhammer.
munities. Berkeley, CA: Peachpit Press.
Gerum, E., Sjurts, I., & Stieglitz, N. (2003).
Koch, M., Groh, G., & Hillebrand, C. (2002).
Der mobilfunkmarkt im umbruch - eine innova-
Mobile communities - extending online communi-
tionsökonomische und unternehmensstrategis-
ties into the real world. Retrieved May 27, 2008,
che analyse. Wiesbaden, Germany: Deutscher
from http://www11.informatik.tumuenchen.de/
Universitäts-Verlag.
publications/pdf/Koch2002c.pdf
Groh, G. (2003). Ortsbezug in kontext-sensitiven
Koesch, S., Magdanz, F., & Stadler, R. (2007).
diensten für mobile communities. Retrieved
Soziale netzwerke - mobile kontaktbörsen. Re-
June 20, 2008, from http://www11.informatik.
trieved June 15, 2008, from http://www.spiegel.
tu-muenchen.de/lehrstuhl/personen/groh/pub/
de/netzwelt/mobil/0,1518,470250,00.html
gis.pdf
Kurose, J. F., & Ross, K. W. (2005). Computer
Gross, R., & Acquisti, A. (2005). Information
networking - a top-down approach featuring the
revelation and privacy in online social networks
Internet. Boston: Pearson Education.
(the Facebook case). Retrieved June 22, 2008,
from http://www.heinz.cmu.edu/~acquisti/papers/
privacy-facebook-gross-acquisti.pdf

19
Mobile Social Web

Larsen, J., Urry, J., & Axhausen, K. W. (2006). Schneider, U. (2003). Online-community - neues
Mobilities, networks, geographies. Aldershot, medium und/oder neue sozialform? In C. Eigner,
UK: Ashgate. H. Leitner, P. Nausner, & U. Schneider (Eds.),
Online-communities, weblogs und die soziale
Leitner, H. (2003). Online-community, ‘hands
rückeroberung des netzes (pp. 95-114). Graz,
on!’. In C. Eigner, H. Leitner, P. Nausner, & U.
Austria: Nausner & Nausner.
Schneider (Eds.), Online-communities, weblogs
und die soziale rückeroberung des netzes (pp. Schubert, P., & Ginsburg, M. (2000). Virtual com-
11-51). Graz, Austria: Nausner & Nausner. munities of transaction: The role of personalization
in electronic commerce. EM Electronic Markets,
Markus, U. (2002). Integration der virtuellen com-
10(1), 45–56. doi:10.1080/10196780050033971
munity in das CRM: Konzeption, rahmenmodell,
realisierung. Electronic Commerce, 15. Stader, R. (2001). Client- und endgerätetechnolo-
gien für mobile community-support-systeme. In
Maslow, A. H. (1943). A theory of human mo-
U. Baumgarten, H. Krcmar, R. Reichwald, & J.
tivation. Psychological Review, 50, 370–396.
Schlichter (Eds.), Community online services and
doi:10.1037/h0054346
mobile solutions - projektstartbericht des ver-
Michelsen, D., & Schaale, A. (2002). Handy-busi- bundvorhabens COSMOS (pp. 31-44). München,
ness: M-commerce als massenmarkt. München, Germany: Institut für Informatik, Technische
Germany: Financial Times Prentice Hall. Universität München.
Petrzek, D. (2007). Süchtig nach Myspace & Co. Tasch, A., & Brakel, O. (2004). Location based
Internet World Business, 26, 1. community services – new services for a new
type of Web communities. Retrieved May 25,
Reichwald, R., Erben, R., Fremuth, N., & Tasch,
2008, from http://www.cosmoscommunity.org/
A. (2002). Mobile communities: Phänomen und
downloadFiles/Lisbon-format-final.pdf von
erlösungspotenziale. In R. Reichwald (Ed.), Ar-
Tetzchner, J. S. (2008). Mobile browsing report,
beitsberichte des Lehrstuhls für Allgemeine und
state of the mobile Web: First quarter, 2008.
Industrielle Betriebswirtschaftslehre (no. 36).
Retrieved June 23, 2008, from http://www.opera.
München, Germany: Lehrstuhl für Allgemeine
com/mobile_report/
und Industrielle Betriebswirtschaftslehre.
Wiedmann, K.-P., Buckler, F., & Buxel, H. (2000).
Reichwald, R., Fremuth, N., & Ney, M. (2002).
Chancenpotentiale und gestaltungsperspektiven
Mobile communities - erweiterung von virtuellen
des m-commerce. Der Markt, 39(153), 84–96.
communities mit mobilen diensten. In R. Reich-
doi:10.1007/BF03036349
wald (Ed.), Mobile kommunikation (pp. 521-537).
Wiesbaden, Germany: Gabler. Winkler, K., & Mandl, H. (2004). Virtuelle com-
munities - kennzeichen, gestaltungsprinzipien und
Reisinger, D. (2007). ‘Bluejacking,’ ‘bluesnarf-
wissensmanagement-prozesse. Retrieved June
ing’ and other mobile woes. Retrieved June 18,
12, 2008, from http://epub.ub.unimuenchen.de/
2008, from http://news.cnet.com/8301-10784_3-
archive/00000323/01/ FB_166.pdf
9764450-7.html

20
Mobile Social Web

Yom, M. (2002). Utility und usabilty im mobile to the Internet. Examples include mobile phones,
commerce. In G. Silberer, J. Wohlfahrt, & T. smart phones, and personal digital assistants.
Wilhelm (Eds), Mobile commerce - grundlagen,
geschäftsmodelle, erfolgsfaktoren (pp. 173-184). Mobility: Unrestricted transfer of text, voice
Wiesbaden, Germany: Gabler. or data independent of user’s physical location.

Online Community: A social group that in-


teracts through a web platform over an extended
Key Terms and Definitions
period of time.
Cosmos: The Community-Driven Systems
Management in Open Source (COSMOS) project Social Web: Refers to Web 2.0-based tech-
is a collaboration of software developers aimed nologies and applications that are used to support
to create standards-based tools for system man- communication and facilitate social contact, such
agement. as, social networking sites, massively multiplayer
online role-playing games, photo and video shar-
Mobile Internet: Use of TCP/IP based services ing, online stores and auction houses, virtual
and protocols with mobile devices via wireless worlds, and collaborative wikis.
communications technologies.

Mobile Social Web: Refers to mobile social Endnotes


networks and other Web 2.0-based applications in 1
http://www.shozu.com
which people access and form online communities 2
See http://www.eclipse.org/cosmos
by using mobile devices.

Mobile Wireless Devices: Handheld elec-


tronic devices with wireless capability to connect

This work was previously published in Handbook of Research on Social Interaction Technologies and Collaboration Software;
Concepts and Trends, edited by T. Dumova; R. Fiordo, pp. 11-21, copyright 2010 by Information Science Reference (an imprint
of IGI Global).

21
22

Chapter 1.3
Social Semantic Web and
Semantic Web Services
Stelios Sfakianakis
ICS-FORTH, Greece

Abstract interconnected islands of data where documents are


linked, searched for, and shared, forming a massive,
In this chapter the authors aim to portray the social albeit not always well organized, digital library.
aspects of the World Wide Web and the current and Sharing of digital content has always been the major
emerging trends in “Social Web”. The Social Web requirement for the Web since its inception and will
(or Web 2.0) is the term that is used frequently to continue to be one of its most important features in
characterize Web sites that feature user provided the years to come. Nevertheless, what we experience
content as their primary data source and leverage nowadays is the endeavor for extending this shar-
the creation of online communities based on shared ing to cover also additional artifacts beyond plain
interests or other socially driven criteria. The need documents, like data, information, and knowledge.
for adding more meaning and semantics to these The power of the hyperlinks, connecting different,
social Web sites has been identified and to this possibly disparate entities, can also be exploited in
end the Semantic Web initiative is described and order to connect information sources and people:
its methodologies, standards, and architecture are not just “dumb” machine readable data but dynamic
examined in the context of the “Semantic Social content like user profiles and ultimately people
Web”. Finally the embellishment of Web Services themselves for building virtual communities. The
with semantic annotations and semantic discovery vision is that the current web of computers and
functionality is described and the relevant technolo- documents will be broadened to the web of people.
gies are explored A “People Web” is the one where users are the nodes
of the graph, the edges being their relationships and
Introduction interactions in space and time, thus constructing
new virtual societies (see Figure 1).
The World Wide Web (WWW or, simply, the “Web”) This new environment is leveraged by the in-
has been used extensively as a huge network of troduction of an array of technologies collectively
identified as Semantic Web (Berners-Lee, Hendler,
& Lassila, 2001). The Semantic Web builds upon
DOI: 10.4018/978-1-60566-340-1.ch011

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Social Semantic Web and Semantic Web Services

Figure 1. A social graph

the existing Web and provides the necessary systems, one-way, inexpensive links (Universal
substrate for giving “meaning” and “Semantics” Resource Identifiers – URIs) but also the employ-
to Web resources and Web interactions. The ment of the Internet as its networking substrate
benefits will be many in a number of application led to its wide adoption and success.
domains and while the challenges, technological In spite of its success and popularity the early
and other, are numerous, the momentum is strong version of the Web lacked in many respects,
and the Semantic Web slowly but steadily enters ranging from user accessibility and user interface
in a number of diverse domains like health and design to the ability to repurpose and remix exist-
life sciences. ing Web-based data in not pre-established ways.
Furthermore the Semantic Web promises a Although the hyper linking facility allowed the
great potential for supporting the construction and interconnection of different documents on the
smooth operation of Web communities of people. Web, the “traditional” Web suffers from frag-
In this chapter we study its fusion with social mentation in the sense that the huge graph lacks
software and software for machine to machine any organization and discipline. This “anarchy”
communication over the Web for supporting this in the Web has also been its driving force for the
vision. success it has enjoyed so far but created the need
for having special machinery, e.g. search engines
like Google, to deal with the efficient indexing and
Background discovery of the available information. Despite
the fact that the search engine technology has
Since its launching in 1990, the Web has grown made important steps in indexing and searching
exponentially both in terms of size and in terms massive amounts of data on the Web, there’s still
of use and utility to people and organizations. The the issue that keyword based searching is limited
inherent simplicity of hypertext and its feature on its potential and usually finding “what the user
limited, in comparison to previous hyper linking wants” proves to be a tedious task. Another major

23
Social Semantic Web and Semantic Web Services

limitation of this environment is that the people and recently the convergence of the Social and
are not part of the equation. Users are expected to the Semantic Web and the experimentation of the
be the actors triggering the Web interactions but two working in complementary ways have gained
they are not allowed to participate and be involved a lot of attention and research interest.
enough in these interactions:

• Content delivered is not personalized. Social Web or Web 2.0


What the user gets back is usually not in-
line with her interests or other preferences The situation described in the previous section led
and there’s no feedback link going from to the emergence of a new breed of Web applica-
the user back to the system she interacts tions and sites that were collectively identified as
with so as to guide future interactions. “Web 2.0” by Tim O’Reilly (2005) and whose
• Contextual information is not taken into major design principle is to “harness network
consideration. The people as complex sys- effects to get better the more people use them”.
tems do not act in an easily predetermined The value of “Web 2.0” sites and applications
way and the context of their actions is usu- therefore comes to a large extent by the number
ally ignored or not taken advantage of. This of users participating and actively communicat-
context information ranges from the user’s ing and sharing through them so the term “Social
profile, which is also dynamic in nature, Web” is actually a synonym. The social nature
to the specific objective she/he is trying to of this Web is evident when the collaboration of
achieve at a specific point in time. people and their active contribution is considered.
• Content is passive and static, stored and The very essence of such sites is the building and
maintained in back end databases, which maintenance of Web based virtual communities
the users do not have the ability to enrich of people that produce and maintain collective
or customize to their own needs knowledge. Examples of such community oriented
• Communication and collaboration of the and social networking sites include:
users to build Web communities are not
supported enough. Discussion forums were • Blogs, i.e. Web sites managed by indi-
the sole way to build such communities but viduals that provide news or opinions on
with no means to support intelligent inte- certain subjects (typically personal online
gration of the different forums or to en- diaries), where typically other people are
hance the user collaboration experience. able to leave comments. In addition to
comments, the hyperlinking facility of the
These and other requirements are the ones that Web has been extensively used to provide
the Social Web tries to tackle. Social Web does not “trackbacks” (i.e. reverse hyperlinks that
represent a shift or radical change in technology identify who is talking about me) and rec-
per se but rather a shift on the perception of the ommended blogs (“blogrolls”). Therefore
human – machine interaction by placing the users blogging has been emerged as a method for
in the centre of the system and in control of these anyone to publish content on the Web and
interactions. But from the other end of the spec- building online communities of people that
trum there is also a clear need for making the Web communicate, share, and integrate.
itself more intelligent to support these machine • “Social bookmarking” sites (e.g. http://
facilitated social interactions. The Semantic Web del.icio.us/) where users can store and
could provide for such an enabling technology share their bookmarks with the additional

24
Social Semantic Web and Semantic Web Services

possibility to provide metadata through openwetware.org/) for biology and biolog-


the means of tags, i.e. terms that denote ical engineering
concepts, meaning, intent, etc. These sites • As open educational centers to support
provide for user maintained and collabora- learning activities, e.g. Wikiversity (http://
tive indexing of the Web content in a way en.Wikiversity.org/)
that it may be more efficient to search there • As social event calendars, like Upcoming
for something than in general purpose Web (http://upcoming.yahoo.com/)
search engines.
• “Wikis” (e.g. http://en.Wikipedia.org), The single distinctive feature of Wikis and
which are collaboratively built Web sites a central trait of the social Web sites is the user
where the users, through custom made and generated content and its “open” editing: any-
user friendly interfaces, are able to create, one can edit an existing Wiki article or create a
share, enhance, and manage the content. new one for a particular topic if it doesn’t exist
• Content sharing sites, e.g. YouTube (http:// already. The users therefore are responsible for
www.youtube.com/) for videos or Flickr adding content and maintaining the information
(http://www.flickr.com/) for photographs, that is available from these sites. Of course such
where the users upload their multime- an approach can raise a lot of concerns about the
dia content and share it online with other validity of the content, the lack of authority, etc.1
users. and there have been cases in the past where such
• Social networking sites, such as Facebook skepticism was proven true, such as the Seigen-
(http://www.facebook.com/) and MySpace, thaler incident2. Nevertheless this open model
for online communities of people who has worked quite well in practice and in general
share interests and activities or who are so that at the time of this writing Wikipedia is
interested in exploring the interests and ac- considered by many a serious competitor to the
tivities of others. Encyclopedia Britannica. The reason for this can be
• Classified advertisement sites, e.g. Craigslist explained as another instantiation of the “wisdom
(http://www.craigslist.org), which offer of crowds” phenomenon (Surowiecki, 2004): the
advertisements for jobs, resumes, services, participation of many people, possibly with dif-
etc. grouped in categories. ferent background, habits, way of thinking, and
so on, in a decision making process usually yields
If we take only “Wikis” as an example we better results than when the individual opinions
can see that these Web sites have been used in a are considered separately from one another.
multitude of ways: The contribution of user content and the shar-
ing of the uploaded information are the main
• As online encyclopedias, e.g. Wikipedia forces for the formation of online communities
• As free dictionaries, e.g. Wiktionary (http:// of people. In Figure 2 an example of this com-
en.wiktionary.org) munity creation process is shown for the Del.icio.
• As free libraries of educational books, e.g. us online bookmarking site. Online bookmarking
Wikibooks (http://en.Wikibooks.org) sites like this provide the means for storing and
• As software development repositories and organizing bookmarks of Web sites on the Web
issue tracking systems, e.g. Trac (http:// instead of the users’ desktop browsers. By storing
trac.edgewall.org/) their bookmarks in a central area the users are
• As open forums to promote research additionally enabled to create their online social
interests, like OpenNetWare (http:// networks by registering other users as members

25
Social Semantic Web and Semantic Web Services

Figure 2. Del.icio.us networks of users

of their network so that they can be notified about of the business and control logic resides on the
the bookmarking activity of these users. These client (i.e. the Web browser), leveraged by tech-
networks therefore connect users with their friends, nologies like AJaX3 and Comet4 which provide
family, coworkers, or even totally strangers when more responsive user interfaces.
they unexpectedly meet each other on the Internet The Social Web offers a meeting point for
and discover they have similar interests. Facilitated people to collaborate and share information in an
by these network links the users can subsequently open environment. The openness is a distinctive
observe each other’s online behavior and even characteristic of Web 2.0 and it’s supported by
proactively send interesting Web sites addresses Open Data APIs like content syndication via RSS/
to their peers, easier and quicker than using email Atom5 and lightweight Web services interfaces
or instant messaging. like Open Search6 . These technologies enable
What the previous examples show is that in the the view of Web sites as Web applications and
Social Web users are in the limelight: they are the their synthesis (“mashup”) in more complex ap-
primary actors in the data sharing process through plications. An example of such combination of
their contributions and online behavior. They are existing Web sites and their data to create new/
usually indulged by the low cost entry and partici- aggregated content is Housing-Maps (http://www.
pation in these Web sites, and, to a lesser extent, housingmaps.com/) where houses to rent or buy
by the visual appeal the Web 2.0 sites offer to the are located through Craiglists and projected over
viewer. The modern Web sites are actually Rich geographic maps drawn from Google Maps (http://
Internet Applications (RIA), where the majority maps.google.com) so that a user can easily locate

26
Social Semantic Web and Semantic Web Services

the information he wants in an interactive and are suboptimal the user participation and their
visual way. A more general and reusable way to socialization needs have driven the evolution of
combine and “mix” content from different Web Web of documents to the Web of People (Ra-
sites is offered by Yahoo! Pipes7 which can be makrishnan & Tomkins, 2007).
thought of a simple but effective way to build
“workflows” and “dataflows” on the Web.
The above discussion shows that collabora- Semantic Web
tion between people but also between Web sites/
applications supports the notion of “collective To the other end of the spectrum, with roots in
intelligence” to the Social Web. An instance of Artificial Intelligence research, the Semantic Web
this intelligence built collectively is the creation of emanated as an extension to the current version
“folksonomies” for categorization of resources. A of the Web that aims to enhance it by the promo-
quite popular way of classifying content in Web 2.0 tion of higher level sharing and integration of
Web sites is through “tagging”. A tag is a keyword data and information. According to Berners-Lee
which acts like a subject or category. The user is et al. (2001):
allowed to attach whatever keywords she wants The Semantic Web is not a separate Web but an
to identifiable content such as links in the case of extension of the current one, in which information
social bookmarking, or videos and photographs in is given well-defined meaning, better enabling
the case of digital content sharing. The important computers and people to work in cooperation.
thing is that tags can be shared, used in searches, The Semantic Web aims to support the rep-
or recommended based on the choices of other resentation and exchange of information in a
users for the same content. meaningful way so as to make possible the auto-
The new term “folksonomy”, as a fusion of the mated processing of descriptions on the Web. The
words “folks” and “taxonomy”, has been suggested objective is to enrich the unstructured information
to describe this method of classifying content in the current Web with machine processable
through tags that are collaboratively generated descriptions of the Semantics in order to make
and shared. Of course these “poor man’s” clas- its navigation and exploration by software agents
sification schemes are informal in nature, could as easy as it’s for the human users today, or even
contain duplication in meaning, or be simply easier. In this context Semantic Web promotes
erroneous but again they are contributed by the a shift from the current “syntactic” world to the
users and the more people contributing the more future “Semantic” world of services, applications,
robust and stable these “folksonomies” become. and people and aims to make the machine to
A self adapting and auto regulating method is machine communication feasible so that not only
usually followed through the use of tag clouds data but also information and finally knowledge
(Figure 3). In simple terms a tag cloud is a visual are shared.
representation of a user’s tags where each tag is
weighted based on the user preferences and how The Semantic Web
many times he has used the tag. Through such an Technology Infrastructure
approach “good” tags are likely to prevail assum-
ing that the user participation is high. In technological terms the Semantic Web architec-
Collaboration, sharing, “mashing”, annotating ture consists of an array of technologies that can
and “tagging” content are roughly the distinctive roughly be visualized in a layered design layout
features of Web 2.0 and although in most of the as depicted in Figure 4. The basic infrastructure in
cases the approach is not formal or the solutions the bottom layers in this stack of technologies is

27
Social Semantic Web and Semantic Web Services

Figure 3. A tag “cloud”

the exactly the same to the syntactic Web: Uniform practice to describe, by the means of RDF and
Resource Identifiers (URIs) used for identification the various URI schemes, real world entities like
of Web resources, universal encoding schemes people, or more abstract things like relationships
for characters, i.e. Unicode, and XML and its and concepts. The use of URIs and especially the
related technologies (e.g. XML Namespaces) as HTTP based ones for identifying persons or other
a ubiquitous data serialization format. Some of physical entities may seem strange at first but
the upper layers like Proof and Trust are missing this is in compliance with the architecture of the
or are work in progress. Here we will concentrate World Wide Web (Berners-Lee et al., n.d.) which
on the middle layers where the core infrastructure strongly suggests the use of URIs for identifying
technologies of the Semantic Web reside: RDF, anything that can be of importance irrespective
RDF Schema/OWL, and SPARQL. of how abstract or tangible it may be.
The Resource Description Framework (RDF) The properties serve both to represent attri-
is a syntax neutral data model that enables the butes of resources and to represent relationships
description of Web resources in a simple way between resources. They are also identified though
(Lassila, Swick, et al., 1999). At the core of RDF URIs to make them unique. The combination of
there is a model for representing and describing resources and the properties that connect them
resources through named properties (also known builds the simple RDF data model. In this data
as predicates) and their values. The resources model the primary informational building block
can be anything that can be identified with a is the “triple” which denotes the subject – prop-
URI. Although in the initial specification of RDF erty - object expressions (Figure 5). The subject
resources were limited to Web documents and denotes the resource, and the predicate denotes
Web sites, it is possible and quite frequent in traits or aspects of the resource and expresses a

28
Social Semantic Web and Semantic Web Services

Figure 4. The Semantic Web stack of technologies

relationship between the subject and the object. any specific serialization syntax. The normative
Since an object of a triple can be the subject of representation syntax for RDF graphs is XML but
another one, a set of RDF triples forms a directed more lightweight formats, such as Turtle (Beckett
graph where the RDF resources, both subjects & Berners-Lee, 2008), exist.
and objects, are the nodes of the graph and the The simplicity and flexibility of RDF is evi-
predicates are the labeled arcs. As an example, in dent but in certain cases its generality must be
Figure 6 there’s a simple RDF graph. The graph formally confined so that software entities are able
shown in the figure describes an entity identified to correctly exchange the encoded information.
through the URI “http://ssfak.org/stelios/”, ap- For example, stating that an animal is the creator
parently denoting a person, which has a “name” of a Web page does not make sense in the real
property with the value “Stelios Sfakianakis”, a world but RDF does not forbid anyone for making
property denoting the homepage of an organiza- such a claim. Ontologies (Uschold & Gruninger,
tion a person works for relating it to the resource 1996) provide such a tool to specify what can be
“http://www.ics.forth.gr/cmi-hta/”, and a “maker” expressed in the context of an application domain
property that connects it (backwards, as an object) or in a real world scenario, what is the underlying
to the resource identified as “http://ssfak.org”. meaning, and how the information presented can
RDF as an abstract model is independent of be further processed to generate more information.

Figure 5. RDF Data Model

29
Social Semantic Web and Semantic Web Services

Figure 6. Abstract representation of RDF triples

Moreover ontologies and their less powerful rela- resources and their URI references are used, the
tives like taxonomies and thesaurus provide the open world assumption is followed, since partial
means for achieving a common interpretation of a information on the Web is a quite frequent phe-
domain and a shared understanding of the concepts nomenon, and the ontologies themselves can be
and relationships involved. In the Semantic Web freely intermixed and meshed since hyperlinks
there are two main technologies for providing are employed everywhere.
such rigor: RDF Schema and OWL (Brickley & Since RDF is the common interchange and
Guha, 2004; Dean, Schreiber, et al., 2004). RDF representation model of information, the Semantic
Schema provides the means for defining classes, Web transforms the hyperlinked syntactic World
class hierarchies, properties, property hierarchies, Wide Web to a huge database or a Global Giant
and property restrictions. Its expressive power is Graph, as Tim Berners-Lee put it. The standard
basically limited to the representation of concepts, query language for this huge database is SPARQL
their relations, and taxonomies of concepts. On the (Prudhommeaux & Seaborne, 2008), which is
other hand the Web Ontology Language (OWL) similar to SQL. In addition to the query language
was introduced to address the need for more the SPARQL standard defines an application
expressiveness and extends the RDF Schema by protocol for the submission of queries to RDF
providing three variants: OWL-Lite, OWL-DL, sources and the retrieval of results. With the
and OWL-Full. Without delving into details, query language and the access protocol defined,
the different species of OWL provide different the SPARQL specifies a Web friendly interface to
degrees of expressiveness and are able to define RDF information, whether this is actually stored as
existential restrictions, cardinality constraints in RDF triples or not. It is therefore feasible to make
properties, property types like inverse, transitive, SPARQL queries to relational or other databases
and symmetric, and a lot more. The added features through an appropriate wrapper or transformation
of OWL allow the ontologies built in conformance process that translates, either online or in some
to it to be formally treated and the data represented preprocessing step, the internal data to an RDF
are amenable to “reasoning” and inference, i.e. compliant format. As a result these Semantic
they can be processed according to formal logic Web technologies enable the connection of data
rules to deduce new information. All these hap- between different and heterogeneous data sources,
pen on the basis of the Web infrastructure: RDF effectively allowing data in one data source to

30
Social Semantic Web and Semantic Web Services

be linked to data in another data source. (Bizer, If Web 2.0 could be summarized as interaction,
Heath, Idehen, & Berners-Lee, 2008) Web 3.0 must be about recommendation and
personalization.

Social Semantic Web An example of such added value is the case


of Semantic Wikis (e.g. Schaert, 2006; Völkel,
In recent years the cross pollination of Semantic M., Krötzsch, M., Vrandecic, D., Haller, H., &
Web technologies and Social Networking has Studer, R., 2006). The Semantic Wikis support the
emerged as an interesting roadmap. The Seman- annotation with Semantics descriptions the links
tic Web technology can significantly enrich and and the content they provide and take advantage
expedite the Social Web in order to establish the of these annotations for providing more intelligent
Semantic Social Web (Greaves, 2007;Gruber, search and navigation. The annotation is usu-
2007).  In the Semantics-enabled social Web con- ally done by some extended version of the Wiki
tent can be easily connected, integrated, navigated, editing syntax so that every link to another page
and queried so that the benefits of today’s Social or any important attribute of the current page is
Web can be greatly enhanced and augmented be- annotated with a property identifier. For example
yond the limited user experience offered by social in a Semantic Wiki’s page about the Europe the
networking sites alone or the restricted keyword amount of its population, which is a number, can
based search and matching. be wrapped with the appropriate metadata that
What does the Semantic Web offer to the So- denote that this number represents the population.
cial Web? First and foremost, the Semantic Web Such metadata annotation makes structured search
technologies can be used to provide rigor and easy, e.g. for queries like what is the population
structure to the content of the user contributions of Europe, or which continents have population
in a form that enables more powerful computa- above a certain amount. Additionally it facilitates
tion.  Currently social Web applications are more the users in providing more active content in the
focused on the distribution and the management pages by incorporating “online queries” in the
of content and the social interactions around it Wiki pages, in the sense that the page’s content can
rather than the provision of Semantically rich de- be dynamically generated by the results of these
scriptions of the data. Although there are popular, queries on the metadata annotations. Although
“low end” technologies like “microformats” and the details may vary from one implementation to
tagging/“folksonomies” to cater for the annota- another, there’s usually an underlying model based
tion and the description of data, these seem to be on RDF and OWL to support these Wikis and the
ad hoc and unstructured efforts in comparison to content can be exported in a Semantic Web com-
the formal Web ontologies and metadata descrip- pliant format. DBPedia is an interesting example
tions. On the other hand, as already described, the of a truly Semantic Web Wiki which offers the
Semantic Web promotes the global distribution content of Wikipedia in a machine-readable and
and integration of resources in a single, giant, searchable format (Auer et. al., 2007).
interoperable graph. So, additionally, the standards In another application area, Semantic Web
and infrastructure of the Semantic Web can enable technologies can facilitate the browsing experi-
data sharing and computation across independent, ence of people and the searching capabilities of
heterogeneous social Web applications. the Web search engines. Unlike traditional search
Furthermore, the Semantic Web can enhance engines, which “crawl” the Web gathering Web
the Social Web with additional intelligence as pages information, Semantic Web search engines
Jemima Kiss (2008) wrote: index RDF data stored on the Web and provide

31
Social Semantic Web and Semantic Web Services

an interface to search through the crawled data. using Semantic Web technologies. It defines meth-
Because of the inherent Semantics of RDF and the ods to describe the information that communities
other Semantic Web technologies, the search and have about their structure and contents, and to find
information retrieval capabilities of these search related information and new connections between
engines are potentially much more powerful than content items and other community objects. SIOC
those of current search engines. Examples of such again is based around the use of machine-readable
early Semantic Search Engines include the Seman- information provided by these sites.
tic Web Search Engine (SWSE, http://www.swse. The main entities of SIOP are shown in Figure
org/), Swoogle (http://swoogle.umbc.edu/), and 7 and it’s easy to see the role and the function of
Zitgist Search (http://www.zitgist.com/). These the main concepts. The entity Site refers to the
and other Semantic Web search engines explore location of an online community or set of com-
and index the documents of the Semantic Web munities, which hosts one or many blogs. A Forum
and its ontologies by the means of user friendly can be thought of a discussion area on which
interfaces that hide the details and complexities posts are made. In a Forum a number of posts are
of the technology. contained where a Post represents an article or a
Blogs, which are one of the most prominent message send by a user to the forum. Posts can be
examples of the Social Web, can also be enhanced connected as people reply to previous posts and
with Semantics. Augmenting a blog with content these connections can cross site boundaries since
and structural metadata is usually called Seman- the identifiers of posts (as well as any Semantic
tic Blogging (Cayzer, 2004; Bojars, Breslin, & Web resource) are universal and unique.
Moller, 2006). Putting Semantics in a blog’s From the figure above it can be said that SIOP
contents means that the topic of the content is defines a common schema for the different blog
described in a machine processable way. On the sites and discussion forums. This of course needs
other hand describing the structure of the blog not be their internal schema but a common, shared,
Semantically entails the description of the entities and standard representation of their information
that compose it: the posts, the comments, the users, model. Adopting SIOP therefore is a major step
etc. To this end there are a number of efforts to in achieving the integration of social content in
make the Semantic Web more social by building Web 2.0.
new ontologies to support people in their social
interactions and provide Semantics to the Social
Web. Two of such ontologies, SIOC and FOAF, FOAF
are of particular importance in the context of Se-
mantic Blogging and are described below. The Friend-Of-A-Friend (FOAF) project focuses
on expressing mostly personal information and
relationships in a machine-readable form. A central
SIOC entity in the FOAF vocabulary and the one most
frequently used is the Person (Figure 8). Accord-
Existing online community sites usually provide ing to FOAF a Person may have names, e-mails,
rich information for specific interest groups but interests, publications, etc. It can also be connected
they are isolated from one another, which makes to other resources like the Web site of the organiza-
difficult the linking and merging of comple- tion he/she works for (foaf:workplaceHomepage
mentary information among different sites. The property), a personal blog site (foaf:weblog), the
Semantically-Interlinked Online Communities Website of his/her school (foaf:schoolHomepage),
(SIOC) project aims to link online community sites or to other people that he/she knows (foaf:knows).

32
Social Semantic Web and Semantic Web Services

Figure 7. The main classes and relationships of SIOC

A lot of personal information can be therefore of information over the network. Over the last
represented and parts of the real world’s social couple of decades there have been several at-
graph can be inferred by following the foaf:knows tempts for defining such an infrastructure, such
relationship. as Sun/RPC, CORBA, Microsoft’s DCOM, Java
Of particular importance to the Social Web is RMI, and others. Currently Web Services are the
the support the FOAF vocabulary offers to link favorite and most popular technology for building
the physical persons (foaf:Person) to the accounts distributed systems over the Internet. As a middle-
they hold in a number of social Web sites (e.g. ware technology Web Services represent a new
Flickr, Facebook) through the foaf:holdsAccount generation that tries to mitigate the problems of
property. It is therefore possible through a single legacy integration technologies such as CORBA
FOAF document that could be indexed in a Se- by adopting a more Web friendly substrate. Such
mantic Search engine to join all these different a different approach seems to be needed in order
accounts and the information each of them ex- to support business-to-business integration over
poses. Integration of different social content and the Internet where crossing organization borders
behavior can be achieved and the resulting graph has implications on the security, interoperability,
of information is searchable in an unambiguous scalability, maintenance, flexibility, and other
and machine interpretable way. aspects of application integration. In order to
achieve these goals the Service Oriented Archi-
Semantic Web Services tecture (SOA) has been proposed. Informally
speaking, in such architecture (Web) Services are
Application integration requires an agreed network accessible entities that offer a number of
infrastructure to be in place for the exchange functionalities to their callers. The SOA environ-

33
Social Semantic Web and Semantic Web Services

Figure 8. The main classes and relationships of FOAF

ment should be highly dynamic as suggested by The Web itself is built around these very
a number of real world phenomena, like network directions: open protocols, text based (markup,
instability, changing real world requirements and e.g. HTML) message and document content, and
settings, etc. The need for “late binding” of services abstraction over implementation details. In essence
and clients is important and Figure 9, depicting the underlying infrastructure is roughly based on
the main entities of SOA and their interactions, the following technologies:
shows that a middle service repository or registry
is introduced. This repository stores “offers” of • SOAP messaging format, which is based
functionality as these are published by service on XML, to provide a wrapper format and
providers, and subsequently performs matching protocol for data interchange between Web
with the corresponding “requests”. After some services
matching has been performed the correspond- • Web Service Description Language
ing parties (services and their clients) are free to (WSDL) documents to describe the servic-
communicate and exchange data. es’ functionality and data exchange
On the technology side Web Services put more
emphasis on the following: On top of these a number of standard technolo-
gies have been specified for handling discovery
• Transport over widely accepted Web and (UDDI), security (WS-Security), trust (WS-Trust),
Internet protocols like HTTP/HTTPS and composition (WSBPEL, WSCL), etc. Neverthe-
SMTP less for this discussion the WSDL standard is the
• XML message payloads to provide the most pertinent specification because it specifies
extensibility, introspection, and interoper- in a machine readable format the structure of the
ability required in building complex multi XML messages exchanged.
party systems Integration of computation and functional-
• Platform and programming language ity is an additional field where Semantic Web
independence shows a great potential of use because the Web

34
Social Semantic Web and Semantic Web Services

Figure 9. Web service architecture

Services, at their present incarnation, provide • UDDI (Universal Description, Discovery


syntactic interoperability only. The WSDL service and Integration) allows the discovery of
descriptions are restricted to the syntactic aspects potential business partners on the basis of
of service interaction: how the service can be in- the services they provide. Each business
voked, which operations may be called, what are description in UDDI consists of a busin-
the number and the type of the parameters each essEntity element that describes a busi-
operation needs, etc. However, what the service ness by name, a key value, categorization,
does and in what order its operations have to be services offered (businessServices) and
called in order to achieve certain functionalities is contact information for the business. Each
usually described only in natural language either in businessService element contains descrip-
the comments of a WSDL description or in UDDI tive information such as names and de-
entries or other documentation. Semantic Web scriptions, and also classification informa-
Services (McIlraith, Cao Son, & Zeng, 2001) is tion describing the purpose of the relevant
an “umbrella” definition to include the annotation Web service. Using UDDI, a Web service
of existing Web services with Semantics and their provider registers its advertisements along
publication, discovery, and composition. with keywords for categorization. A Web
The vision again is to make feasible the ma- services user retrieves advertisements out
chine to machine communication by providing of the registry based on keyword search.
machine interpretable descriptions of the services. So far, the UDDI search mechanism relied
Such descriptions will make possible the automatic on predefined categorization through key-
discovery, composition, and invocation of ser- words, but more recently specifications to
vices. Because of this a lot of efforts in metadata use OWL in UDDI are emerging as a uni-
descriptions are centered on the Semantic Web and form way to express business taxonomies.
its technologies, namely RDF and OWL. In the • Semantic Annotations for WSDL and
area of the Semantic Web Services the following XML Schema (SAWDL; Kopecký et. al.,
technologies and standards are relevant to the 2007) is a means to add Semantics inline to
Semantic description of Web services: WSDL. It is actually a set of extensions to

35
Social Semantic Web and Semantic Web Services

WSDL 2.0 but can also be used for WSDL soon as it’s finalized.
1.1. With these extensions the service pro- • The Semantic Web Services Framework
vider can attach references to Semantic (SWSF), initiated by the Semantic Web
concepts for the functionality of an opera- Services Initiative (SWSI, 2004), includes
tion or the type/meaning of a parameter and the Semantic Web Services Language
additional information for the transforma- (SWSL) and the Semantic Web Services
tion (mapping) of the XML data either to Ontology (SWSO). SWSL is a logic-based
(“lift”) or from (“lower”) the correspond- language for specifying formal character-
ing Semantic terms. The Semantic domain izations of Web service concepts and de-
model used is external to these annotations scriptions of individual services. SWSO is
and could be expressed in OWL or other an ontology of service concepts defined us-
ontology language of choice. ing SWSL and incorporates a formal char-
• OWL-S (formerly DAML-S) builds on top acterization (“axiomatization”) of these
of OWL and allows for the description of concepts in first-order logic.
a Web service in terms of a Profile, which • WSMO (Web Services Modeling
tells “what the service does/provides”, a Ontology) defines the modeling elements
Process Model, which tells “how the ser- for describing several aspects of Semantic
vice works”, and a Grounding, which tells Web services (Feier et al., 2005). These el-
“how to access the service” (Martin et al., ements are Ontologies, which provide the
2004). The service profile describes what formal Semantics to the information used
is accomplished by the service, any limita- by all other elements, Goals which specify
tions on service applicability and quality of objectives that a client might have when
service, and requirements that the service consulting a Web service, Web services
requester must satisfy in order to use the that represent the functional and behavior-
service successfully. The process model al aspects which must be Semantically de-
gives details about the Semantic content scribed in order to allow semi-automated
of requests, the conditions under which use, and Mediators that are used as connec-
particular outcomes will occur, and, where tors and they provide interoperability facil-
necessary, the step by step processes lead- ities among the other elements. It also de-
ing to those outcomes. In the process mod- fines the Web Service Modelling Language
el a service can be described as an atomic (WSML) which formalizes WSMO and
process that can be executed in a single aims to provide a rule-based language for
step or a composite process that, similar the Semantic Web.
to a workflow, can be decomposed in other • BioMOBY (http://www.biomoby.org/)
processes based on control structures like is a Web Service interoperability initia-
‘if-then-else’ and ‘repeat-while’. Finally, tive in the field of bioinformatics aiming
Grounding descriptions supply informa- to facilitate the integration of Web-based
tion about the communication protocol and bioinformatics resources. Currently there
other transport information (such as port are two approaches to achieve such inte-
numbers) and the message formats and gration: The first approach, based on the
serialization methods used in contacting Web Services paradigm, is referred to as
the service. The only currently specified “MOBY Services” (MOBY-S), while the
grounding mechanism is based on WSDL second one is called “Semantic MOBY”
1.1 and will be extended to WSDL 2.0 as (S-MOBY) and is based on concepts from

36
Social Semantic Web and Semantic Web Services

the Semantic Web. MOBY-S uses a set of can greatly help.


simple, end-user-extensible ontologies as There have been a lot of discussions about what
its framework to describe data Semantics, will be the “killer application” of the Semantic
data structure, and classes of bioinformat- Web, which means some breakthrough in the
ics services. These ontologies are shared domain that will show beyond any doubt the full
through a Web Service registry system, potential of the Semantic Web. Nevertheless we
MOBY Central, which uses the ontologies think that Semantic Web technologies are used
to Semantically bind incoming service re- slowly and without much “noise” in a lot of dif-
quests to service providers capable of ex- ferent areas and as “extension to the existing Web”
ecuting them. S-MOBY on the other hand are not clearly visible but are certainly catching
employs RDF and OWL and the document on. There is a common view nowadays that the
oriented infrastructure of the WWW (the Semantic Web will not supersede the Syntactic
GET/POST methods of HTTP) for pub- Web in any way but they will happily coexist in a
lishing and retrieving information from its symbiotic manner: the Web of documents will be
discovery servers. enriched by the Web of data and information.
In terms of the core infrastructure what we see
As shown above this is an area of active as emerging trend is the use of simple REST Web
research. So far SAWDL enjoys the approval of services (Fielding & Taylor, 2002) that present a
W3C being one of its recommendations but of small entry barrier and a transition from the SOAP
course is lacking when compared with WSMO and and WSDL Web Services technologies backed
OWL-S. Nevertheless SAWSDL can be combined by big commercial corporations like IBM and
with these most prominent technologies and it Microsoft to more flexible and agile architectures.
remains to be seen whether such approaches are These architectures are more bound to the existing
adequate or something more powerful should be Web and also are more Semantic Web friendly
introduced. since they share common basic infrastructure
and interaction protocols (e.g. Web protocols like
HTTP used as application protocol and not for
Conclusion transport, full support for URI to access network
resources and Semantic concepts, etc.). The whole
The social aspects of the Web show an uprising history of the Web clearly shows that successful
evolution and all the indications imply that this distributed systems of this scale are built on open
trend will continue. The current Web 2.0 sites are access, open protocols, and open source method-
quite successful in attracting people share their ologies combined with collaborative behavior by
data and interests, and build online communi- the people (developers, users) involved.
ties, but the next step will be to enrich them with Research questions and issues for further inves-
more Semantics in the lines of the Semantic Web tigation abound in this Semantic new world. First
to provide a unifying platform for people and of all the issue of trust and security and how this
machines to use and collaborate. The need for is incorporated in the Semantic Web machinery
Semantics (Breslin & Decker, 2007) is evident should be tackled on. For example currently a user
for enhancing the social networking sites with can claim anything in his FOAF document, or a
advanced filtering and recommendation services malicious application can publish RDF informa-
and also to provide data portability and integration tion that contains false statements. The notion of
between different sites and networks. This is an identity and validation of the identity is important
active area where the Semantic Web technologies and there is ongoing work in this area, e.g. the

37
Social Semantic Web and Semantic Web Services

incorporation of user certificates or Web based au- Breslin, J., & Decker, S. (2007). The future of
thentication mechanisms like OpenId8. Semantic social networks on the Internet: The need for
Web has also increased demands for supporting Semantics. IEEE Internet Computing, 86–90.
indexing and reasoning over the managed content. doi:10.1109/MIC.2007.138
The scalability concerns are real when we think
Breslin, J., Harth, A., Bojars, U., & Decker, S.
about a Semantic Web search engine of the size
(2005). Towards Semantically-interlinked online
of Google. Finally the adoption of these technolo-
communities. In Proceedings of the 2nd European
gies by the users needs work to be done in the
Semantic Web Conference (ESWC05), Heraklion,
presentation layers as well. Easy to use, friendly,
Greece, LNCS, 3532, 500-514.
and functional user interfaces are necessary for
making the transition to the Semantic Web more Brickley, D., & Guha, R. (2004). RDF Vocabulary
painless and transparent for the users. description language 1.0: RDF schema. W3C
Recommendation 10 February 2004. World Wide
Web Consortium.
References
Cayzer, S. (2004). Semantic blogging and
Auer, S., Bizer, C., Kobilarov, G., Lehmann, J., decentralized knowledge management. Com-
Cyganiak, R., & Ives, Z. (2007). DBpedia: A munications of the ACM, 47(12), 47–52.
nucleus for a Web of open data. The 6th Interna- doi:10.1145/1035134.1035164
tional Semantic Web Conference (ISWC 2007). Dean, M., Schreiber, G., et al. (2004). OWL Web
Ayers, D. (2007). Evolving the link. IEEE In- ontology language reference. W3C Recommen-
ternet Computing, 11(3), 94–96. doi:10.1109/ dation, 10.
MIC.2007.53 Decker, S. (2006). The social Semantic desktop:
Beckett, D., & Berners-Lee, T. (2008). Turtle Next generation collaboration infrastructure. In-
- Terse RDF triple language. W3C Team Submis- formation Services & Use, 26(2), 139–144.
sion. Feier, C., Roman, D., Polleres, A., Domingue, J.,
Berners-Lee, T., Bray, T., Connolly, D., Cotton, P., Stollberg, M., & Fensel, D. (2005). Towards intel-
Fielding, R., Jeckle, M., et al. (2004). Architecture ligent Web services: The Web service modeling
of the World Wide Web, Volume One. W3C, ontology (WSMO). International Conference on
Intelligent Computing (ICIC).
Berners-Lee, T., Hendler, J., & Lassila, O. (2001).
The Semantic Web. Scientific American, 284(5), Fielding, R. T., & Taylor, R. N. (2002). Prin-
28–37. cipled design of the modern Web architecture.
ACM Transactions on Internet Technology, 2(2),
Bizer, C., Heath, T., Idehen, K., & Berners-Lee, 115–150. doi:10.1145/514183.514185
T. (2008). Linked data on the Web (ldow2008). In
WWW ‘08: Proceeding of the 17th International Greaves, M. (2007). Semantic Web 2.0. IEEE
Conference on World Wide Web (pp. 1265-1266). Intelligent Systems, 22(2), 94–96. doi:10.1109/
New York: ACM. MIS.2007.40

Bojars, U., Breslin, J., & Moller, K. (2006). Using Gruber, T. (2007). Collective knowledge systems:
Semantics to enhance the blogging experience. Where the social Web meets the Semantic Web. To
In Proceedings of 3rd European Semantic Web appear in Journal of Web Semantics, 2007
Conference,ESWC 2006, (pp. 679-696).

38
Social Semantic Web and Semantic Web Services

Kiss, J. (2008). Web 3.0 is all about rank and rec- Surowiecki, J. (2005). The wisdom of crowds.
ommendation. The Guardian, February 4 2008, Anchor. Paperback.
Kopecký, J., Vitvar, T., Bournez, C., & Farrell, SWSI. (2004). Semantic Web services initiative
J. (2007). SAWSDL: Semantic annotations for (SWSI).
WSDL and XML schema. IEEE Internet Comput-
Uschold, M., & Gruninger, M. (1996). Ontolo-
ing, 60–67. doi:10.1109/MIC.2007.134
gies: Principles, methods and applications. The
Lassila, O., Swick, R., et al. (1999). Resource Knowledge Engineering Review, 11(2).
description framework (RDF) model and syntax
Völkel, M., Krötzsch, M., Vrandecic, D., Haller,
specification. W3C Recommendation, 22, 2004-
H., & Studer, R. (2006). Semantic Wikipedia.
03.
Proceedings of the 15th international conference
Martin, D., Paolucci, M., McIlraith, S., Burst- on World Wide Web, (pp. 585-594).
ein, M., McDermott, D., McGuinness, D., et al.
(2004). Bringing Semantics to Web Services:
The OWL-S Approach. In Proceedings of the
Key Terms
First International Workshop on Semantic Web
Services and Web Process Composition (SWSWPC Semantic Web (SW): The creator of the World
2004), (pp. 6-9). Wide Web Tim Berners-Lee defines the SW as
McIlraith, Sheila A., Cao Son, Tran, & Zeng, “a web of data that can be processed directly
Honglei. (2001). Semantic Web Services. or indirectly by machines”. A similar definition
IEEE Intelligent Systems, 16(2), 46–53. coming from the World Wide Web Consortium
doi:10.1109/5254.920599 (W3C) describes the Semantic Web as “a Web that
includes documents, or portions of documents,
Moller, K., Bojars, U., & Breslin, J. (2006). Using describing explicit relationships between things
Semantics to enhance the blogging experience. and containing Semantic information intended for
In I3rd European Semantic Web Conference automated processing by our machines.”
(ESWC2006), LNCS, 4011, 679-696.
O’Reilly, T. (2005). What is Web 2.0: Design pat- Social Web/Web 2.0: The way people so-
terns and business models for the next generation cialize and interact with each other through the
of software. World Wide Web. This term is also used to denote
a large number of Web sites that are devoted to
Prudhommeaux, E., & Seaborne, A. (2008). people and their social interactions through the
SPARQL query language for RDF. W3C Recom- creation of online communities of users that
mendation 15 January 2008. share digital content and information, discuss,
Ramakrishnan, R., & Tomkins, A. (2007). To- or enable communication in any possible, Web-
ward a peopleweb. Computer, 40(8), 63–72. facilitated way.
doi:10.1109/MC.2007.294
Web Service (WS): A Web Service is defined
Schaert, S. (2006). IkeWiki: A Semantic Wiki by the World Wide Web Consortium (W3C) as “a
for collaborative knowledge management. In software system designed to support interoperable
Proceedings of the 15th IEEE International Work- machine-to-machine interaction over a network”.
shops on Enabling Technologies: Infrastructure Since this definition is quite general the term
for Collaborative Enterprises, (pp. 388-396). “Web Service” commonly refers to systems that

39
Social Semantic Web and Semantic Web Services

communicate using XML messages that comply Wiki/Criticism_of_Wikipedia


2
with the SOAP messaging format. In such systems, http://en.Wikipedia.org/Wiki/Seigentha-
there is often machine-readable description of the ler_incident
3
operations offered by the service written in the http://en.Wikipedia.org/Wiki/Ajax_(pro-
Web Services Description Language (WSDL). gramming)
4
http://en.Wikipedia.org/Wiki/Comet_(pro-
gramming)
ENDNOTES 5
http://en.Wikipedia.org/Wiki/Web_syndica-
tion
1
Despite the irony in itself as a fact, an exten- 6
http://www.opensearch.org
sive survey of this criticism can be found in 7
http://pipes.yahoo.com
the Wikipedia at http://en.WikipedSia.org/ 8
http://openid.net/

This work was previously published in Virtual Community Practices and Social Interactive Media: Technology Lifecycle and
Workflow Analysis, edited by D. Akoumianakis, pp. 207-225, copyright 2009 by Information Science Reference (an imprint of
IGI Global)

40
41

Chapter 1.4
Semantic Web in Ubiquitous
Mobile Communications
Anna V. Zhdanova
The Telecommunications Research Center Vienna, Austria

Ning Li
University of Surrey, UK

Klaus Moessner
University of Surrey, UK

Abstract INTRODUCTION

The world becomes ubiquitous, and mobile com- Nowadays, mobile and Web environments converge
munication platforms become oriented towards in one shared communication sphere. Technologies
integration with the web, getting benefits from the stemming from Semantic Web and Mobile Com-
large amount of information available there, and munication fields get combined to achieve this
creation of the new types of value-added services. convergence towards the vision of ontology-enabled
Semantic and ontology technologies are seen as ubiquitous mobile communication. Knowledge
being able to advance the seamless integration of Management and Semantic technologies fields
the mobile and the Web worlds. We provide back- produce ways to describe, specify and manage infor-
ground information on the Semantic Web field, mation in a machine processable form, in particular,
discuss other research fields that bring semantics into acquire, evolve, reuse, and combine knowledge
play for reaching the ontology-enabled ubiquitous (Fensel, 2001). Certain formats and protocols stem-
mobile communication vision, and exemplify the ming from these fields are already being applied
state of the art of ontology development and use in to telecommunications: vCard1, CC/PP2, UAProf3.
telecommunication projects. However, these specifications are only applicable to
a limited number of telecommunication scenarios,
DOI: 10.4018/978-1-60566-028-8.ch011

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Semantic Web in Ubiquitous Mobile Communications

and management of information about resources and development fields and challenges in the
in mobile environment could be substantially area. In Section 4, two illustrative case studies
improved, e.g., by alignment of heterogeneous for the converged area are described. Section 5
information sources in knowledge-based service concludes the chapter and Sections 6 indicates
enablers. future research directions.
Ontologies and architecture knowledge layers
play an ever-increasing role in service platforms
and mobile communications. As integration of WHY SEMATICS IN UBIQUITOUS
Telco, Internet and the Web takes place, in order MOBILE COMMUNICATIONS?
to achieve interoperability, telecommunication
systems and services tend to rely on knowledge In this section we motivate why combination
represented with the use of shared schema, i.e., of Semantic Web technology with ubiquitous
on ontologies similar to as envisioned on the mobile communications is beneficial. Semantic
Semantic Web (Tarkoma et al., 2007). However, technologies in mobile communication have been
specific ontology-based implementation solutions somewhat considered to the less extent comparing
for mobile systems are rare, and best practices for to other fields, such as semantics in e-sciences,
such interoperability are not established. In this e-government, e-enterprise, e-communities, etc.
chapter, we address a problem of ontology-based However, as the mobile world starts to integrate
interoperation in order to integrate independent with the Web world in delivering new value-
components in a system providing value-added added services, the area of semantics ubiquitous
mobile services. mobile communication inevitably gains a larger
We present the overall state of the art ontology- importance and potential.
related developments in mobile communication Ubiquitous computing, also referred to as
systems, namely, the work towards construction, pervasive computing, is the seamless integration
sharing and maintenance of ontologies for mobile of devices into the users every day life. Applica-
communications, reuse and application of ontolo- tions should vanish into the background to make
gies and existing Semantic Web technologies in the user and his tasks the central focus rather than
the prototypes. Social, collaborative and technical computing devices and technical issues (Weiser,
challenges experienced in the project showcase the 1991). When applying to mobile communication
need in alignment of ontology experts’ work across scenarios, ubiquitous computing can be viewed
the mobile communication projects to establish as when user moves around and changes circum-
the best practices in the area and drive standard- stances, he can always be connected and well
ization efforts. We indicate certain milestones in served without being aware of the technical issues
integration of Semantic Web-based intelligence under the scene. To achieve the goal, information
with Mobile Communications, such as performing from all the involving participants, such as user,
ontology construction, matching, and evolution network, service provider etc., needs to be col-
in mobile service systems and alignment with lected, shared and interoperable with each other,
existing heterogeneous data models. known by one or more operational agents but
The chapter is structured as follows. In Sec- agnostic to the user. Such information is diverse
tion 2 we provide a motivation for discussing the in their language, format and lack of semantic
convergence between the areas of Semantic Web meaning for autonomous processing by computer
and ubiquitous mobile communications. Section or operational agent. The Semantic Web can be a
3 gives an overview of the core ontology tech- rescue with its vision to achieve global informa-
nologies involved, related and relevant research tion sharing and integration.

42
Semantic Web in Ubiquitous Mobile Communications

An example of combination of the two fields MPEG-21, multimedia content description, such
can be a service enabler that could be used by as MPEG-7, service discovery and service context
other services and thus make their construction modelling, such as UPnP, Jini, Bluetooth SDP,
simpler. For example, such an enabler could ac- agent technologies, such as FIPA, ACL. Network
cess distributed information on user’s location, mobility, security and QoS management, such as
availability and friends and inform other services Hierarchical mobile IP, IPSec, DiffServ etc. The
about which groups of friends are available and combination of these technologies with Semantic
located at the same place. Services that assist Web has become an inevitable trend in a ubiqui-
with scheduling business meetings or parties, or tous mobile communication environment. Some
the ones that are targeted at selling products for of such joint developments are presented later in
groups of users can be among the services that need Section: Relevant research fields.
such information from the enabler. To achieve the
output, the enabler would take external structured Semantic Web Existing
data represented in a formal way (e.g., in RDF): Developments
for instance, information about user’s availability
from PIDF files, information on who is a friend Existing developments of the Semantic Web in-
of whom from FOAF profiles, information about clude languages (i.e. core formalisms to specify
location from data stemming from such standards domain knowledge or services), methodologies
as IETF RFC4119, RFC4589. Then the enabler and tools. In this section we outline the major
would combine the gathered information, apply developments in these areas and indicate their
certain rules to deduce the result and pass it to role and contributions in the area of Semantic
other mobile services via an interface. Web enabled mobile platforms.

Languages and Formalisms


GROUNDING FOR SEMANTICS
IN UBIQUITOUS MOBILE RDF(S)
COMMUNICATIONS
RDF (Lassila & Swick, 1999; Manola & Miller,
In this section, we describe existing developments 2004) became a W3C recommendation in 1999.
relevant for the combined field of ubiquitous mo- It is a general-purpose language for representing
bile communications and the Semantic Web. resources on the web in terms of named proper-
ties and values (McBride, 2004). With RDF it is
Ubiquitous Mobile Communication not possible to define the relationships between
Existing Developments properties and resources. For this purpose, RDF
Schema (Brickley & Guha, 2004) has been speci-
A lot of work has been undertaken to implement fied. It became a W3C recommendation in 2004
the vision of ubiquitous mobile communications and is basically an extension of RDF. More specifi-
by investigating the underlying technologies. cally, it is a formal description language for eligible
Examples include, but not limited to, user-related RDF expressions. In particular, a schema defines
context collection, such as sensor network, Blu- the kinds of properties available for resources
etooth, GPS, user-related context modelling and (e.g., title, author, subject, size, colour, etc.) and
transmission, such as CC/PP, UAProf, 3GPP, the kind of resource classes being described (e.g.,

43
Semantic Web in Ubiquitous Mobile Communications

books, Web pages, people, companies, etc.). RDF classification hierarchies and simple constraints.
Schema is a simple ontology and a simple ontol- OWL-Lite corresponds semantically to the formal
ogy definition language. RDF and RDF Schema Description Logic SHIF(D) and cannot express
are usually denoted RDF(S). the whole RDF vocabulary.
RDF(S) bases on some syntactical principles OWL-DL contains the language constructs of
of XML (e.g. URIs) and has been equipped with OWL-Lite. OWL-DL corresponds semantically
an XML syntax as well. The most basic Semantic to the Description Logic SHOIN(D). Although
Web language which provides the syntactical strictly more expressive than OWL-Lite, it still
basis for all other Semantic Web languages is provides computational completeness and decid-
RDF(S). RDF(S) is not provided completely with ability.
a formal logical semantics, thus reasoning is only OWL Full does not correspond to a formal logic
on partially supported. anymore as it builds upon the complete RDF(S)
vocabulary which also lacks a correspondence to
Topic Maps a formal logic. The language incorporates maxi-
mum expressive power and syntactic freedom,
Topic Maps are a data modelling language and but offers no computational guarantees.
became an ISO standard (ISO/IEC 13250) in
2000. A Topic Map offers a means to create an Semantic Web Languages
index of information which resides outside of that in Progress
information. It describes the information in docu-
ments and databases by linking into them using In this subsection, we consider Semantic Web
URIs. A Topic Map consists of topics, associations languages which have been submitted to the W3C
(relationships between topics), occurrences (infor- and thus have communities promoting them. At
mation resources relevant to a topic). Topics and least some of them can be expected to become
occurrences can by typed. Types in Topic Maps W3C recommendations. Examples of such lan-
are themselves topics and thus there is no real guages are:
difference between a topic and a type.
There exists SGML, XML and RDF language • Languages based on the Logic
support for Topic Maps. However, they are very Programming Knowledge Representation
simple and do not have a formal semantics and thus paradigm: The trend to the aforementioned
no sophisticated inference support. Nevertheless, paradigm exists already since the year
because of their simplicity, they are often used in 2000 when the development of RuleML4
industry applications. has started. RuleML is a set of languages
revolving around the Logic Programming
OWL paradigm and being equipped with an RDF
syntax. Other examples of Semantic Web
OWL (Dean & Schreiber, 2004) became a W3C Languages with Logic Programming se-
recommendation in 2004. OWL is mainly based mantics are WRL5, a set of three layered
on OIL and DAML+OIL, which are obsolete Se- rule languages of increasing expressivity,
mantic Web languages and therefore not mentioned and SWRL6, a language which combines
further here. OWL is equipped by an RDF syntax OWL and RuleML but is computationally
and includes three sub languages: intractable. Furthermore, a W3C work-
OWL-Lite roughly consists of RDF(S) plus ing group7 has been formed for estab-
equality and 0/1-cardinality. It is intended for lishing standards for Semantic Web rule

44
Semantic Web in Ubiquitous Mobile Communications

languages. As the ontology data are processed within the


• Semantic Web Service Modelling ubiquitous mobile communication applications,
Languages: Semantic Web Services will exploration and reuse of further ontology manage-
play an important role in the Semantic Web ment technology is on the roadmap of the research
as they combine Web Services with seman- and development field.
tics. Examples for Semantic Web Services
Languages are WSML8 and SWSL9. The Major Challenges and Approaches
languages serve for the specification of on-
tologies describing Semantic Web Services. In this section, we define major challenges that
E.g., WSML is used to describe WSMO10 are under development or need to be developed
and SWSL is used to describe SWSO11. in the joint area. From ubiquitous mobile com-
munication point of view, the challenges can be
Ontologies and Tools viewed from three perspective, i.e., from the user,
the network operator, and the service provider.
Apart from the languages to describe data, specific From user’s point of view, there is an expectation
ontologies related to the mobile communication of autonomous, non-stop service being provided
domains and appropriate ontology management with satisfying quality whatever terminal he/she
tools are necessary to implement the vision of uses, whenever he/she needs and wherever he/
ubiquitous mobile communications. she goes without having to set the configuration.
In a nutshell, the state of the art in development From network perspective, there is a challenge
of the ontologies addressed by ubiquitous mobile of ensuring the service delivery by providing a
communications comprises: smooth handover, a guaranteed QoS and security
level etc. when delivery circumstance changes,
• Ad-hoc small-size schemata on certain for example, from one type of network to another.
general purpose topics are specified in on- From service provider point of view, the antici-
tology languages. pation is to provide only one version of service,
• Detailed, XML-based standards on certain which, however, can be used by any device via
narrow telecommunications topics. any networks. To face these challenges, when
designing a ubiquitous service delivery based on
Certain standardisation schemata and activi- ontologies, the information from and about the
ties to be considered for the development of the user, network condition, the content/service being
ontology framework for mobile communications provided, together with any context information
are listed later in Section: Mobile Ontology. Ad- (to support personalization and service push to
ditional efforts coming from the Semantic Web the user), need to be described in an unambigu-
community are listed in Table 1. ous and interoperable manner in order to perform
Typically, existing ontology management effective service delivery. Therefore, common
tools are adopted and explored in the semantic ontologies describing such domain knowledge, as
telecommunications projects. Stemming from the well as best practices in their reuse are required.
SPICE project (Zhdanova et al., 2006), examples Generally, challenges faced from ontology point
of relevant and popular ontology management of view include:
tools used, as well as encountered problems in
their exploitation and an expected resolution times, • Heterogeneity: Resolving inconsistencies,
are provided in Boxes 1 and 2. format differences (syntactic and semantic

45
Semantic Web in Ubiquitous Mobile Communications

Table 1. Ontologies related to the mobile ontology

Development Sta-
Ontology Name Producer URL Description
tus
The motivation for the MeNow schema is to be able
to describe a variety of aspects of the current status of
someone, either online or off, in a way that the data can
http://crschmidt.net/
Chris be easily aggregated or retrieved. This schema allows
MeNow foaf/menow/menow.
Schmidt the definition of a variety of terms that would be com-
rdf
mon in many applications: describing the current book
you are reading, music you are listening to, mood you
are in, and more.
Pervasive Computing Standard Ontology (PERVA-
SIVE-SO) is a set of RDF/OWL ontology documents.
Each ontology document is identified by a unique XML
namespace and defines the ontologies of a specific
domain.
In a pervasive computing environment, computer
Pervasive SO – systems often need to access the profiles and the
http://pervasive.se-
Describing User Harry Chen, preferences of a user in order to provide services and
manticweb.org/doc/ Frozen in 2004
Profile and Pref- UMBC information that are tailored to the user. The profile of
ont-guide/part1/
erences a user includes typical contact information (telephone
numbers, email addresses, name, etc.) and informa-
tion that describe other computing entities that can act
on the behalf of the user (e.g., the personal agent of a
user). The preference of a user is a description of the
environment state that the user desires the computer
systems to honor or achieve whenever it is possible.
The Platform for Privacy Preferences Project (P3P)
enables Web sites to express their privacy practices in
a standard format that can be retrieved automatically
Brian and interpreted easily by user agents. P3P user agents
Platform for Pri- http://www.w3.org/
McBride, will allow users to be informed of site practices (in both Frozen in 2002
vacy Preferences TR/p3p-rdfschema/
HP machine- and human-readable formats) and to automate
decision-making based on these practices when ap-
propriate. Thus users need not read the privacy policies
at every site they visit.
http://www.wasab. Definitions of various terms related to (typically)
Morten Fred-
Gadget dk/morten/2004/10/ electronic gadgets such as GPS receivers, cameras and Frozen in 2004
eriksen
gadget mobile phones.
ConOnto describes the different aspects of context-
aware systems.
http://www.site.uot- ConOnto includes location, time, activities, software
ConOnto: Con- Mohamed
tawa.ca/~mkhedr/ and hardware profiles.
text Ontology Khedr
contexto.html ConOnto also includes meta-information that describes
negotiation and fuzzy ontologies to be used in systems
that will negotiate and infer about context information.
Ambient Net-
Anders
works: General, http://kiwi.intra.son-
Karlsson, Created in 2005
Cost, QoS On- era.fi/an_costs.owl
TeliaSonera
tology

46
Semantic Web in Ubiquitous Mobile Communications

Box 1.
Name……………………….. Protege
Website…………………….. http://protege.stanford.edu/
White page………………… n/a
Main characteristics……….. Ontology editor
Open problems Relevance Term
1 2 3 4 5 0-3 3-6 6-12
Very low Low Normal High Very high short medium long
Needs improvement of usability
X X
features, robustness

Box 2.
Name………………………. Jena
Website…………………….. http://jena.sourceforge.net/
White page………………… n/a
Main characteristics……….. A Semantic Web Framework for Java. Ontology API and implementation,
supports RDF(S), OWL, performs basic ontology management and reasoning (similar idea as Xerces
for XML)
Open problems Relevance Term
1 2 3 4 5 0-3 3-6 6-12
Very low Low Normal High Very high short medium long
Scalability: works slowly on large
X X
volumes of data

differences in formalization), business pro- alike to YouTube, Flickr, LinkedIn, Google


cess mediation. Base, etc.).
• Versioning when merging or combining • Semantically described mobile and web
ontologies representing different knowl- services (no yet widely accepted “stan-
edge domains, ontology and instance dard” solutions).
data evolution and maintenance during • Service composition and discovery, user-
updates. driven creation of new services by com-
• Scalability: Scalable repositories for on- posing service enablers, mash-ups.
tology instance data (currently popular on- • Integration with the non-semantic web
tology management toolkits such as Jena services and formats, which use tradition-
and Sesame do not always meet industrial al technologies such as WSDL, SOAP and
standards). UDDI, XML.
• Ontology and instance data visualiza- • Integration with legacy applications,
tion, user interfaces. which are not terminologically a service,
• User/community generated content: for- but can be upgraded to be a service, e.g.
malization, acquisition and employment MPEG codec.
(when building innovative mobile servic-
es in Web 2.0 style of social applications

47
Semantic Web in Ubiquitous Mobile Communications

Relevant Research Fields description tools for generic media entities, such
as vector, time and more complex media entities.
The following research areas are related to, impact The latter can be grouped into 5 different classes
and will be potentially impacted by the described according to their functionality: Content descrip-
involvement of the Semantic Web in Mobile tion, Content management, Content organization,
Communications. Navigation and access and User interaction.
MPEG-21, the 21st century multimedia frame-
Multimedia work, goes further and provides tools to describe
the environment to enable transparent multime-
Today, the amount of digital multimedia infor- dia creation, delivery and consumption between
mation is growing over the World Wide Web, heterogeneous environments. The main parts
in broadcast data streams and in personal and are Digital Item Declaration (DID), Digital Item
professional databases. One of the major chal- Identification (DII), Intellectual Property Manage-
lenges for ubiquitous mobile communication is ments and Protection (IPMP), Rights Expression
to enable any mobile devices, e.g. mobile phone, Language (REL), Rights Data Dictionary (RDD)
PDA, to access, exchange and consume a rich set and Digital Item Adaptation (DIA). The Digital
of multimedia content seamlessly over dynamic Item Declaration (DID) specification contains
and heterogeneous networks. The need of se- three normative sections, a model to describe a
mantic description of the multimedia information set of abstract terms and concepts to form a useful
becomes apparent. The MPEG-7 and MPEG-21 model for defining Digital Items, a representation
are the dominant efforts for multimedia content to describe the syntax and semantics of each of the
and service description framework. Digital Item Declaration elements, and a Schema
MPEG-7 is known as the multimedia descrip- comprising the entire grammar of the Digital
tion standard and offers several tools, i.e. Descrip- Item Declaration representation in XML. The DII
tion Schemes, to annotate multimedia content at specification provides mechanisms to uniquely
different levels. The main parts are: Description identify Digital Items, Intellectual Property re-
Definition Language (DDL), Visual, Audio and lated to the Digital Items such as abstractions,
Multimedia Description Schemes (MDS). The Description Schemes and types of Digital Items.
DDL is a language that allows the creation of new IPMP is an extended efforts based on MPEG-4
Description Schemes and, possibly, Descriptors. to develop new systems and tools with enhanced
It also allows the extension and modification of interoperability. The REL, together with the
existing Description Schemes. The DDL is based RDD that supports the REL and provides exten-
on XML Schema Language, but with MPEG-7 sive semantics, provides a universal method for
extensions specifically for audiovisual description. specifying rights and conditions associated with
The Visual description tools provide structures the distribution and use of digital items and thus
to describe basic visual features, such as color, facilitates the creation of an open DRM architec-
texture, shape, motion and localization etc. The ture. DIA provides tools to describe the Digital
Audio description tools provides structures for Item usage environment including: Usage char-
the description of audio features that are com- acteristics, such as user info, usage history, User
mon across many applications, such as spectral, preferences and physical characteristics such as
parametric, and temporal features, and that are disabilities, Device characteristics such as display,
application-specific features, such as audio memory and battery, Network characteristics,
indexing, recognition and signature. MPEG-7 such as error characteristics and bandwidth, and
Multimedia Description Schemes provides the Natural environment characteristics such as noise

48
Semantic Web in Ubiquitous Mobile Communications

and illumination. This is to facilitate transparent to the service description by extending original
access to distributed digital items by shielding WSDL elements. SOAP, as the message exchange
users from the technical complexity, such as protocol and originally XML-based, can be com-
network and terminal installation, management bined with RDF and OWL in order to introduce
and implementation issues. semantics to assist the flexible service invocation
(Zhao, 2004). Similarly, enabling UDDI to store
Web Service (SOA) semantic markup and handle semantic enquiries
has been investigated in recent years (Luo et al.,
Service-Oriented Architecture (SOA), especially a 2006). Correspondingly, service-oriented archi-
Web service-based SOA, has the potential in speed- tectures require machine-processable semantics
ing up the application development process and to achieve its full potential. In particular, DERI14,
the agility in responding to the change of business targets this challenge by offering a set of tools and
needs. This is due to the loose coupling of client techniques ranging from ontology construction to
from service and the set of standard protocols description language. The later include Web Ser-
and technologies used by the Web service, such vice Modelling Ontology (WSMO), Web Service
as XML, WSDL, SOAP, and UDDI. The inherent Modelling Language (WSML) and a Web Service
features of SOA, i.e. reusability, interoperability, Execution Environment (WSMX)15.
scalability and flexibility, can virtually meet the
requirement of a supportive framework for ubiq- Security, Privacy, Trust
uitous mobile communication.
Coming along with Semantic Web is Semantic In recent years, the security and trust aspects of
Web service, where Web service is described with Web services are standardised by OASIS16 with
added computer-processable semantics, and thus a WS-security specification released in 2004.
a number of services can be concatenated au- WS-security insets security-related information
tonomously to compose a new service for a more to Web service messaging that provides for mes-
complex task. This will benefit service provision sage integrity and confidentiality using security
in a ubiquitous environment where all information token and digital signatures. The use of Semantic
from user, network, together with the requested Web technologies enables Web into a genuinely
service and any intermediate service are required distributed and global content and service pro-
to make a delivery decision autonomously. In ac- vider. Inherent with this are the issues of more
cordance with this advance, the set of standard widespread security, trust, information quality
protocols and technologies for Web services are and privacy.
evolving to reach their semantic counterparts or To achieve a security solution for Semantic
brand new standards are created. For example, the Web service, the traditional security solutions can
ontology languages, RDF, RDFS, and OWL are be described as one of the contextual information
developed to add computer-processable semantics attached to the service and can be interpreted on the
on top of the exiting syntax provided by XML. other end at semantic level. This solution has the
WSDL specifies a way to describe the abstract advantage of not requiring to design a bottom-up
functionalities of a Web service and concretely how Semantic Web service security architecture and
and where to invoke it. Semantic Annotations for thus provides the service provider with flexibility
WSDL (SAWSDL)12 defines mechanisms using of control. However, embedding security and trust
which semantic annotations can be added to WSDL policies into every Web service may not appear
components based on an earlier effort, namely to be an attractive solution and can result in tight
WSDL-S13, which adds semantic expressivity coupling between services and particular security

49
Semantic Web in Ubiquitous Mobile Communications

implementations. An alternative is to design an knowledge from various heterogeneous parties,


integrated Semantic Web security framework with embedding Semantic Web technology into the HCI
security mechanisms available at various layers of design is envisioned to be a necessity for automat-
the network. This may provide a comprehensive ing the deployment of rich user interfaces.
solution when more security and trust issues and
challenges arise from the traditional communica- Lower Layer of Mobile Communication
tion domain. For example, when seamless inter-
connecting heterogeneous networks, particularly One of the important contextual information to
when security issue are jointly considered with be included, and maybe semantically described,
other issues, such as QoS and mobility manage- when customizing Semantic Web content and
ment, in the overall communication process. service to the user ubiquitously are the network
conditions in the delivery path, e.g. bandwidth,
Human Communication Interface QoS, Security, latency, jitter etc. Signalling and/
or information exchange is required between the
The design of the Human Communication In- application layer and the underlying network
terface (HCI) for Web applications has always layers for request and response. Enhancements
been of great importance. The Web technology to the communication protocols with semantic
is evolving with Semantic Web, the interaction capabilities are envisaged to be required in order
must also evolve. With the emergence of the Se- to assure user’s satisfaction in ubiquitous service
mantic Web and Semantic Web services, which delivery. In addition, there have been extensive
give information well-defined meaning and enable efforts to tackle the network layer integration of
computer work in cooperation with humans, the QoS, security and mobility management over het-
interactions with Web-delivered information has erogeneous network in a mobile environment.
become possible and thus the complexity of the
human communication interface has increased. Autonomous Computing
For example, instead of being an information re-
ceiver only, user can interact with information to The vision of autonomous computing is to enable
construct new information and build knowledge. computing system operate in a fully autonomous
In addition, with the user terminal getting smaller manner. The challenges of autonomous computing
and smaller in size like PDA, Pocket PC, together are robustness, simplicity for the end-user and
with the service being more customized to the seamless integration (Hercock, 2002). With the
user’s personal need and preference, the human vision of being a global information integrator
interface design becomes even more challenging by making information computer-interpretable,
than ever. Semantic Web technologies can help realizing the
In ubiquitous mobile communication environ- vision of autonomous computing, particularly in
ment, human computing interfaces form one of the a ubiquitous mobile communication environment
major contextual information of the user as well as where constant changes take place and autono-
one of the major component in the delivery path. mous process are expected.
Therefore, it is essential to bring interaction design
principles with other contextual information into Grid, Semantic Grid
the semantically structured information in order to
facilitate this evolution of the Web. On the other Grid development is targeted at the problem of
hand, with the popularity of Semantic Web and efficiently using computing power of distributed
wide acceptance of its technology in integrating resources for achieving a common set of tasks.

50
Semantic Web in Ubiquitous Mobile Communications

Typical Grid toolkits include GLOBUS17. Seman- applications with the Internet increase. To ensure
tic Grid is oriented towards enhancing typical interoperation of mobile and Web services, ap-
Grid services or processes with ontology-based plications and tools (running on heterogeneous
descriptions (Goble et al., 2004). Different types various service platforms in such a sphere), de-
of Grid resources can be used by ontology–enabled velopers need to have a shared specification of
mobile services to reach their goals. objects belonging to the sphere and their roles.
Certain ontologies have already been developed
for the mobile communication domain by em-
ONTOLOGY FRAMEWORK STUDIES ploying area with employment of Semantic Web
formalisms (Korpipää et al., 2004; Pfoser et al.,
In this section, we describe ongoing research and 2002). However, widespread and global adoption
development combining the areas of ubiquitous of such ontologies remains a challenge.
mobile communication and Semantic Web. One Approaching the problem of interoperation
can view convergence of Semantic technologies between the Web and mobile service technologies,
with mobile communication from two sides: in- Mobile ontology, a comprehensive “higher-level”
clusion of the Semantic technologies in solutions ontology for mobile communication domain, is
delivered by mobile communication project from being developed. Currently, definition and imple-
one side; and mobile communication use cases in mentation of the Mobile ontology is managed as
core Semantic technology projects from the other a collaborative effort amongst participants of the
side. On the one hand, work on development of EU IST SPICE Integrated Project.
mobile communication applications, enablers
and services with involvement of ontologies has Mobile Ontology Introduction
been carried out in the following mobile com-
munication projects: SPICE18, MobileVCE19, What’s Mobile Ontology for?
Mobilife20, OPUCE21, Ambient Networks22, etc.
On the other hand, research and development Mobile Ontology is being developed as a com-
involving mobile aspects has been carried out in prehensive “higher-level” ontology for mobile
the following Semantic Web projects: TripCom23, communication domain. The ontology is a machine
SWING24, ASG25, SmartWeb: Mobile Access to readable schema intended for sharing knowledge
the Semantic Web26, etc. In this section, we pro- and exchanging information both across people
vide detailed illustrating examples of the ontol- and across services/applications, and it covers
ogy work carried out for mobile communication domains related to mobile communications, spe-
solutions, specifically, mobile ontology (SPICE cifically, addressing persons, terminals, services,
project) and an ontology solution for multimedia networks.The added values of Mobile Ontology
(MobileVCE project). are:

Mobile Ontology • Providing an easy and formal way to refer-


ence objects from the mobile communica-
Mobile environments and the Web converge tion domain (in particular, to serve as an
forming a shared Distributed Communication exchange format between mobile service
Sphere (DCS). This causes the appearance of enablers).
new settings to be supported, e.g., when the user • Providing an opportunity to implement en-
utilizes mobile and fixed devices to interact with hanced, ontology-based reasoning.
systems. Interaction and connectivity of mobile • Providing a formal representation of the

51
Semantic Web in Ubiquitous Mobile Communications

domain to be used in research and de- mobile_ontology.owl”>


velopment projects, and for educational …
purposes. <Device
rdf:ID=”BlueSonyEriccson”>
Mobile Ontology Overview <belongsTo>
<Person rdf:ID=”AnnaZ”>
DCS-related vocabulary terms, grouped in broad <owns rdf:resource=”#BlueSonyEri
categories, are presented in Figure 1. ccson”/>
Mobile Ontology, in particular, its DCS Vo- <rdfs:comment
cabulary (Zhdanova et al., 2006) definitions are rdf:datatype=”http://www.
written using RDF and OWL (Manola & Miller, w3.org/2001/XMLSchema#string”
2004; Dean & Schreiber, 2006) that makes it ></rdfs:comment>
easy for software (both Web-based and mobile- </Person>
oriented) to process facts about the terms in the </belongsTo>
DCS vocabulary, and consequently about the </Device>
things described in DCS documents. A DCS …
document/instance data can be combined with </rdf:RDF>
other DCS documents to create unified sources
of information. RDF/S and OWL have been chosen as formats
to represent the mobile ontology, as they are
Example current recommendation ontology languages of
W3C and have a relatively large tool support for
A very basic annotation describing the state of implementation of enablers and applications.
the communication model is: Starting from the DCS ontology, the Mobile
ontology has developed a new structure and
evolved in Mobile ontology Core and Mobile
<?xml version=”1.0”?> subontologies. Mobile ontology Core comprises
<rdf:RDF the telecommunications domain concepts and
xmlns:rdf=”http://www. properties that occur most commonly and in
w3.org/1999/02/22-rdf-syntax- various subdomains or application types. Mobile
ns#” subontologies contain further details related to
xmlns:xsd=”http://www. specific topics in telecommunications, and its
w3.org/2001/XMLSchema#” items are linked to the items of the Mobile ontol-
xmlns:rdfs=”http://www. ogy Core. The Mobile ontology Core overview is
w3.org/2000/01/rdf-schema#” depicted at Figure 2.
xmlns:owl=”http://www. Currently the subontologies on the follow-
w3.org/2002/07/owl#” ing topics are being represented in the Mobile
xmlns:p1=”http://www.owl-ontolo- ontology infrastructure: Profile, Service, Service
gies.com/assert.owl#” Context, DCS, Service Roaming, Rules and rec-
xmlns=”http://www.ist-spice.org/ ommendations, Presence, Location, and Content.
mobile_ontology/2006/5/26/mo- The up-to-date ontology versions and the status
bile_ontology.owl#” of the work are represented at the Mobile Ontol-
xml:base=”http://www.ist-spice. ogy website27.
org/mobile_ontology/2006/5/26/

52
Semantic Web in Ubiquitous Mobile Communications

Figure 1. DCS-related Classes and Properties of Mobile Ontology

Reuse of Schemata and Ontologies significant communities. Such schemata can be


either an output of the standardization bodies or
Certain schemata covering the domains of the coming in a “bottom-up” manner from companies
Mobile ontology exist and have already acquired and individuals and being widely accepted by the

53
Semantic Web in Ubiquitous Mobile Communications

Figure 2. Mobile ontology core visualization

masses. These schemata and ontologies can be only via XML.


specified in different knowledge representation
formalisms. We address external sources rep- Approach 1: RDF/S or
resented via the most popular formats, namely OWL Encoding
OWL, RDF/S and XML/S.
Relating and mapping these schemata to the The following principles are valid when consider-
Mobile ontology is mainly beneficial for interop- ing integration of Mobile ontology with ontologies
eration of the Mobile ontology community with of relevant topics expressed via RDF/S or OWL
other mobile communities. Thus: formalisms:

• Mobile ontology developers and users ben- • When necessary directly reusing the agreed
efit acquiring additional knowledge in the ontologies or their parts when modelling
mobile communication domain captured processes;
in the existing OWL, RDF, XML schemas • Establishing and using the library of map-
(i.e., reusing the present knowledge). pings of these ontologies with the “higher”
• Users of the related ontologies and schemas level Mobile ontology classes and prop-
benefit from a straightforward mapping of erties that have similar items as the used
their schemas to the Mobile ontology that external ontology. Such a mapping library
enables a simpler move to/involvement or would not be re-modelling, but stating rela-
extension of the Semantic technologies for tions between items in a machine readable
these communities. format. Equivalence, for example, can be
stated using constructions “owl:sameAs”
Technically, two different approaches to com- so that applications and enablers can “un-
bine the Mobile ontology with the existing com- derstand” that an item from the Mobile
mon ontologies and schemata will be considered, ontology and an “imported” agreed upon
depending on whether the data is encoded via an ontology are the same.
ontology language (such as PDF/S and OWL) or

54
Semantic Web in Ubiquitous Mobile Communications

The RDFS and OWL-based standard sche- So the ontology work with the existing XML
mata considered for this approach are listed in schemas would focus on ontologizing/consider-
Table 2. ing the knowledge present these schemas, and
combining it with the Mobile ontology, and not
Approach 2: XML Encoding extending these schemata.
The XML-based standard schemata considered
The following principles are valid when con- for this approach are listed in Table 3.The follow-
sidering integration of Mobile ontology with ing goals addressed by Mobile Ontology are open
schemata of relevant topics expressed via XML challenges for the current state of the art:
formalisms:
• The first comprehensive higher level on-
• Re-modelling XML schemata in OWL tology for mobile communication domain
and providing the new sub-ontologies that is constructed with involvement/sup-
as relatively independent ontology sub- port of major players in mobile communi-
modules under the umbrella of the Mobile cation area, i.e. the ontology (i) responds to
ontology; the needs of mobile service developers, (ii)
• Creation of the converters lifting up the in- is evolving, (iii) representatively captures
stance data represented solely in the XML the domain in an unbiased fashion.
format to RDF. • The most large scale international investi-
gation on the use of Semantic technology

Table 2. OWL and RDFS -based relevant standards

Ontology name Ontology Web address


UAProf http://www.openmobilealliance.org/release_program/uap_v2_0.html
FOAF http://www.foaf-project.org/
vCard http://www.w3.org/TR/vcard-rdf

Table 3. XML-based Relevant Standards

Schema name Schema Web address


Presence simple http://www.openmobilealliance.org/release_program/Presence_simple_v1_0.html
specification
Basic Presence http://www.ietf.org/rfc/rfc3863.txt
Data model
Generic Presence http://www.rfc-editor.org/rfc/rfc4479.txt
Data Model
Rich Presence In- http://www.rfc-editor.org/rfc/rfc4480.txt
formation
Location Types http://www.ietf.org/rfc/rfc4589.txt
Registry
A Presence-based http://www.ietf.org/rfc/rfc4119.txt
GEOPRIV Loca-
tion Object For-
mat

55
Semantic Web in Ubiquitous Mobile Communications

in mobile communication domain. knowledge on ontologies require at least


an introduction to the field.
Collaboration Aspects • Methodology: As of yet no widely ac-
cepted or best practice solutions on how
Mobile Ontology construction has been initially to acquire ontologies from people in such
implemented within two major deliverables of a setting.
SPICE EU project. 25 persons from 10 industry • Basic technology: Current ontology lan-
and research organizations stemming from 6 Eu- guage standards (such as OWL) cause con-
ropean countries have been initially involved in fusion and awkward modelling solutions.
this specific cross issue. Therefore, apart from the • Tool support:Better tools for ontology
definition of the up-to-date ontology infrastructure construction process coordination, docu-
for Mobile services, the results of this study include mentation would help to avoid ad-hoc so-
observation of collaboration aspects of developers lutions and manual work.
and users in the ontology construction.
Here we show how involved parties col- Ontology-Based Multimedia
laborated on the ontology construction, and what
personal involvement expectations for a larger Nowadays, network access technologies, such
scale ontology infrastructure would be. In Figure as Bluetooth, WiFi, WiMAX, are bringing the
3, the extent to which the developers have been dream of ubiquitous service provision closer
typically involved in initial definition of the ontol- to reality. Ubiquitous service provision is of
ogy is demonstrated. The figure shows that most interest to service providers, telecommunication
contributors tend to provide minor extensions to operators and technology manufactures for their
the ontology or choose a role of the user, which future revenue prospect. However, the barriers to
also confirms the previous research (Zhdanova, the delivery of ubiquitous services arise from the
2006).Summarising, the main challenges identi- desire to deliver a wide service mix to users hav-
fied in collaborative ontology construction as they ing a wide range of access devices via a multitude
have been observed are as follows: of heterogeneous access networks with different
preferences and likings. Most of the existing ap-
• Educational: People with no/little plications and services are created and provided

Figure 3. Collaboration patterns in mobile ontology construction

56
Semantic Web in Ubiquitous Mobile Communications

assuming a traditional pre-set delivery method and ment description besides their well-established
homogeneous transport media, which indicates comprehensiveness in describing the respective
a great importance and necessity for content and domains. MPEG-7 offers several tools, i.e. De-
service adaptation in order to deliver them in such scription Schemes (DS), to annotate multimedia
a ubiquitous environment. Such adaptation must content at different levels. These include Descrip-
be ‘context-aware’ and must facilitate user and tion Definition Language (DDL), Visual Schemes,
situation specific content and service provision. Audio Schemes and Multimedia Description
Therefore, the information from all the par- Schemes etc. MPEG-21 provides tools to describe
ties involved in the content and service delivery the environment to enable transparent multime-
chain will form a contextual knowledge base that dia creation, delivery and consumption between
is shared by different operational agents in order heterogeneous environments. The most relevant
to come up with a delivery or adaptation deci- part within MPEG-21 standard for the adapta-
sion autonomously. These agents are generally tion domain is Digital Item Adaptation (DIA). It
heterogeneous in nature and distributed across provides tools to describe the user environment
networks. How to describe and represent such a including: user characteristics, such as user info,
knowledge base is fundamental to the development preferences, usage history and physical charac-
of a content and service adaptation framework teristics, Device characteristics, such as display,
which supports ubiquitous service delivery. This memory and battery, Network characteristics,
work has formed part of the Ubiquitous Services such as error characteristics and bandwidth, and
Core Research Programme of the Virtual Centre Natural Environment characteristics such as noise
of Excellence in Mobile & Personal Communica- and illumination.
tions, Mobile VCE. Recent research efforts have reflected the
Ontology has been recognized as the knowl- recognition of using MPEG-7 and MPEG-21
edge representation scheme and OWL as the DIA, together with ontology-based technologies,
knowledge representation language. However, to to construct an ontology to support the content/
define a set of commonly-agreed vocabularies for service adaptation (Soetens et al., 2004; Jannach
the adaptation domain remains as a challenging et al., 2006). Though OWL has been chosen as
issue. The ubiquitous content/service adaptation the description language, Soetens et al. adopted
domain involves multiple sub-domains, i.e. the limited usage of MPEG-21 vocabularies due to the
user domain, the content/service domain and immaturity of this standard at the time of writing
the adaptation operation domain. Many efforts (Soetens et al., 2004). In (Jannach et al., 2006),
have been seen in recent years aiming to reach though MPEG vocabularies are adopted to form
a description standard including vocabularies the domain ontology, the representation remains
in order to achieve maximum acceptance and its original format of XML. With the actual adap-
interoperability among communities. So far, the tation operations being described in OWL-based
widely-acknowledged standards include usage language, this work realizes the integration of
environment description standards describing user the different representation formats on the tech-
information, device and network characteristics nical level using XML-based and logic-based
etc., such as CC/PP, UAProf and MPEG-21, and technologies. Therefore, although MPEG-7 and
content description standards such as MPEG-7. MPEG-21 standards have been acknowledged for
Among those, MPEG-7 (ISO/IEC JTC1/SC29/ their strengths in multimedia domain description
WG11 N3752, 2000) and MPEG-21 DIA (ISO/ and delivery, their strengths can be still greatly
IEC 21000-7, 2004) provide a good combination enhanced by adding machine-processable seman-
to linking content description with user environ- tics via ontology representation languages, such

57
Semantic Web in Ubiquitous Mobile Communications

as OWL and RDF(S). and use of ontology infrastructures that can serve
There exist several efforts to construct ontol- as a semantic basis for applications and enablers
ogy representations of MPEG-7 and MPEG-21 within the convergence of the mobile and the
(Hunter, 2001; Garcia, 2005). Those efforts con- Web worlds.
struct ontology automatically by means of XSLT In a nutshell, one may conclude that (i) there
transformation according to the rules specified exist a large number of ontologies addressing
in (Garcia, 2005). By automatically converting context-awareness and mobile communication
the XML tree structure, the obtained ontology issues, (ii) these ontologies are difficult to find
describes the relationship between the types of and they are not or weakly linked and connected
the tree element instead of describing the relation- to each other. Factually, most of the time they do
ships between the semantics embodied by the tree not form Semantic Web as the Web is about linking
elements. Although this approach expresses the the data (and the users thus obtaining the typical
XML-based standards in an OWL or RDF format, Semantic Web benefits, such as interoperability),
it does not add much semantic expressiveness to which is not the case for the current Semantic
them. Such approach would be applied in any Mobile Communications.
automatic XML schema to OWL conversion
regardless of the semantics of the respective
domain. In (Li et al., 2007), it argues that, for an FUTURE RESEARCH DIRECTIONS
expressive OWL representation of the XML-based
standards, manual conversion is necessary. The The questions for the further research and devel-
manual conversion may result in some XML ele- opment include: How to make these ontologies
ments being discarded or treated with another XML collaboratively constructed, linked to each other,
construct as one OWL concept as a consequence easily found, used and evolved? And more specific,
of its semantic interpretation. There are no rules follow-up questions thus are:
on how to manually convert an XML schema
description into OWL ontology. Different from 1. How to involve developers and users in
automatic conversion, which merely translates the community-driven ontology construction
XML syntax to OWL syntax, manual conversion and to what extent one should expect their
has to examine the elements and the attributes involvement?
of the XML schema, study their semantics, and 2. Which technical infrastructure is needed to
translate them into OWL constructs. implement this vision?

Ontologies are evolving as the domain is


CONCLUSION evolving and capturing the whole domain and
all the needs by a (small) group of ontology de-
State of the art and trends in convergence of velopers alone is ineffective (Zhdanova, 2006).
the Semantic Web and mobile communication The ontology infrastructure for ubiquitous mobile
fields are presented in this article. Knowledge communications should provide a user-friendly
representation formalisms, relevant research support to the ontology-based context-aware
fields, relevant ontologies are detailed, and the application and service developer with the fol-
challenges of Semantic technology application to lowing methods:
the mobile communications have been discussed.
State-of-the-art examples of the work in this area 1. Key-word based ontology search (e.g.,
have been outlined, including the development similar to OntoSelect28, Swoogle29)

58
Semantic Web in Ubiquitous Mobile Communications

2. Extraction and segmentation of the required Dean, M., & Schreiber, G. (Eds.). (2004 Febru-
ontology parts (e.g., operated on a level of ary). OWL Web Ontology Language Reference.
triples and eventually on demand arranged W3C Recommendation.
in a stand-alone schemata)
Fensel, D. (Ed.). (2001). Ontologies: A Silver
3. In case no existing relevant ontology parts
Bullet for Knowledge Management and Electronic
are found in the ontology platform infrastruc-
Commerce, Springer-Verlag.
ture, possibility to plug in freshly developed
ontologies and extensions Garcia, R., & Celma, O. (2005). Semantic inte-
4. Simple ontology instantiation, with a sub- gration and retrieval of multimedia metadata. In
sequent simple discovery and use of the Proceedings of the 5th International Workshop on
instance data Knowledge Mark-up and Semantic, Annotation at
5. Ontology matching and alignment to the ex- the 4th International Semantic Web Conference.
isting ontologies in case of duplicated model-
Goble, C. A., De Roure, D., Shadbolt, N., &
ing discovery, e.g., in a community-driven
Fernandes, A. A. A. (2004). Enhancing services
manner (Zhdanova & Shvaiko, 2006)
and applications with knowledge and semantics.
6. Search of relevant data within the instances
In Foster, I. and Kesselman, C. (Ed.). The Grid:
(e.g., employing technologies such as YARS
Blueprint for a New Computing Infrastructure
(Harth et al., 2006)
(2nd. Ed., Ch. 23, pp. 431-458). Morgan Kauf-
7. In case the developer made new ontology
mann.
design/extensions, allow him/her easily
plugging in the evolved versions of his/her Harth, A., Umbrich, J., & Decker, S. (2006). Mul-
ontologies/extensions, keeping up with the tiCrawler: A pipelined architecture for crawling
agreed ontology versioning practices and indexing Semantic Web data. In Proceedings
of the 5th International Semantic Web Conference,
In conclusion, the starting points for the Athens, GA, USA.
integration of the fields of Semantic Web and
Hercock, R. G. (2002). Autonomous Comput-
ubiquitous mobile communications exist both
ing. In Proceedings of the Workshop on Grand
on the ontology schemata level as well as on the
Challenges for Computing Research, Edinburgh,
tools level. However, practices and processes for
UK.
the common usage of the technologies originating
from these two fields are still to be acquired. It Hunter, J. (2001). Adding multimedia to the
is expected, that the new practices and processes semantic Web-building and MPEG-7 ontology.
are also to influence the future development of In Proceedings of the 1st International Semantic
applications, services and tools appearing in the Web Working Symposium.
unified field.
ISO/IEC 21000-7, (2004). Information Technol-
ogy — Multimedia Framework (MPEG-21) — Part
7: Digital Item Adaptation .
REFERENCES
ISO/IEC JTC1/SC29/WG11 N3752, (2000). Over-
Brickley, D., & Guha, R. V. (Eds.). (2004 Febru- view of the MPEG-7 Standard (version 4.0).
ary). RDF Vocabulary Description Language 1.0:
RDF Schema. W3C Recommendation.

59
Semantic Web in Ubiquitous Mobile Communications

Jannach, D., Leopold, K., Timmerer, C., & Hell- Tarkoma, S., Prehofer, C., Zhdanova, A. V., Moess-
wagner, H. (2006). A knowledge-based frame- ner, K., & Kovacs, E. (2007). SPICE: Evolving
work for multimedia adaptation. [Special Issue IMS to next generation service platforms. In
on Innovations in Applied Artificial Intelligence, Proceedings of the 3rd Workshop on Next Genera-
Springer. ]. The International Journal of Artificial tion Service Platforms for Future Mobile Systems
Intelligence, 24(2), 109–125. (SPMS 2007) at the 2007 International Symposium
on Applications and the Internet, IEEE Computer
Korpipää, P., Häkkilä, J., Kela, J., Ronkainen, S.,
Society Press.
& Känsälä, I. (2004). Utilising context ontology in
mobile device application personalization. ACM Weiser, M. (1991). The Computer for the Twenty-
International Conference Proceeding Series; Vol. First Century . Scientific American, 94–10.
83 Proceedings of the 3rd international confer-
Zhao, Y. X. (2004). Combining RDF and OWL
ence on Mobile and ubiquitous multimedia, pp.
with SOAP for Semantic Web Services. In Pro-
133-140, ACM Press.
ceedings of the 3rd annual Nordic Conference on
Lassila, O., & Swick, R. (1999). Resource De- Web Services (NCWS’04).
scription Framework (RDF) Model and Syntax
Zhdanova, A. V. (2006 January). An Approach
Specification. W3C Recommendation.
to Ontology Construction and its Application to
Li, N., Attou, A., & Moessner, K. (2007). A Community Portals. PhD thesis, University of
MPEG-based ontology for ubiquitous content/ Innsbruck, Austria.
service adaptation. In Proceedings of the 3rd
Zhdanova, A.V., Boussard, M., Cesar, P., Clavier,
Workshop on Context Awareness for Proactive
E., Gessler, S., Hesselman, C., Kernchen, R.,
Systems (CAPS’2007), Guildford, UK.
Le Berre, O., Melpignano, D., Nani, R., Patrini,
Luo, J., Montrose, B., Kim, A., Khashnobish, A., L., Räck, C., Strohbach, M., Sutterer, M., van
& Kang, M. (2006). Adding OWL-S support to the Kranenburg, H., Villalonga, C., Vitale, A. (2006).
existing UDDI infrastructure. In Proceedings of Ontology Definition for the DCS and DCS Re-
IEEE International Conference on Web Services source Description, User Rules, EU IST SPICE
(ICWS’06) pp. 153-162. IP deliverable (D3.1).
Manola, F., & Miller, E. (2004). RDF Primer. Zhdanova, A. V., & Shvaiko, P. (2006). Commu-
W3C Recommendation. nity-driven ontology matching. In Proceedings
of the Third European Semantic Web Conference,
Pfoser, D., Pitoura, E., & Tryfona, N. (2002).
11-14 June 2006, Budva, Montenegro, Springer-
Metadata modeling in a global computing envi-
Verlag, LNCS 4011, pp. 34-49.
ronment. GIS’02, November 8-9, 2002, McLean,
Virginia, USA, ACM Press (2002).
Soetens, P., Geyter, M. D., & Decneut, S. (2004).
ADDITIONAL READING
Multi-step media adaptation with semantic Web
services. In Proceedings of International Semantic Cho, M. J., Kim, H., Kim, S. H., Lee, H. J., Hong,
Web conference. S. C., & Jung, M. J. (2007). Context knowledge
modelling method for sharing and reusing con-
text knowledge in context-aware system, vol.
US2007038438, pp. 10.

60
Semantic Web in Ubiquitous Mobile Communications

De, S., & Moessner, K. (2007). Context gathering Reif, G., Gall, H., & Jazayeri, M. (2005). WEESA
in ubiquitous environments: enhanced service - Web engineering for Semantic Web applications.
discovery. In Proceedings of the 3rd Workshop In Proceedings of 14th International World Wide
on Context Awareness for Proactive Systems Web Conference, Chiba, Japan.
(CAPS’2007), Guildford, UK.
Smith, J. R., & Schirling, P. (2006). Metadata
Franz, B., Diego, C., Deborah, L. M., Daniele, N., standards roundup. IEEE MultiMedia, 13, 84–88.
& Peter, F. P.-S. (Eds.). (2003). The description doi:10.1109/MMUL.2006.34
logic handbook: theory, implementation, and ap-
Stamou, G., van Ossenbruggen, J., Pan, J. Z.,
plications, Cambridge University Press.
Schreiber, G., & Smith, J. R. (2006). Multimedia
Kobeissy, N., Genet, M. G., & Zeghlache, D. annotations on the semantic Web. IEEE MultiMe-
(2007). Mapping XML to OWL for seamless dia, 13, 86–90. doi:10.1109/MMUL.2006.15
information retrieval in context-aware environ-
Tsinaraki, C. (2007). Interoperability support
ments. Second IEEE International Workshop on
between MPEG-7/21 and OWL in DS-MIRF.
Services Integration in Pervasive Environments
Knowledge and Data Engineering . IEEE Trans-
(SEPS’07) at IEEE International Conference on
actions on, 19, 219–232.
Pervasive Services (ICPS’2007), pp. 349-354.
Villalonga, C., Strohbach, M., Snoeck, N., Sutterer,
Li, N., & Moessner, K. (2007). The MVCE
M., Belaunde, M., Kovacs, E., et al. (2007). Mo-
knowledge-based content and service adapta-
bile Ontology: Towards a standardized semantic
tion management framework. In Workshop on
model for the mobile domain. In Proceedings of
Applications and Services in Wireless Networks,
the 1st International Workshop on Telecom Service
Santander, Spain.
Oriented Architectures (TSOA 2007) at the 5th
Li, N., & Moessner, K. (2007). Design of content/ International Conference on Service-Oriented
service adaptation framework. Mobile VCE proj- Computing, Vienna, Austria.
ect Deliverable (D-U3.3).
Zhdanova, A. V., Du, Y., & Moessner, K. (2007).
Mukherjee, D., Delfosse, E., Jae-Gon, K., & Yong, Mobile experience enhancement by ontology-
W. (2005). Optimal adaptation decision-taking for enabled interoperation in a service platform. In
terminal and network quality-of-service. Multime- Proceedings of the 3rd Workshop on Context
dia . IEEE Transactions on, 7, 454–462. Awareness for Proactive Systems (CAPS’2007),
Guildford, UK, ISBN: 978-0-9556240-0-1.
Noy, N. F., & Musen, M. A. (1999). An algorithm
for merging and aligning ontologies: automation
and tool support. In Proceedings of the Work-
shop on Ontology Management at the Sixteenth ENDNOTES
National Conference on Artificial Intelligence
1
(AAAI-99). vCard: http://www.w3.org/TR/vcard-rdf
2
CC/PP: http://www.w3.org/Mobile/CCPP/
3
UAProf: http://www.openmobilealliance.
org/release_program/uap_v2_0.html

61
Semantic Web in Ubiquitous Mobile Communications

4 21
http://www.ruleml.org OPUCE: http://www.opuce.tid.es
5 22
http://www.w3.org/Submission/WRL/ Ambient Networks: http://www.ambient-
6
http://www.w3.org/Submission/SWRL/ networks.org
7 23
http://www.w3.org/2005/rules/wg TripCom: http://www.tripcom.org
8 24
http://www.w3.org/Submission/WSML/ SWING: http://www.swing-project.org
9 25
http://www.w3.org/Submission/SWSF- ASG: http://asg-platform.org
26
SWSL/ SmartWeb: http://www.smartweb-project.
10
http://www.w3.org/Submission/WSMO/ de
11 27
http://www.w3.org/Submission/SWSF- Mobile Ontology website: http://ontology.
SWSO/ ist-spice.org
12 28
http://www.w3.org/2002/ws/sawsdl/ OntoSelect: http://olp.dfki.de/ontoselect
13 29
http://www.w3.org/Submission/WSDL-S/ Swoogle: http://swoogle.umbc.edu
14
http://www.deri.org
15
http://www.w3.org/Submission/WSMX/
16
http://www.oasis-open.org/committees/
wss/
17
The GLOBUS Alliance: http://www.globus.
org
18
SPICE: http://www.ist-spice.org
19
MobileVCE: http://www.mobilevce.com
20
Mobilife: http://www.ist-mobilife.org

This work was previously published in The Semantic Web for Knowledge and Data Management: Technologies and Prac-
tices, edited by Z. Ma; H. Wang, pp. 266-287, copyright 2009 by Information Science Reference (an imprint of IGI Global).

62
63

Chapter 1.5
A Review of Fuzzy Models
for the Semantic Web
Hailong Wang
Northeastern University, China

Zongmin Ma
Northeastern University, China

Li Yan
Northeastern University, China

Jingwei Cheng
Northeastern University, China

Abstract the existing proposals for extending the theoretical


counterpart of the semantic web languages, descrip-
In the Semantic Web context, information would be tion logics (DLs), and the languages themselves.
retrieved, processed, shared, reused and aligned in The following statements will include the expressive
the maximum automatic way possible. Our experi- power of the fuzzy DLs formalism and its syntax
ence with such applications in the Semantic Web and semantic, knowledge base, the decidability
has shown that these are rarely a matter of true or of the tableaux algorithm and its computational
false but rather procedures that require degrees of complexity etc. Also the fuzzy extension to OWL
relatedness, similarity, or ranking. Apart from the is discussed in this chapter.
wealth of applications that are inherently imprecise,
information itself is many times imprecise or vague.
In order to be able to represent and reason with such INTRODUCTION
type of information in the Semantic Web, different
general approaches for extending semantic web lan- The Semantic Web is an extension of the current
guages with the ability to represent imprecision and web in which the web information can be given
uncertainty has been explored. In this chapter, we well-defined semantic meaning, and thus enabling
focus our attention on fuzzy extension approaches better cooperation between computers and people.
which are based on fuzzy set theory. We review From this point of view, we should find some
methods which can describe the semantic mean-
DOI: 10.4018/978-1-60566-028-8.ch002

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
A Review of Fuzzy Models for the Semantic Web

ing of the web. Fortunately, “ontology” can do Bayesian probability as well as ad hoc heuristics
this. The core of the Semantic Web is “ontology” into ontology definition and reasoning.
which refers to a set of vocabulary to describe the In this paper, we review existing proposals to
conceptualization of a particular domain. Over extend semantic web languages with the capa-
the past few years, several ontology definition bility to handle uncertain information to better
languages for the Semantic Web have emerged, deal with the situations mentioned above. There
including RDF(S), OIL, DAML, DAML+OIL, are many ways of representing and dealing with
and OWL. Among them, OWL is the newly re- uncertainty. In this paper, we restrict our attention
leased standard recommended by W3C. As the to approaches that use fuzzy methods for repre-
Semantic Web expects, OWL has the reasoning senting uncertain information. In particular, we
nature because description logics (DLs) (Baader, will not cover recent proposals for probabilistic
2003) are essentially the theoretical counterpart extensions of semantic web languages. We will
of OWL and play a crucial role in this context. also not discuss non-monotonic and non-standard
DLs provide a logical reconstruction of object- logics for representing uncertainty. As described
centric and frame-based knowledge representation above, existing Semantic Web languages are
languages. It is a subset of first-order logic that mainly based on logic and do not support represent-
provides sound and decidable reasoning support ing imprecise and uncertain information. In this
(Baader, 2003). paper, we therefore review a number of proposals
It is clear that DLs play a key role in the Se- for extending logical languages with fuzzy exten-
mantic Web. As with traditional crisp logic, any sions in more details. We focused on:
sentence in OWL, being asserted facts, domain
knowledge, or reasoning results, must be either 1. Approaches that extend description logics
true or false and nothing in between. However, which play as the theoretical counterpart of
most real world domains contain uncertainty the semantic web languages.
knowledge and incomplete or imprecise informa- 2. Approaches that directly extend semantic
tion that is true only to a certain degree. Ontologies web languages, in particular OWL.
defined by these languages thus cannot quantify
the degree of the overlap or inclusion between In the first category, we cover fuzzy exten-
two concepts, and cannot support reasoning in sions of description logics which are commonly
which only partial information about a concept accepted as being the formal basis of OWL. Even
or individual in the domain can be obtained. though most approaches only cover logics that are
Uncertainty becomes more prevalent when more much weaker than OWL, the methods proposed
than on ontologies are involved where it is often can directly be applied to the corresponding subset
the case that a concept defined in on ontology can of OWL without changes because the description
only find partial matches to one or more concepts logics play as the theory counterpart of the OWL.
in another ontology. To overcome the difficulty When talking about the different approaches,
arising from the crisp logics, existing ontology we will survey them according to the expressive
languages need to be extended to be able to capture power from weaker to stronger. And in the fol-
uncertainty knowledge about the concepts, prop- lowing survey, we should discuss the following
erties and instances in the domain and to support issues of the different approaches:
reasoning with partial, imprecise information.
Along this direction, researchers in the past have • Expressiveness of the logical language
attempted to apply different formalisms such as • The syntax and semantic of the fuzzy ex-
Fuzzy logic (Zadeh, 1965), Rough set theory and tension to description logics

64
A Review of Fuzzy Models for the Semantic Web

• The components of the knowledge base tics to capture the meaning of the most popular
• Tableaux algorithm for the description features of structured representation of knowledge.
logics Nowadays, DLs have gained even more popular-
• The decidability and complexity of the tab- ity due to their application in the context of the
leaux algorithm Semantic Web (Berners-Lee, 2001).The recent
research about description logics can be divided
Indeed, the balance of expressive power into three categories:
and the computability of the fuzzy extension of
description logics is a hot topic of the research. • Introducing the theoretical foundations of
Generally speaking, the more expressive of the description logics, addressing some of the
description logic, the higher computational com- most recent developments in theoretical re-
plexity of it, so we should consider the balance search in the area;
of the two factors in a real application. At last, in • Focusing on the implementation of knowl-
the latter category, we also review a number of edge representation systems based on
proposals for extending the ontology description Descriptions Logics, describing the basic
language OWL. functionality of a DL system, surveying
The paper is structured as follows. We first the most influential knowledge represen-
present some background information on semantic tation systems based on descriptions, and
web languages and related formalisms that are the addressing specialized implementation
basis for the logical languages used in the different techniques;
approaches discussed later in the paper. We also • Addressing the use of description logics
provide a brief introduction to fuzzy set theory and of DL-based systems in the design of
which the fuzzy description logics are based on. several applications of practical interest.
In the mainly part of this chapter, we survey the
different approaches to extend the description In the following statements, we mainly focus
logics to represent the imprecise and uncertainty on the first category, especially the theoretical
information according their expressive power. formalism of the description logics, with respect
We also discuss proposals for fuzzy languages to the balance between its expressive power and
OWL for the semantic web. Finally, we conclude its computational complexity. Indeed, subsequent
with a critical review of the state of the art and an results on the tradeoff between the expressiveness
analysis of directions for future research. of a DL language and the complexity of reasoning
with it, and more generally, the identification of
the sources of complexity in DL systems, showed
PRELIMINARIES AND that a careful selection of language constructs was
BACKGROUND needed and that the reasoning services provided
by the system are deeply influenced by the set
Description Logics of constructs provided to the user. We can thus
characterize three different approaches to the
In the last decade a substantial amount of work implementation of reasoning services. The first can
has been carried out in the context of Description be referred to as limited + complete, and includes
Logics. DLs are a logical reconstruction of the systems that are designed by restricting the set of
so-called frame-based knowledge representation constructs in such a way that subsumption would be
languages, with the aim of providing a simple computed efficiently, possibly in polynomial time.
well-established Tarski-style declarative seman- The second approach can be denoted as expressive

65
A Review of Fuzzy Models for the Semantic Web

+ incomplete, since the idea is to provide both an A | (atomic concept)


expressive language and efficient reasoning. The
drawback is, however, that reasoning algorithms C  D | (intersection)
turn out to be incomplete in these systems. After
some of the sources of incompleteness were dis- ∀R . C (value restriction)
covered, often by identifying the constructs—or,
more precisely, combinations of constructs—that ∃R . ⊥ | (limited existential quantification)
would require an exponential algorithm to preserve
the completeness of reasoning, systems with In AL, negation can only be applied to atomic
complete reasoning algorithms were designed. concepts, and only the top concept is allowed in
Systems of this sort are therefore characterized as the scope of an existential quantification over
expressive + complete; they were not as efficient a role. In order to define a formal semantics of
as those following the other approaches, but they AL-concepts, we consider interpretations I that
provided a test bed for the implementation of consist of a non-empty set ΔI (the domain of the
reasoning techniques developed in the theoretical interpretation) and an interpretation function •I,
investigations, and they played an important role which assigns to every atomic concept A a set AI
in stimulating comparison and benchmarking with ⊆ΔI and to every atomic role R a binary relation
other systems. RI ⊆ΔI×ΔI . The interpretation function is extended
Now, we survey the languages of the descrip- to concept descriptions by the following inductive
tion logics according to their expressive power definitions:
with the beginning of AL. Elementary descrip-
tions are atomic concepts and atomic roles (also I = ΔI
called concept names and role names). Complex
descriptions can be built from them inductively ⊥I = Φ
with concept constructors and role constructors.
(¬A)I = ΔI \ AI
In abstract notation, we use the letters A and B
for atomic concepts, the letter R for atomic roles,
(C  D)I= CI ∩DI
and the letters C and D for concept descriptions.
Description languages are distinguished by the
(∀R .C)I = {a ∈ ΔI | ∀b. (a, b) ∈ RI →b ∈ C I}
constructors they provide. In the sequel we shall
discuss various languages from the family of
(∃R .⊥)I = {a ∈ ΔI | ∃b. (a, b) ∈ RI}
AL-languages. The language AL (= attributive lan-
guage) has been introduced as a minimal language
We say that two concepts C, D are equivalent,
that is of practical interest. The other languages
and write C ≡ D, if CI =DI for all interpretations
of this family are extensions of AL.
I.

The Basic Description Logic AL The Family of AL-languages


Concept descriptions in AL are formed according
We obtain more expressive languages if we add
to the following syntax rule:
further constructors to AL. The union of concepts
(indicated by the letter U) is written as C D, and
C, D::= | (universal concept)
interpreted as
⊥| (bottom concept)

66
A Review of Fuzzy Models for the Semantic Web

(C D)I = CI ∪ DI: where a letter in the name stands for the presence
of the corresponding constructor. For instance,
ALEN is the extension of AL by full existential
Full existential quantification (indicated by the quantification and number restrictions.
letter E) is written as $R .C, and interpreted as
The More Expressive
(∃R .C)I = {a ∈ ΔI | ∃b. (a, b) ∈ RI ∧ b ∈ C I} Description Logics

Note that ∃R .C differs from ∃R . in that arbi- There are several possibilities for extending AL in
trary concepts are allowed to occur in the scope order to obtain a more expressive DL. The three
of the existential quantifier. most prominent are adding additional concept
Number restrictions (indicated by the letter constructors, adding role constructors, and formu-
N) are written as ≥nR (at-least restriction) and as lating restrictions on role interpretations. Below,
≤nR (at-most restriction), where n ranges over the we start with the third possibility, since we need
nonnegative integers. They are interpreted as to refer to restrictions on roles when defining
certain concept constructors. For these extensions,
(≥nR)I = {a ∈ ΔI | |{b | (a, b) ∈ RI }| ≥ n} we also introduce a naming scheme. Basically,
each extension is assigned a letter or symbol.
For concept constructors, the letters/symbols are
and written after the starting AL, for role constructors,
we write the letters/symbols as superscripts, and
(≤nR)I = {a ∈ ΔI | |{b | (a, b) ∈ RI }| ≤ n} for restrictions on the interpretation of roles as
subscripts. As an example, the DL ALCQ-1R+, ex-
tends AL with the concept constructors negation
respectively, where“|•|”denotes the cardinality (C) and qualified number restrictions (Q), the role
of a set. From a semantic view point, the coding constructor inverse (-1), and the restriction that
of numbers in number restrictions is immaterial. some roles are transitive(R+).
However, for the complexity analysis of inferences Restrictions on role interpretations These
it can matter whether a number n is represented restrictions enforce the interpretations of roles
in binary (or decimal) notation or by a string of to satisfy certain properties, such as functionality
length n, since binary (decimal) notation allows and transitivity. We consider these two prominent
for a more compact representation. examples in more detail. Others would be sym-
The negation of arbitrary concepts (indicated metry or connections between different roles.
by the letter C, for “complement”) is written as
¬C, and interpreted as (i) Functional roles. Here one considers a subset
NF of the set of role names NR, whose ele-
(¬C)I = ΔI \ CI ments are called features. An interpretation
must map features f to functional binary rela-
Extending AL by any subset of the above con- tions f I ⊆ΔI×ΔI. AL extended with features
structors yields a particular AL-language. We name is denoted by ALf.
each AL-language by a string of the form (ii) Transitive roles. Here one considers a subset
NR+ of NR. Role names R∈NR+ are called
AL[U][E][N][C]; transitive roles. An interpretation must map
transitive roles R∈NR+ to transitive binary

67
A Review of Fuzzy Models for the Semantic Web

relations R I ⊆ΔI×ΔI. AL extended with existential predicate restrictions according to their


transitive roles is denoted by ALR+. semantics. The main new feature is that, in ad-
dition to the usual “abstract” clashes, there may
All the DLs mentioned until now contain be concrete ones, i.e., one must test whether the
the concept constructors intersection and value given combination of concrete predicate assertions
restriction as a common core. DLs that allow for is non-contradictory. This is the reason why we
intersection of concepts and existential quantifi- must require that the satisfiability problem for D
cation (but not value restriction) are collected in is decidable. As described in (Baader and Han-
the EL-family. The only constructors available schke, 1991), the algorithm is not in PSpace. Using
in EL are intersection of concepts and existen- techniques similar to the ones employed for ALC
tial quantification. Extensions of EL are again it can be shown, however, that the algorithm can
obtained by adding appropriate letters/symbols. be modified such that it needs only polynomial
In order to avoid very long names for expres- space (Lutz, 1999), provided that the satisfiability
sive DLs, the abbreviation S was introduced for procedure for D is in PSpace. In the presence of
ALCR+, i.e., the DL that extends ALC by transitive acyclic TBoxes, reasoning in ALC(D) may become
roles. Prominent members of the S-family are SIN NExpTime-hard even for rather simple concrete
(which extends ALCR+ with number restrictions domains with a polynomial satisfiability problem
and inverse roles), SHIF (which extends ALCR+ (Lutz, 2001).
with role hierarchies, inverse roles, and number The more expressive description logics
restrictions of the form ≤1R), and SHIQ (which SHOQ(D) which can represent data informa-
extends ALCR+ with role hierarchies, inverse roles, tion is proposed in (Horrocks, 2001). Although
and qualified number restrictions). Actually, the SHOQ(D) is rather expressive, it has a very serious
DLs SIN, SHIF, and SHIQ are somewhat less limitation on data types; i.e., it does not support
expressive than indicated by their name since the customised data types. It has been pointed out that
use of roles in number restrictions is restricted: many potential users will not adopt it unless this
roles that have a transitive sub-role must not occur limitation is overcome. Pan and Horrocks release
in number restrictions. a series of papers about data types to solve the
problem, in (Pan & Horrocks, 2006; Pan, 2007).
Description Logics with Data In the two papers, they summarize the limitations
Type Representation of OWL datatyping and propose the data type ap-
proach. For example, the SHIQ(G) and SHOQ(G)
A drawback that all DLs introduced until now share DLs presented in (Pan & Horrocks, 2006; Pan,
is that all the knowledge must be represented on 2007) can support user-defined data type and
the abstract logical level. In many applications, user-defined data type predicates.
one would like to be able to refer to concrete do-
mains and predefined predicates on these domains Fuzzy Set Theory
when defining concepts. To solve the problem,
Baader and Hanschke prompt two extensions Fuzzy data is originally described as fuzzy set
(Hanschke, 1992; Haarslev et al., 1999). In the (Zadeh, 1965). Let U be a universe of discourse,
two papers, the definition of concrete domain is then a fuzzy value on U is characterized by a fuzzy
given and a tableau-based algorithm for deciding set F in U. A membership function μF: U → [0,
consistency of ALC(D)-ABoxes for admissible D 1] is defined for the fuzzy set F, where μF (u), for
was introduced in (Baader & Hanschke, 1991). each u ∈ U, denotes the degree of membership
The algorithm has an additional rule that treats of u in the fuzzy set F. Thus the fuzzy set F is

68
A Review of Fuzzy Models for the Semantic Web

described as follows. operations used to interpret the concept construc-


tors in all these approaches were the same ones as
F = {μF (u1)/u1, μF (u2)/u2, ..., μF (un)/un} in our context. (Tresp & Molitor, 1998) contains
complete algorithms for solving these inference
When the μF (u) above is explained to be a problems in the respective fuzzy extension of
measure of the possibility that a variable X has ALC. Although both algorithms are extensions of
the value u in this approach, where X takes values the usual tableau-based algorithm for ALC, they
in U, a fuzzy value is described by a possibility differ considerably. For example, the algorithm
distribution πX. Let πX and F be the possibil- in (Tresp & Molitor, 1998) introduces numerical
ity distribution representation and the fuzzy set variables for the degrees, and produces a linear
representation for a fuzzy value, respectively. optimization problem, which must be solved in
In the fuzzy set theory, each object ui ∈ U is as- place of the usual clash test. In contrast, (Straccia,
signed a single value between 0 and 1, called the 2001) deals with the membership degrees within
degree of membership, where U is a universe of his tableau-based algorithm.
discourse. The fuzzy description logic FALCDefinition.
Fuzzy set theory is the theory basis for fuzzy Let NI, NC and NR be three disjoint sets: NI is a set
extensions to description logics to represent im- of individual names, NC is a set of fuzzy concept
precise and uncertain information. names and NR is a set of fuzzy role names. Fuzzy
ALC-concepts are defined as

FUZZY EXTENSIONS OF C, D::= ⊥|Τ|A|¬C|C∪D|C∩D|∃R.C|∀R.C ;


SEMANTIC WEB LANGUAGES
Here A∈NC, R∈NR. Fuzzy ALC semantics is
Extensions of Description Logics defined by a fuzzy interpretation I = <ΔI, ·I>, Here
ΔI is a nonempty set and ·I is an function which
Much work has been carried out towards combin- maps every a∈NI to an element aI∈ΔI, maps
ing fuzzy logic and description logics during the every A ∈NC into a function A I: ΔI →[0, 1], and
last decade. The initial idea was presented by Yen maps every R∈NR into a function RI: ΔI×ΔI →[0,
in (Yen, 1991), where a structural subsumption 1].Furthermore, for any fuzzy ALC-concepts C
algorithm was provided in order to perform rea- and D, R ∈ NR and x ∈ ΔI, we have:
soning. The following statements will illustrate
all the fuzzy extensions to DLs from weaker to ΤI(x) = 1;
stronger in expressive power.
⊥I(x) = 0;
The Family of FALC Languages
(¬C)I(x) = 1 - CI(x);
ALC is the basic format of the description logics.
Reasoning in fuzzy ALC was latter presented in (C∩D)I(x) = CI(x) ∧ DI(x);
(Straccia, 2001), as well as in other approaches
(Straccia, 1998), where an additional concept (C∪D)I(x) = CI(x) ∨ DI(x));
constructor, called membership manipulator was
included in the extended language. In all these (∃R.C)I(x) = sup y∈ΔI {min (RI(x, y), CI(y))};
approaches tableaux decision procedures were
(∀R.C)I(x) = inf y∈ΔI {max (1 - RI(x, y), CI(y))};
presented for performing reasoning services. The

69
A Review of Fuzzy Models for the Semantic Web

With the introduction of the fuzzy sets into use fuzzy logic for defining the semantics. Thus, an
the classical ALC, the form of the knowledge interpretation now assigns fuzzy sets to concepts
base is changed accordingly:Definition. A fuzzy and roles, i.e., concept names A are interpreted
ALC knowledge base is composed of a TBox and by membership degree functions of the form AI:
an ABox: ΔI →[0, 1], and role names R by membership
degree functions of the form RI: ΔI×ΔI→[0, 1].
• A TBox is a finite set of terminology axioms The interpretation of the Boolean operators and
of the form C ⊆ D. Any interpretation I satis- the quantifiers must then be extended from {0,
fies C ⊆ D iff for any x∈∆I, CI(x) ≤DI(x). I is 1} to the interval [0, 1].
a model of TBox T iff I satisfies all axioms Tresp & Molitor (1998) also propose an ex-
in T. tension of the syntax by so-called manipulators,
• An ABox is a finite set of assertions of the which are unary operators that can be applied to
form < α⋈n >, Here ⋈∈{>, ≥, <, ≤}, n∈ concepts. Examples of manipulators could be
[0, 1], α (called a fuzzy assertion) is ei- “mostly”, “more or less”, or “very”. For example,
ther of the form a: C or (a, b): R(a,b∈NI if Tall is a concept (standing for the fuzzy set of
). Especially, in order to giving a uniform all tall persons), then VeryTall, which is obtained
format of the ABox, we define: when n =1, by applying the manipulator Very to the concept
the form <α≥1> is equivalent to <α=1>. Tall, is a new concept (standing for the fuzzy set of
Concretely speaking, <a: C≥1>means that all very tall persons). Intuitively, the manipulators
a is determinately an individual of C; <(a, modify the membership degree functions of the
b): R ≥1> means that (a, b) determinately concepts they are applied to appropriately. In our
has the relationship R. Any interpretation I example, the membership function for VeryTall
satisfies <a:C⋈n> iff CI (aI)⋈n and satis- should have its largest values at larger heights
fies <(a, b): R⋈n > iff RI (aI, bI)⋈n. Then than the membership function for Tall. Formally,
I is a model of ABox A iff I satisfies all the semantics of manipulators is defined by a
assertions in A. function that maps membership degree functions
to membership degree functions. The manipula-
Reasoning algorithms for FALC and their tors considered in (Tresp & Molitor, 1998) are,
proofs can be found in (Straccia, 2001). It can be however, of a very restricted form. Lets us now
seen from syntax and semantics presented above consider what kind of inference problems are of
that the entailment and subsumption relationships interest in this context. (Yen, 1991) considers
may hold to some degree in the interval [0, 1]. crisp subsumption of fuzzy concepts, i.e., given
Complete algorithms for reasoning in FALC have two concepts C, D defined in the fuzzy DL, he is
been presented, that is, we have devised algorithms interested in the question whether CI(d) ≤ DI(d)
for solving the entailment problem, the subsump- for all fuzzy interpretations I and d∈ΔI. Thus, the
tion problem as well as the best truth-value bound subsumption relationship itself is not fuzzified. He
problem. The complexity result shows that the describes a structural subsumption algorithm for a
additional expressive power has no impact from rather small fuzzy DL, which is almost identical to
a computational complexity point of view. the subsumption algorithm for the corresponding
The fuzzy description logic ALCFM The DL classical DL. In contrast, (Tresp & Molitor, 1998)
language used was a sub-language of the basic DL are interested in determining fuzzy subsumption
ALC. The main idea underlying the fuzzy exten- between fuzzy concepts, i.e., given concepts C, D,
sions of description logics proposed in (Tresp & they want to know to which degree C is a subset
Molitor, 1998) is to leave the syntax as it is, but to of D. In (Tresp & Molitor, 1998), also ABoxes

70
A Review of Fuzzy Models for the Semantic Web

are considered, where the ABox assertions are specialisation of the general form C⊆D and in the
equipped with a degree. case cycles are allowed in a fuzzy KB. Another
Fuzzy description logics with hedges and interesting topic for further research concerns the
modifiers Reasoning in fuzzy ALC was also semantics of fuzzy connectives. While for a huge
presented in other approaches (Hölldobler, 2002; number of proposals given in the literature their
2003; 2004; 2005; 2006), where an additional impact from a semantics point of view is well
concept constructor, called membership manipula- understood, the question how they impact from
tor was included in the extended language. In all a computational complexity and algorithms point
these approaches tableaux decision procedures of view remains still open.
were presented for performing reasoning services. According to the above statements, the family
(Hölldobler, 2002) presents a fuzzy description of fuzzy ALC description logics can be summaried
logic ALCFH, where primitive concepts are modi- in Table 1.
fied by means of hedges. ALCFH is strictly more
expressive than Fuzzy ALC defined in (Straccia, The More Expressive Fuzzy
2001). The paper shows that given a linearly Description Logics
ordered set of hedges primitive concepts can be
modified to any desired degree by prefixing them However, FALC offers limited expressive power
with appropriate chains of hedges. Furthermore, it of complex fuzzy information. Some discus-
defines a decision procedure for the unsatisfiability sions about reducing FALC into classical ALC
problem in ALCFH, and discusses truth bounds, and providing a tableau for FALC with General
expressivity as well as complexity issues. Concept Inclusions (GCIs) were given in (Strac-
Strictly speaking, the language defined by cia, 2004a) and (Stoilos, 2006c), respectively. In
(Tresp & Molitor, 1998) is more expressive, as (Stoilos, 2006c), fuzzy description logics have
we do not consider concept modifiers. From a been proposed as a language to describe struc-
semantics point of view, the extension to Tresp tured knowledge with vague concepts. A major
and Molitor’s language is quite straightforward. theoretical and computational limitation so far is
But, the cost that we have to pay for this increasing the inability to deal with General Concept Inclu-
expressive power is that, from a computational sions (GCIs), which is an important feature of
complexity and algorithms point of view, things classical DLs. It addresses this issue and develops
changes radically. The fuzzy extension to ALC can a calculus for fuzzy DLs with GCIs. (Meghini et
be used as a basis both for extending existing DL al. 1998) proposed a preliminary fuzzy DL, which
based systems and for further research. In this latter lacks reasoning algorithm, as a modeling tool for
case, there are several open points. For instance, it multimedia document retrieval.
is not clear yet how to reason both in case of fuzzy In the following statements, we survey the dif-

Table 1. The family of FALC languages

Fuzzy description logics Corresponding references Mainly discussed issues


FALC (Straccia, 1998; 2001) Syntax,semantics, properties and reasoning
services
ALCFM (Tresp and Molitor, 1998) Syntax,semantics and a mothod for computing the
degree of subsumption
ALCFH (Hölldobler, 2002; 2003; 2004; 2005; 2006) Syntax,semantics and reasoning services
ALCFL (Dinh-Khac, 2006) Syntax,semantics and reasoning services

71
A Review of Fuzzy Models for the Semantic Web

ferent formalisms of the description logics accord- and reason with vague and imprecise knowledge.
ing to the concept constructors or role restrictions Both these properties fit well into the framework
that have been constraint on them. of knowledge based multimedia processing where
Fuzzy description logics with number re- both part-whole relationships, as well as, imprecise
strictions Approaches towards more expressive and vague knowledge appear in applications like
DLs, are presented in (Sánchez, D, 2004), where multimedia information retrieval and processing.
the DL is ALCQF+. It includes fuzzy quantifiers, In the latter of the paper, a tableau algorithm for
which is a new novel idea for fuzzy DLs. Un- checking the consistency of ABox is given and
fortunately, in the approach only the semantics has been proved that it will terminates. (Stoilos,
of the extended languages are provided and no 2005b) provides an extension of the above fKD-SI
reasoning algorithms. But in (Sánchez, D, 2006), to an even more expressive DL, namely fKD-SHIN.
it introduces reasoning procedures for ALCQF+, The paper has presented an extension of the very
the fuzzy description logic with extended qualified expressive description logic SHIN with fuzzy set
quantification. The language allows for the defini- theory. It shows the semantics as well as detailed
tion of fuzzy quantifiers of the absolute and relative reasoning algorithms for the extend languages. A
kind by means of piecewise linear functions on N fuzzy tableau for fKD-SHIN ABoxes is shown and
and Q∩[0, 1] respectively. In order to reason about it proves the flowing lemma.Lemma:Let A be an
instances, the semantics of quantified expressions fKD-SHIN ABox and R a fuzzy RBox. Then
is defined based on recently developed measures
of the cardinality of fuzzy sets. A procedure is i. when started for A and R the tableaux algo-
described to calculate the fuzzy satisfiability rithm terminates
of a fuzzy assertion, which is a very important ii. A has a fuzzy tableau w.r.t. R if and only
reasoning task. The procedure considers several if the expansion rules can be applied to A
different cases and provides direct solutions for and R such that they yield a complete and
the most frequent types of fuzzy assertions. In ad- clash-free completion forest.
dition, (Sánchez, D, 2006) defined independence
of fuzzy assertions and obtained some results that In (Stoilos, 2006b), the syntax and semantics
speed up the calculation of fuzzy satisfiability in of fuzzy SHOIQ were presented and the properties
some (the most common) cases. of the semantics of transitivity, qualified cardinal-
The series of F-SHOIQ Because today quite ity restrictions and reasoning capabilities were
a lot of multimedia systems and applications use investigated. (Stoilos, 2006b) extends the current
knowledge representation formalisms to encode state-of-the-art on fuzzy extensions to Semantic
and reason with knowledge that exists within the Web languages by presenting the syntax and
multimedia documents, (Stoilos, 2005a) presents a semantics of the fuzzy-SROIQ DL as well as the
more expressive fuzzy DL fKD-SI. The goal of this abstract, XML syntax and semantics of a fuzzy
direction is to narrow the semantic gab between extension to OWL 1.1. Moreover, it provides rea-
the content of a multimedia object, as perceived soning support for a fuzzy version of fuzzy-SROIQ
by a human being, and as “viewed” by an infor- by extending well-known reduction techniques
mation system. (Stoilos, 2005a) has extended of fuzzy DLs to classical DLs for the additional
the DL language SI with fuzzy set theory. The axioms and constructors of fuzzy-SROIQ.
combination of transitive and inverse roles can Fuzzy description logics with concrete do-
capture knowledge about part-whole relationships main The Semantic Web is expected to process
and aggregated objects. Furthermore, the incor- knowledge information and data information
poration of fuzziness allows the users to encode in an intelligent and automatic way. But recent

72
A Review of Fuzzy Models for the Semantic Web

research has shown that the OWL DL ontology which means the relationship of data types v1,…,vn
language is very limited in representing data infor- satisfies predicate p in a degree in [0,1]. Based on
mation. Furthermore, the OWL DL can’t process the definition of the fuzzy concrete domain, the
imprecision and uncertainty which widely exists fuzzy data information in the semantic web can
in human knowledge and natural language. In be represented and reasoning.
(Straccia, 2004b), a fuzzy extension of ALC(D) According to the above statements, some more
(the ALC extended with concrete domains) was expressive fuzzy description logics can be sum-
presented. The paper presents a fuzzy description maried in Table 2.
logic where the representation of concept mem-
bership functions and fuzzy modifiers is allowed, Extensions of OWL
together with a inference procedure based on a
mixture of a tableaux and bounded mixed integer More recently, little work has been carried out
programming. towards combining fuzzy logic and the Semantic
The more expressive DLs which can support Web ontology. In (Stoilos, 2005c), the OWL web
fuzzy concrete domains is shown in (Straccia, ontology language was extended with fuzzy set
2005), where the language is SHOIN(D+). It is the theory, which is called f-OWL, in order to capture,
corresponding Description Logic of the ontology represent and reason with imprecise information.
description language OWL DL. It shows that the Based on a fuzzy extension to OWL called Fuzzy
representation and reasoning capabilities of fuzzy OWL, (Stoilos, 2006a) developed a reasoning plat-
SHOIN(D) go clearly beyond classical SHOIN(D). form, Fuzzy Reasoning Engine (FiRE), which lets
Interesting features are: (i) concept constructors Fuzzy OWL capture and reason about imprecise
are based on t-norm, t-conorm, negation and and uncertain knowledge. Several connections
implication; (ii) concrete domains are fuzzy sets; between Fuzzy Logic, the Semantic Web, and its
(iii) fuzzy modifiers are allowed; and (iv) entail- components were presented in (Stoilos, 2005c).
ment and subsumption relationships may hold to For most recent research issues about fuzzy
some degree in the unit interval [0, 1]. The fuzzy logic, and more generally soft computing, in the
concrete domain is defined as following: (ΔD, •D) description logics, ontologies and the Semantic
is an interpretation. •D is an interpretation which Web, ones can refer to (Ma, 2006), (Sanchez,
assigns each concrete individual to an element in 2006a; 2006b).
ΔD; assigns each simple data type role T ∈ RD to
a function TI : ΔI ×ΔD →[0, 1]; assigns each n-ary
predicate p to the fuzzy relation pD: ΔDn → [0,1]

Table 2. Some important fuzzy description logics

Representation of fuzzy terminologies and concepts Representation of fuzzy data information


f-TSL f-ALCQF+ f-SI f-SHIN f-SHOIQ f-ALC(D) f-SHOIQ (D) f-SHOIQ
(G)
Syntax & semantic (Yen, (Sánchez, (Stoilos, (Stoilos, (Stoilos, (Straccia, (Straccia,
2001) 2004,2006) 2005a) 2005b) 2006b) 2004b) 2005)
Tableau algorithm (Yen, (Sánchez, (Stoilos, (Stoilos, (Stoilos, (Straccia,
2001) 2004,2006) 2005a) 2005b) 2006b) 2004b)
Decidability (Yen, (Sánchez, (Stoilos, (Stoilos, (Stoilos,
2001) 2006) 2005a) 2005b) 2006b)

73
A Review of Fuzzy Models for the Semantic Web

DISCUSSION AND CONCLUSIONS can process fuzzy data information should be de-
veloped. Especially, we should pay more attention
In this chapter, we focus our attention on the on the fuzzy extension of user-defined data type
recent research achievements on fuzzy extension and user-defined data type predicates.
approaches which are based on fuzzy set theory. The computational complexity of the fuzzy
We survey existing proposals for extending the tableau algorithm and the optimal technologies
theoretical counterpart of the semantic web to the fuzzy tableau algorithm. After prompting a
languages, description logics (DLs), and the lan- fuzzy tableau algorithm to an expressive DL lan-
guages themselves. The above statements include guage, although the decidability of the algorithm
the expressive power of the fuzzy DLs formalism is investigated, the computational complexity is
and its syntax and semantic, knowledge base, still an open problem. Furthermore, the optimal
the decidability of the tableaux algorithm and technologies can be investigated which can reduce
its computational complexity etc. Also the fuzzy the computational complexity of the algorithm.
extension to OWL is discussed in this chapter. Fuzzy extensions to description logics based
After reviewing all the proposals of fuzzy exten- on vague sets. A single membership degree in the
sions, we find that it is a paradox in the balance fuzzy sets is inaccurate to represent the impreci-
between the expressive power of the DL formalism sion in the membership degrees. As a result, based
and its computational complexity. It is sure that on vague sets (Gau, 1993), a fuzzy extension of
the computational complexity will become more description logic should be presented. Instead of
complex with the more expressive power of the a crisp degree of membership, two degrees of
DL formalism. membership (lower and upper degrees of mem-
bership) are used in the newly proposed fuzzy
description logic version. Its syntax, semantics
FUTURE RESEARCH DIRECTIONS and inference problems and tableaux also should
be investigated.
As overall conclusion, we can summarize that
until recently, research has not paid much at-
tention to fuzzy extensions to DLs in the area of REFERENCES
the Semantic Web. However, it gains more and
more interest and new approaches considering Baader, F., Calvanese, D., & McGuinness, D.
imprecise and uncertainty tend to emerge. Still, (2003). The Description Logic Handbook: Theory,
many of these approaches are rather half-baked Implementation and Applications. Cambridge
and a lot of things are missing: University Press.
Fuzzy extensions to concrete domain to support
Baader, F., & Hanschke P. (1991). A scheme for
the fuzzy user-defined data type and fuzzy user-
integrating concrete domains into concept lan-
defined data type predicate. It should be pointed
guages. In Proc. of the 12th Int. Joint Conf. On
out that, however, the SHIQ(G) and SHOQ(G)
Artificial Intelligence (IJCAI’91), pp. 452–457.
DLs presented in (Pan & Horrocks, 2006; Pan,
2007) which can support the user-defined data Berners-Lee, T., Hendler, J., & Lassila, O. (2001).
type and user-defined data type predicate can only The Semantic Web. The Scientific American, 284
deal with crisp knowledge. In the real world, hu- (5), 34- 43.
man knowledge and natural language have a big
Dinh-Khac, D., Hölldobler, S. & Tran, D. K.
deal of imprecision and uncertainty, as a result,
(2006). The fuzzy linguistic description logic
a fuzzy extension version of DL language which

74
A Review of Fuzzy Models for the Semantic Web

ALCFL. Proceedings of the 11th International Management of Uncertainty in Knowledge-Based


Conference on Information Processing and Systems, pp. 243-250.
Management of Uncertainty in Knowledge-Based
Pan., J. Z. (2007). A Flexible Ontology Reason-
Systems, 2096-2103.
ing Architecture for the Semantic Web. IEEE
Gau, W. L. & Buehrer, D. J., (1993) Vague sets. Transaction on Knowledge and Data Engineering.
IEEE Transactions on Systems, Man, and Cyber- 19(2), 246 - 260.
netics, 23 (2), 610-614.
Pan, J. Z. & Horrocks, I. (2006). OWL-Eu: Add-
Haarslev V., Lutz C., & Möller R. (1999). A ing Customised Data types into OWL. Journal
description logicwith concrete domains and role- of Web Semantics, 4(1), 29-49.
forming predicates. J. of Logic and Computation,
Lutz, C. (1999). Reasoning with concrete domains.
9(3), 351–384.
In Proc. of the 16th Int. Joint Conf. on Artificial
Hanschke P. (1992). Specifying role interaction in Intelligence (IJCAI’99), Stockholm, Sweden,
concept languages. In Proc. of the 3rd Int. Conf. pp. 90–95.
on the Principles of Knowledge Representation
Lutz, C. (2001). NEXPTIME-complete descrip-
and Reasoning (KR’92), Morgan Kaufmann, Los
tion logics with concrete domains. In Proc. of the
Altos, pp. 318–329.
Int. Joint Conf. on Automated Reasoning (IJCAR
Horrocks I., & Sattler, U. (2001). Ontology 2001), volume 2083 of Lecture Notes in Artificial
reasoning in the SHOQ(D) description logic. Intelligence, Springer, pp. 45–60.
Proceedings of the 17th International Joint Con-
Meghini, C., Sebastiani F. & Straccia, U. (1997).
ference on Artificial Intelligence (IJCAI 2001),
Reasoning about the form and content for multi-
pp. 199–204.
media objects. Proceedings of AAAI 1997 Spring
Hölldobler, S., Khang, T.D., & Störr, H.P. (2002) Symposium on Intelligent Integration and Use of
A fuzzy description logic with hedges as concept Text, Image, Video and Audio, pp. 89-94.
modifiers. In Proceedings In Tech/VJFuzzy’2002.
Ma. Z. M. (2006) Soft Computing in Ontologies
pp. 25–34.
and Semantic Web. Springer-Verlag.
Hölldobler, S, Störr, H. P. & Tran, D. K. (2003).
Sanchez, E., (2006a). Fuzzy Logic and the Se-
The fuzzy description logic ALCFH with hedge
mantic Web. Elsevier.
algebras as concept modifiers. Journal of Ad-
vanced Computational Intelligence and Intelligent Sanchez, E. & Yamanoi, T. (2006b). Fuzzy on-
Informatics, 7 (3), 294-305. tologies for the Semantic Web. Lecture Notes in
Artificial Intelligence, 4027: pp. 691-699.
Hölldobler, S., Nga, N. H. & Khang, T. D. (2005).
The fuzzy description logic ALCFLH. Proceedings Sánchez, D., & Tettamanzi, G. (2004). General-
of the 9th IASTED International Conference on izing quantification in fuzzy description logic. In
Artificial Intelligence and Soft Computing, pp. Proceedings 8th Fuzzy Days in Dortmund.
99-104.
Sánchez, D. & Tettamanzi, G. (2006). Reasoning
Hölldobler, S., Störr, H. P., & Khang, T. D. (2004). and quantification in fuzzy description logics.
The subsumption problem in the fuzzy description Lecture Notes in Artificial Intelligence, 3846,
logic ALCFH. Proceedings of the 10th Interna- pp. 81-88.
tional Conference on Information Processing and
Shadbolt, N., Hall, W. & Berners-Lee, T. (2006).

75
A Review of Fuzzy Models for the Semantic Web

The Semantic Web revisited. IEEE Intelligent Conference on Logics in Artificial Intelligence,
Systems, 21 (3), 96-101. pp. 385-399.
Stoilos, G., Stamou, G., & Tzouvaras, V. (2005a) Straccia, U. (2005). Towards a fuzzy description
A fuzzy description logic for multimedia knowl- logic for the semantic web. In Proceedings of the
edge representation. Proceedings of the 2005 2nd European Semantic Web Conference.
International Workshop on Multimedia and the
Tresp, C., & Molitor, R. (1998). A description
Semantic Web.
logic for vague knowledge. In Proc of the 13th
Stoilos, G., Stamou, G., & Tzouvaras V. (2005b) European Conf. on Artificial Intelligence (ECAI-
The fuzzy description logic f-SHIN. Proceedings 98). Brighton, England.
of the International Workshop on Uncertainty
Yen, J. (1991). Generalising term subsumption
Reasoning for the Semantic Web, pp. 67-76.
languages to fuzzy logic. In Proc of the 12th Int.
Stoilos, G., Stamou, G., & Tzouvaras, V. (2005c) Joint Conf on Artificial Intelligence (IJCAI-91),
Fuzzy OWL: Uncertainty and the Semantic Web. pp: 472-477, Sydney, Australia
Proceedings of the 2005 International Workshop
Zadeh, L. A. (1965). Fuzzy sets. Information and
on OWL: Experience and Directions.
Control, 8(3), 338-353.
Stoilos, G., Simous, N., & Stamou, G. (2006a),
Uncertainty and the Semantic Web. IEEE Intel-
ligent Systems, 21 (5), 84-87.
ADDITIONAL READING
Stoilos, G., Stamou, G. & Pan, J. Z., (2006b) Han-
dling imprecise knowledge with fuzzy descrip- General Logic
tion logic. Proceedings of the 2006 International
Workshop on Description Logics. Ben-Ari, M. (2001). Mathematical Logic for
Computer Scientists. Springer-Verlag.
Stoilos, G., Straccia, U., & Stamou, G. B. (2006c).
General concept inclusions in fuzzy description Ebbinghaus, H.-D., Flum, J. & Thomas, W. (2007)
logics. Proceedings of the 17th European Confer- Mathematical Logic. Springer-Verlag.
ence on Artificial Intelligence, pp. 457-461.
Enderton, H. B. (2002). A mathematical Introduc-
Straccia, U. (1998). A fuzzy description logic. In tion to Logic, 2nd edition. Academic Press.
Proc. of the 15th Nat. Conf. on Artificial Intelli-
Schoening, U. (1994). Logic for Computer Scien-
gence (AAAI-98), pp. 594–599 Madison, USA.
tists. Birkhaeuser Verlag
Straccia, U. (2001) Reasoning within fuzzy de-
scription logics. Journal of Artificial Intelligence
and Research, 14(1), 137–166
Fuzzy Set Theory

Straccia, U. (2004a). Transforming fuzzy de- Zadeh, L.A. (1978) Fuzzy sets as a basis for a
scription logics into classical description logics. theory of possibility, Fuzzy Sets Systems, 1(1),
Proceedings of the 9th European Conference on pp. 3-28.
Logics in Artificial Intelligence, pp. 385-399.
Straccia, U. (2004b). Fuzzy ALC with fuzzy con-
crete domains. Proceedings of the 9th European

76
A Review of Fuzzy Models for the Semantic Web

Semantic Web Smith, M. K., Welty, C. & McGuiness, D. L.


(2004). OWL Web Ontology Language Guide.
Antoniou, G. and Harmelen, F. V. (2003). A Se- W3C Recommendation, from http://www.w3.org/
mantic Web Primer, The MIT Press. TR/2004/REC-owl-ref-20040210/

Description Logic RDF and RDF Schema

Franconi, E., Description Logics Course Infor- A selection of documents on RDF and RDF
mation. http://www.cs.man.ac.uk/~franconi/dl/ Schema (Specification, Use Cases, Recommended
course/ Readings, Tools, Related Technologies, etc.), from
http://www.w3.org/RDF/
OWL
de Bruijn & J., Heymans, S. (2007). Logical Foun-
dations of (e)RDF(S): Complexity and Reasoning.
Horrocks, I. & Patel-Schneider, P. F., van Harmel-
In Proceedings of the International Semantic Web
en, F. (2003). From SHIQ and RDF to OWL: The
Conference (ISWC).
making of a web ontology language. Journal of
Web Semantics, 1(1):7-26.

This work was previously published in The Semantic Web for Knowledge and Data Management: Technologies and Practices,
edited by Z. Ma; H. Wang, pp. 23-37, copyright 2009 by Information Science Reference (an imprint of IGI Global).

77
78

Chapter 1.6
Semantic Web Adaptation
Alexander Mikroyannidis
University of Manchester, UK

Babis Theodoulidis
University of Manchester, UK

Introduction Semantic Web will contain machine-processable


metadata organized in ontologies. This will enhance
The rate of growth in the amount of information the way we search the Web and can even allow for
available in the World Wide Web has not been automatic reasoning on Web data with the use of
followed by similar advances in the way this in- software agents. Semantic Web adaptation brings
formation is organized and exploited. Web adap- traditional Web adaptation techniques into the new
tation seeks to address this issue by transforming era of the Semantic Web. The idea is to enable the
the topology of a Web site to help users in their Semantic Web to be constantly aligned to the users’
browsing tasks. In this sense, Web usage mining preferences. In order to achieve this, Web usage
techniques have been employed for years to study mining and text mining methodologies are employed
how the Web is used in order to make Web sites for the semi-automatic construction and evolution
more user-friendly. of Web ontologies. This usage-driven evolution of
The Semantic Web is an ambitious initiative Web ontologies, in parallel with Web topologies
aiming to transform the Web to a well-organized evolution, can bring the Semantic Web closer to
source of information. In particular, apart from the users’ expectations.
the unstructured information of today’s Web, the

DOI: 10.4018/978-1-59904-845-1.ch093

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Semantic Web Adaptation

Background the extraction of the site’s structure for use in the


mining procedure. The hyperlinks of each Web
Web Usage Mining page build the structure of the Web site. Most Web
sites have nowadays an utterly dynamic topology,
Web usage mining has a wide variety of applica- thus presenting a different structure to different
tions. User profiles can be created for use in Web users. This characteristic should be considered
personalization. Information can also be extracted during the structure preprocessing phase.
that details how a Web site can be reorganized to The preprocessing stage is followed by the
better facilitate users’ navigation through it. In discovery of traversal patterns from the user
e-commerce Web sites, the results of Web usage access data. Traversal patterns reveal the way a
mining can be used to improve sales. Analyzing user navigates through the site during each ses-
user access patterns can also help when targeting sion. Clusters of users can be discovered through
advertisements to specific groups of users. clustering of similar traversal patterns. Moreover,
Srivastava, Cooley, Deshpande, and Tan (2000) association rules can be applied to the pages
divide Web usage mining into three stages: accessed during a session, independent of their
ordering. Examples of association rules that were
i. Preprocessing extracted from an IBM analysis of the Web logs
ii. Pattern discovery of the Official 1996 Olympics Web site (Elo-Dean
iii. Pattern analysis & Viveros, 1997) are:

Preprocessing consists of converting the us- • 45% of the visitors who accessed a page
age, content, and structure information contained about Indoor Volleyball also accessed a
in the various available data sources into the data page on Handball.
abstractions necessary for pattern discovery. Us- • 59.7% of the visitors who accessed pages
age preprocessing involves the identification of about Badminton and Diving also accessed
users and their visiting sessions. In order for this a page about Table Tennis.
to be accomplished, several difficulties need to
be overcome. For example, proxy servers hide the The percentages mentioned in both association
actual IP addresses of the machines that are using rules are called confidence. Confidence can be
them, thus making user identification problematic. defined as the number of transactions containing
A user that uses more than one browser, even on all of the items in a rule, divided by the number
the same machine, will appear as multiple users. of transactions containing the rule antecedents
Tracking repeat visitors can also be complex if a (Cooley, Mobasher, & Srivastava, 1999). Addi-
user uses different machines. tionally, temporal relationships among data items
Content preprocessing consists of converting can be discovered, such as the following (Cooley,
the text, image, scripts, or multimedia files into Mobasher, & Srivastava, 1997):
forms that are useful for the Web usage mining
process. This often involves the application of • 30% of clients who visited the ‘/company/
content mining techniques, such as classifica- products’ page had done a search in Yahoo
tion or clustering. For instance, a classification within the past week on keyword w.
algorithm could be used to limit the discovered • 60% of clients who placed an online or-
patterns to those that contain page views about der in the ‘/company/product1’ page also
a certain subject. Similar to the preprocessing of placed an online order in the ‘/company/
the site’s content, structure preprocessing regards product4’ page within 15 days.

79
Semantic Web Adaptation

Depending on the purpose of the mining, a language for extracting navigation patterns,
traversal pattern may contain backward traversals. called MiDAS, has been proposed by Buchner,
Backward traversals consist of references of pages Baumgarten, Anand, Mulvenna, and Hughes
earlier visited. Studying backward traversals can (1999). Xiao and Dunham (2001) have inves-
help discover missing hyperlinks, which if added tigated techniques to discover frequently used
will reduce these traversals, thus making naviga- contiguous sequences of page references, which
tion paths shorter and more convenient for the they call maximal frequent sequences (MFS).
users. A pattern’s accesses can also be restricted They have also developed an algorithm called
only to contiguous ones, which can be used for online adaptive traversal (OAT) pattern mining, to
prefetching and caching purposes. A frequent pat- mine MFS. Xing and Shen (2004) have proposed
tern is maximal if does not contain any frequent two algorithms, user access matrix (UAM) and
subpatterns. This can reduce significantly the preferred navigation tree (PNT), for mining user
number of meaningful discovered patterns. preferred navigation patterns.
Pattern analysis is the last stage of the Web
usage mining process. The patterns that have Web Adaptation
been produced are reviewed, and useful informa-
tion is extracted from them. Knowledge query Several Web adaptation systems have been devel-
mechanisms, similar to structured query language oped over the years, mainly based on Web usage
(SQL), can be used to filter out the patterns. mining techniques. The WebWatcher system
Another approach involves the use of data cubes (Joachims, Freitag, & Mitchell, 1997) suggests
and OLAP operations. Visualization techniques, links that may interest a user, based on other us-
such as graphing patterns or assigning colors to ers’ online behaviour. The system is implemented
different values, can also be utilized to highlight in the form of a proxy server. Each user is asked,
interesting trends in the data. Last but not least, upon entering the site, what kind of information
content and structure information can be used to he is seeking. Before he departs, the user is asked
filter out patterns containing pages of a certain whether he has found what he was looking for.
usage type, content type, or pages that match a His navigation paths are used to deduce sugges-
certain hyperlink structure. tions for future visitors that seek the same content.
Numerous approaches to Web usage mining These suggestions are visualized by highlighting
have been followed, targeted to a wide range existing hyperlinks.
of applications. Chen, Park, and Yu (1998) and The Avanti project (Fink, Kobsa, & Nill, 1996)
Nanopoulos and Manolopoulos (2000) have tries to predict the visitor’s final objective as well
introduced the concept of using the maximal as her next step. A model for the visitor is built,
forward references to break down user sessions based partly on the personal information of the
into transactions for mining access patterns. Yang, visitor and partly on information extracted from
Pan, and Chung (2001) have proposed an efficient her navigation paths. Visitors are provided with
hash-based method, HMFS, for discovering the direct links to pages that are probably the ones
maximal frequent itemsets. Spiliopoulou (1999) they are looking for. In addition, hyperlinks that
has presented an algorithm for building aggregat- lead to pages of potential interest to each visitor
ing trees from Web logs, then mining the Web ac- are highlighted. The suggestions are extracted
cess patterns by MINT mining language. (Cooley with the use of association rules that are applied
et al. (1999) have provided a query language on to the user’s model.
top of external mining software for association A drawback of both the WebWatcher and the
rules and for sequential patterns. Another query Avanti approaches is that they require the ac-

80
Semantic Web Adaptation

tive participation of the users in the adaptation to work in cooperation” (Berners-Lee, Hendler,
process, by asking them to provide information & Lassila, 2001).
about themselves. On the other hand, Footprints Ontologies are a key enabling technology
(Wexelblat & Maes, 1999) relies entirely on the for the Semantic Web, since they offer a way to
navigation paths of the users. The navigation paths give information a common representation and
of all visitors are recorded and the most frequent semantics. Daconta, Obrst, and Smith (2003)
ones are presented to the visitor. Additionally, distinguish three levels of ontologies: top, middle,
next to each link, the percentage of people who and lower domain levels. At the top level, the
have followed it is displayed. ontological information represented concerns
Perkowitz and Etzioni (2000) have proposed primary semantic distinctions that apply to every
a conceptual framework for adaptive Web sites, ontology. The middle level represents knowledge
focusing on the semi-automatic creation of index that spans domains and may not be as general as
pages from clusters of pages. They have developed the knowledge of the upper level. Finally, the
two cluster mining algorithms, PageGather and lower level represents ontologies at the domain
IndexFinder. The first one relies on a statistical or subdomain level. This is typically knowledge
approach to discover candidate link sets, while about domain-specific subject areas. While an
the second finds link sets that are conceptually ontologist can address the upper and to a cer-
coherent. tain extent the middle level, the domain expert
is absolutely required for the construction and
The Semantic Web maintenance of the lower level.
Mikroyannidis and Theodoulidis (2005)
Since January 2005, the Web has grown by more distinguish between the domain ontology and
than 17 million sites, according to monitoring firm the ontology of a Web site dedicated to a certain
Netcraft. This figure exceeds the growth of 16 domain. A Web site ontology is strongly related
million sites seen in 2000 when net fever reached to the topology of the site and is comprised of the
its most intense pitch.1 Significant progress has thematic categories covered by the site’s pages.
been made in technologies for publication and These categories are the concepts of the ontol-
distribution of knowledge and information on the ogy. Each Web page, depending on its content,
Web. People add private, educational, or corpo- is an instance of one or more concepts of the
rate content. Growth also comes from the rise in ontology. The concepts can be related to each
blogging, in which users write regularly updated other through a number of relationship types,
Web journals on various topics. However, much representing the associations the concepts have
of the published information is not organized, and according to the Webmaster’s perception. Figure
it is hard to find answers to questions that require 1 shows the Web site ontology belonging to the
more than a keyword search. University of Manchester School of Informatics
The Semantic Web (Berners-Lee, Fisch- (www.informatics.manchester.ac.uk). The ontol-
etti, & Dertouzos, 1999) means to address these ogy has been built considering the organization
problems by expressing Web data in forms that of the thematic categories as this is defined in
are machine-processable, in order to be more the current topology of the site. The hierarchy’s
efficiently maintained by software agents, thus top level contains seven classes: School, Un-
enhancing precision of search, as well as logical dergraduate Programmes, Postgraduate Taught
reasoning. The vision behind this concept can be Programmes, Postgraduate Research, Research,
summarized as “giving information a well-defined News and Intranet. These are the main thematic
meaning, better enabling computers and people categories of the site. These categories are then

81
Semantic Web Adaptation

Figure 1. The School of Informatics Web site ontology

expanded to more specific concepts, which are Semantic Web Adaptation


represented by subclasses.
It should be pointed out that the Web site ontol- The Semantic Web is undoubtedly a remarkable
ogy is quite different from the domain ontology. advance in the area of information management.
The latter describes relationships between the The magnitude of Web data necessitates the use
concepts of a domain, whereas the first is based of machine-processable metadata. Nevertheless,
on the organization of the information found in a the Web users’ needs and requirements should
Web site. The ontology of a domain is usually more not be neglected in the process of building and
complex than the ontology of a Web site related to maintaining the Semantic Web. Semantic Web
the same domain. However, the maintenance of a adaptation utilizes traditional Web usage mining
Web site ontology requires considerable effort and methodologies and extends them in order to ad-
has to be performed on a regular basis, since the dress the ontological perspective of the Semantic
content of a Web site is constantly updated. Web. Apart from the topology of the Web, which is
targeted in a conventional Web adaptation system,

82
Semantic Web Adaptation

semantic Web adaptation also aims to the evolu- procedure aims at the evolution of the topology
tion of Semantic Web ontologies. and ontology of the Web site.
The Heraclitus framework (Mikroyannidis, The adaptation starts with a preprocessing
2004; Mikroyannidis & Theodoulidis, 2004, 2005) stage, during which the data stored in the raw
proposes the transformation of the Semantic Web access logs are cleaned and visiting sessions are
based on Web usage data. Web usage mining and identified. The sessions are then mined with the
text mining are used for the extraction of knowl- use of market basket analysis (Bodon, 2003) to
edge from navigation paths and the adaptation of retrieve frequent itemsets. These are then classified
the physical and semantic structure of the Web. based on the Web site topology and ontology in
An implementation of the Heraclitus framework order to produce page sets, that is, sets of pages
has been released as a suite of open source tools that are frequently accessed together throughout
(http://heraclitus.sourceforge.net). the same session. During classification, each
itemset is assigned its position in the topology
Architecture and the ontology of the Web site. For the latter,
automatic categorization is performed with the use
Figure 2 presents the architecture of the Heraclitus of the support vector machines (SVM) algorithm
framework. As it can be seen, the inputs of the (Cortes & Vapnik, 1995). SVM has been trained
adaptation process consist of raw access logs, on the thematic categories that are defined in the
the Web site topology and ontology. The whole concepts of the Webmaster’s ontology. Each page

Figure 2. Semantic Web adaptation architecture

83
Semantic Web Adaptation

of the page sets is then assigned by SVM to one Figure 3 shows an example of a modified
or more concepts of the ontology. Web page, according to the proposed Heraclitus
Based on the topological and ontological fea- adaptations. The page has been modified to fa-
tures of the page sets, a report containing proposals cilitate the navigation of the users during the first
for the improvement of the Web site is generated. semester. Shortcut links to popular courses of the
This report contains proposals for the insertion of first semester have been inserted in the left side
shortcut links from source pages to target pages of the page, under the title “Quick links.” More-
that are frequently accessed together but are cur- over, popular links that already existed, such as
rently not linked. It also contains proposals for the hyperlink leading to the page of the “Personal
the change of the appearance of popular hyper- and Professional Development” course, have been
links. In addition, the report contains proposals highlighted.
for the evolution of the Web site ontology. After The Web site ontology was modified in several
the proposed modifications have been revised by ways, based on the outcomes retrieved from the
the Webmaster, they can be applied to the Web classified page sets. The resulting ontology, after
site. The site topology is then refined through the the application of Heraclitus adaptations, is shown
insertion of new shortcut links, as well as changes in Figure 4. Based on these adaptations, the content
in the appearance of the existing ones. The ontol- organization of the Web site was altered to better
ogy is also refined in a number of ways. satisfy the needs of its visitors. First of all, new
associations were discovered between concepts.
Case Study These associations reflect the interests of the us-
ers, as documents belonging to these concepts are
The Web site of the School of Informatics at the frequently accessed together. In particular, new
University of Manchester (www.informatics. associations were inserted between the following
manchester.ac.uk) was used as a case study for concepts:
the Heraclitus framework. The topology of the
Web site was refined through the insertion of • “Research” and “Students”
new shortcut links between pages that were not • “Research” and “Staff”
previously linked together, as well as through the • “School” and “Students”
highlighting of popular existing links. In addition, • “School” and “Staff”
the Web site ontology was modified in several • “Students” and “Staff”
ways, based on the outcomes retrieved from the
classified page sets. Reorganization of the concepts’ hierarchy was
More specifically, Heraclitus produced two sets also performed. Further improvements included
of reports: shortcut links reports and highlighted the creation of new categories, the removal of
links reports. Page sets of unlinked pages suggest existing categories, as well as changes to the
the insertion of shortcut links between these pages, levels of hierarchy that the concepts belong to.
in order to achieve shorter navigation paths. From For instance, the “Staff” concept was previously a
the page sets of linked pages, changes in the ap- subconcept of the “School” concept, which resided
pearance of existing links can be extracted. For in the highest level of the ontology. It should be
example, if an index page and some of its links noted that the “Staff” concept has as instances all
comprise one or more page sets, then highlighting the Web pages that carry information about the
these links in the index page will provide valuable staff members of the school. However, the high
help to visitors. frequency with which this concept appeared in

84
Semantic Web Adaptation

Figure 3. Proposed Heraclitus adaptations in a sample Web page

the page sets implies the significance that it has corresponds to the “Job Vacancies” concept, was
in the interests of the users. It would be thus ap- found to be an instance of both the “Staff” and
propriate to transfer this concept to the top level “Research” concepts. The information contained
of the ontology, as shown in Figure 4. Based on in this page regards mainly research job posts and
the performed classification, the undergraduate is also highly related to the “Staff” concept. This
and postgraduate programmes were grouped page was previously categorized only under the
under the more general concept “Students.” The “School” concept. In the updated ontology, the
“School” concept was also extended to include “Job Vacancies” concept has been placed both
more subconcepts. under the “Staff” and “Research” concepts. The
The ontology of the Web site was extended to same modification has been applied to the concepts
include multiple instances of concepts or multiple “Visiting Professors,” “Administrative Informa-
subconcepts. The categorization of the Web pages tion for Students,” and so forth.
that was carried out suggested that several pages Finally, useful conclusions were deduced about
belong to more than one concept. Moreover, the usage of the Web site. Particularly, the thematic
in some cases, Web pages and the correspond- category that was the first in the preferences of the
ing concepts were categorized under different users was, as expected, the “Students” concept.
concepts than they previously were in the exist- This concept contains all pages that support the
ing ontology. The Web site ontology should be school’s modules, both undergraduate and post-
therefore updated in order to reflect this fact. For graduate. This is not surprising, since most of
example, the “Job vacancies” Web page, which the traffic is generated by the students. Second,

85
Semantic Web Adaptation

Figure 4. Refined Web site ontology for the School of Informatics

in the users’ interests comes the “Staff” concept. machine-processable metadata. Nevertheless, the
The “Research” concept is third, followed by needs and requirements of the Web users should
the “School” category. These results can be used not be neglected in the process of building and
to enhance the performance of the server, for maintaining the Semantic Web.
example, by the use of additional servers that The Heraclitus framework approaches Se-
will host the popular resources, or to promote mantic Web adaptation from a user-oriented
the problematic concepts by making them more perspective. Transformation of the Web topology
easily accessible. and ontology is carried out having as basis the
information retrieval tasks of the visitors. Web
usage mining and text mining are used for the
Conclusion extraction of knowledge from navigation paths
of the visitors. Practicing semantic Web adapta-
The Semantic Web is undoubtedly a remarkable tion on a real Web site has provided an insight in
advance in the area of information management. navigation difficulties of the users as well as in
The magnitude of Web data necessitates the use of ways to overcome them.

86
Semantic Web Adaptation

References Fink, J., Kobsa, A., & Nill, A. (1996). User-ori-


ented adaptivity and adaptability in the AVANTI
Berners-Lee, T., Fischetti, M., & Dertouzos, M. project. Paper presented at the Designing for the
(1999). Weaving the Web: The original design Web: Empirical Conference.
and ultimate destiny of the World Wide Web by
its inventor. San Francisco: Harper. Joachims, T., Freitag, D., & Mitchell, T. (1997).
WebWatcher: A tour guide for the World Wide
Berners-Lee, T., Hendler, J., & Lassila, O. (2001). Web. Paper presented at the International Joint
The Semantic Web. Scientific American, 279(5), Conference on Artificial Intelligence, Nagoya,
34–43. Japan.
Bodon, F. (2003). A fast APRIORI implementation. Mikroyannidis, A. (2004). Development of a
Paper presented at the IEEE ICDM Workshop framework for self adaptive Web sites. Unpub-
on Frequent Itemset Mining Implementations lished MPhil, University of Manchester, Man-
(FIMI’03), Melbourne, FL. chester, UK.
Buchner, A. G., Baumgarten, M., Anand, S. S., Mikroyannidis, A., & Theodoulidis, B. (2004,
Mulvenna, M. D., & Hughes, J. G. (1999). Naviga- September 20-24). A theoretical framework and
tion pattern discovery from Internet data. Paper an implementation architecture for self adaptive
presented at the WEBKDD’99, San Diego, CA. Web sites. Paper presented at the IEEE/WIC/ACM
Chen, M. S., Park, J. S., & Yu, P. S. (1998). Effi- International Conference on Web Intelligence
cient data mining for path traversal patterns. IEEE (WI’04), Beijing, China.
Transactions on Knowledge and Data Engineer- Mikroyannidis, A., & Theodoulidis, B. (2005,
ing, 10(2), 209–221. doi:10.1109/69.683753 May 29-June 1). Web usage driven adaptation
Cooley, R., Mobasher, B., & Srivastava, J. (1997). of the Semantic Web. Paper presented at the End
Web mining: Information and pattern discovery User Aspects of the Semantic Web Workshop,
on the World Wide Web. Paper presented at the 2nd European Semantic Web Conference (ESWC
IEEE International Conference on Tools with 2005), Heraklion, Greece.
Artificial Intelligence. Nanopoulos, A., & Manolopoulos, Y. (2000).
Cooley, R., Mobasher, B., & Srivastava, J. (1999). Finding generalized path patterns for Web log
Data preparation for mining World Wide Web data mining. Paper presented at the East-European
browsing patterns. Knowledge and Information Conference on Advances in Databases and In-
Systems, 1(1), 5–32. formation Systems (ADBIS’00), Prague, Czech
Republic.
Cortes, C., & Vapnik, V. (1995). Support vector
networks. Machine Learning, 20(3), 273–297. Perkowitz, M., & Etzioni, O. (2000). Adaptive
Web sites. Communications of the ACM, 43(8),
Daconta, M., Obrst, L., & Smith, K. (2003). The 152–158. doi:10.1145/345124.345171
Semantic Web: A guide to the future of XML, Web
services, and knowledge management. John Wiley Spiliopoulou, M. (1999). The laborious way from
and Sons Ltd. data mining to Web mining. International Journal
of Computer Systems, Science and Engineering
Elo-Dean, S., & Viveros, M. (1997). Data mining . Special Issue on Semantics of the Web, 14,
the IBM official 1996 Olympics Web site (Technical 113–126.
report). IBM T.J. Watson Research Center.

87
Semantic Web Adaptation

Srivastava, J., Cooley, R., Deshpande, M., & Web Adaptation: The process of transform-
Tan, P. (2000). Web usage mining: Discovery ing the topology of the Web in order to align it
and applications of usage patterns from Web data. with the preferences of the users, thus facilitating
SIGKDD Explorations, 12-23. their browsing.
Xiao, Y., & Dunham, M. (2001). Efficient mining
Web Mining: Mining data related to the Word
of traversal patterns. Data & Knowledge Engineer-
Wide Web, such as the content of Web pages,
ing archive, 39(2), 191-214.
intrapage structure, which includes the HTML or
Xing, D., & Shen, J. (2004). Efficient data min- XML code of a page, interpage structure that is the
ing for Web navigation patterns. Information and linkage structure between Web pages, usage data
Software Technology, 46(1), 55–63. doi:10.1016/ that describe how Web pages are accessed, and
S0950-5849(03)00109-5 user profiles, including demographic, registration
information, or information found in cookies.
Yang, D., Pan, C., & Chung, Y. (2001). An efficient
hash-based method for discovering the maximal
Web Site Ontology: An ontology whose con-
frequent set. Paper presented at the 25th Annual
cepts are the thematic categories covered by the
International Computer Software and Applications
pages of a Web site. Each Web page, depending on
Conference (COMPSAC’01), Chicago, Illinois.
its content, is an instance of one or more concepts
of the ontology. The concepts are related to each
other through a number of relationship types,
Key Terms representing the associations the concepts have
according to the Webmaster’s perception.
Ontology: A representation of a certain Web Usage Mining: An application of data
domain, through the definition of concepts, re- mining methodologies to Web access logs in order
lationships between concepts, and instances of to discover trends and regularities in navigation
concepts. patterns of Web users.
Semantic Web Adaptation: The process of
transforming the topology and ontology of the
Web in order to improve its usability. EndNote
1
Web Access Log: A listing of page reference “Web enjoys year of biggest growth” (http://
data. Web access logs are created by Web servers news.bbc.co.uk/2/hi/technology/4325918.
in order to keep track of the requests that occur stm), BBC News, 10 October 2005.
on Web sites by Web users.

This work was previously published in Encyclopedia of Information Communication Technology, edited by A. Cartelli; M.
Palma, pp. 704-712, copyright 2009 by Information Science Reference (an imprint of IGI Global).

88
89

Chapter 1.7
Tips for Tracking Web
Information Seeking Behavior
Brian Detlor
McMaster University, Canada

Maureen Hupfer
McMaster University, Canada

Umar Ruhi
University of Ottawa, Canada

Abstract privacy concerns at the start of any collection of


Web tracking information.
This chapter provides various tips for practitioners
and researchers who wish to track end-user Web
information seeking behavior. These tips are derived INTRODUCTION
in large part from the authors’ own experience of
collecting and analyzing individual differences, Upon first consideration, employing Web tracking
task, and Web tracking data to investigate people’s to better understand end-user experiences with the
online information seeking behaviors at a specific Web seems to be a simple process of installing the
municipal community portal site (myhamilton.ca). tracking software, collecting the data over a certain
The tips discussed in this chapter include: (1) the period of time, and conducting the analysis. How-
need to account for both task and individual dif- ever, our own experience in setting up, collecting,
ferences in any Web information seeking behavior and analyzing Web tracking data has shown us
analysis; (2) how to collect Web metrics through that the process is surprisingly more difficult than
deployment of a unique ID that links individual originally expected.
differences, task, and Web tracking data together; To share what we have learned to help others
(3) the types of Web log metrics to collect; (4) set up and better utilize Web tracking tools, we
how to go about collecting and making sense of have reflected upon what we believe are key tips
such metrics; and (5) the importance of addressing concerning the use of Web tracking in any Web infor-
mation seeking analysis. Thus, the overall purpose
DOI: 10.4018/978-1-59904-974-8.ch013

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Tips for Tracking Web Information Seeking Behavior

of this chapter is to discuss the practicalities and • The benefits of using a unique ID to link
usefulness of collecting Web tracking data to help individual differences, task, and Web track-
measure and assess the performance and usage of ing data
a Website or application, particularly with respect • The types of Web metrics to collect
to Web information seeking. • How to gather and make sense of the Web
Note that the ideas presented in this chapter are metric information that is collected in Web
grounded in a research project conducted by the logfiles
authors over the last three years that investigates • The importance of addressing privacy con-
people’s online behaviors at a municipal com- cerns right up-front in the collection of
munity portal site called myhamilton.ca (www. Web tracking information
myhamilton.ca). The ultimate goal of the project is
to understand the relationships among individual We begin by providing background on the
user characteristics such as demographics and need to take both task and individual differences
personality traits, user attitudes toward and per- into consideration when investigating end-user
ceptions about accomplishing certain tasks (Web Web information seeking behavior. To do this,
services) online, and actual usage behavior. We we provide a general model that describes how
believe that an understanding of these relationships task and individual differences affect information
will provide insight into how characteristics of seeking behavior. Next, methods to conduct a Web
the individual, the task, and utilization behaviors information seeking analysis that allows for the
affect task performance in an online community collection of both task and individual differences
environment. We also believe that the capture data are presented. Importantly, these methods
and analysis of Web tracking data is imperative include the collection of Web tracking data via
to reaching such an understanding. the use of Web logs. Using a selective subset
The difficulty in utilizing Web tracking data of variables from the general model presented
successfully is in knowing how to position its earlier, our own myhamilton.ca project serves
collection and use within the larger confines of as a point of illustration. We also provide details
Web information seeking analysis. Web tracking with respect to the types of Web metrics to col-
is just one tool that needs to be coordinated with lect and what needs to be done to make sense of
other data collection methods to yield a more these data. Finally, the importance of addressing
comprehensive understanding than Web tracking privacy in any Web information seeking analysis
alone could ultimately provide. is highlighted.
The objective of this chapter is to raise aware- To help clarify things, find below the following
ness of this point and to suggest techniques and definitions of terms:
approaches for the collection and analysis of Web
tracking information that will aid practitioners in • Information seeking behavior refers to
their performance measurement initiatives and how people seek information in different
understanding of how end-users seek information contexts (Fisher, Erdelez & McKechnie,
on the Web. Various tips are presented: 2005).
• Web information seeking behavior re-
• The need to account for both task and indi- fers to information seeking behaviors
vidual differences in any Web information that occur over the Web. Choo, Detlor &
seeking analysis assessment Turnbull (2000) identify four main modes

90
Tips for Tracking Web Information Seeking Behavior

of information seeking on the Web ranging 2005). Scholars in both fields, as well as those in
from wayward browsing to goal-directed psychology, also have begun to examine the role
search (undirected viewing, conditioned of individual differences in online behavior (e.g.,
viewing, informal search, and formal Bhatnagar & Ghose, 2004; Das, Echambadi, Mc-
search) where each mode is character- Cardle, & Luckett, 2003; Dillon & Watson, 1996;
ized by predominant information seeking Ford, Miller & Moss 2001, 2005a, 2005b; Gugerty,
moves or activities (undirected viewing: Treadaway & Rubinstein, 2006; Heinström, 2005;
starting and chaining; conditioned view- Ho, 2005; Martin, Sherrard & Wentzel, 2005;
ing: browsing and differentiating; informal Tuten & Bosnjak 2001). Unfortunately, the study
search: differentiating, monitoring, and of individual differences in information seeking
extracting; and formal search: monitoring has tended to take a haphazard approach that has
and extracting). failed to link findings with broader theoretical
• Individual differences are the demographic frameworks concerning information seeking be-
and psychological characteristics of people havior and has neglected to study the effects of
that distinguish one person from another. individual differences in conjunction with specific
• Task in this chapter refers to the informa- seeking contexts (Saracevic, 1991). The work
tion seeking task an individual user expe- of Ford et al. (2001, 2005a, 2005b) is a notable
riences that instills a need for information exception both for its use of Wilson’s model of
and motivates the user to satisfy this infor- information behavior (Wilson & Walsh, 1996) as
mation need through some sort of informa- a basis for investigation and for its examination
tion seeking behavior. Task is the context of how information seeking complexity and in-
surrounding a person’s information need. dividual differences in cognitive style interact to
• Web tracking refers to the automated col- result in differing information seeking strategies
lection of Web information seeking behav- (Ford et al., 2005b).
ioral data. To situate individual differences within the
• Web metrics pertains to the measures by Web information seeking context, we propose
which to assess a person’s Web informa- our own model of information seeking behavior
tion seeking behavior or to assess and that utilizes Wilson’s (1999) model as a theoreti-
monitor activity on a Website. Examples of cal foundation (see Figure 1). According to this
commonly used Web metrics include page model, task (analogous to Wilson’s “context of the
views, page transitions, and session times. information need” construct) leads to information
seeking behavior that is mediated by individual
differences variables. The purpose of the model
INDIVIDUAL DIFFERENCES, is to illustrate how task and individual differences
TASKS, AND INFORMATION fit into and influence the end-user information
SEEKING BEHAVIOR seeking process, and to stress the importance of
the need to take both task and individual differ-
Research concerning online information seeking ences into account when planning any type of Web
in both information science and marketing has information seeking analysis assessment.
shown that information seeking strategy depends As Figure 1 shows, with respect to task, there
on the type of information seeking task or its are a variety of characteristics about a task that
context (e.g., Bhatnagar & Ghose, 2004; Moe can influence an end-user’s information seeking
2003; Toms & Trifts, 2006; Wildemuth & Hughes, behavior. For instance, prior research has found

91
Tips for Tracking Web Information Seeking Behavior

Figure 1. How task and individual differences affect information seeking behavior

substantial differences in information seeking is able to mentally formulate the parameters of


patterns across tasks and between product catego- an information search. For example, consumers
ries or information domains (e.g., Bhatnagar & who are well aware of their current product needs
Ghose, 2004; Trifts & Toms, 2006; Wildemuth & may be easily able to articulate this need in the
Hughes, 2005). These differences, in part, can be form of a search query (e.g., buying a particular
explained by the complexity of the information DVD), but a decision made with less specificity
seeking task, the extent to which the task is clearly (e.g., planning a vacation with no particular des-
structured, whether one is seeking information for tination in mind) may require broadening of the
oneself or for someone else (Hupfer & Detlor, search parameters to learn more about the various
2006), and semantic differences between search alternatives available.
domains (Byström & Järvelin, 1995; Vakkari, Even within a specified information domain,
1999, pp. 825). Ford et al. (2003) suggest that task complexity is influenced by such factors as
complex tasks require a conceptual broadening the number of alternatives available, the num-
of useable terminology to reflect broader search ber of dimensions of information on which the
concepts. Conversely, simple tasks would be alternatives vary, and time pressure (Payne, Bet-
ones in which all essential concepts necessary to tman & Johnson, 1993). Greater complexity in
complete the information seeking task are fully an information seeking task often leads to more
specified in the task instructions. In a consumer heuristic-based processing of information. Deci-
decision making context, this conceptual broaden- sion strategies that require processing information
ing may be closely related to how well a consumer by attribute as opposed to alternative are thought

92
Tips for Tracking Web Information Seeking Behavior

to be easier to undertake. an important role in terms of mediating the effect


In the context of online information seeking, of task on information seeking behavior in terms
tasks that require people to search by attribute as of the information seeking strategy or process cho-
opposed to alternative may be cognitively less sen, as well as its effectiveness. These differences
complex and require less time at the general search may include an individual’s familiarity or level of
tool level. Therefore, even within a specified involvement with the information topic (Moorthy,
domain, differences in information seeking effort Ratchford & Talukdar,1997), experience with the
allocation may occur, depending upon a person’s Internet (Bhatnar & Ghose, 2004), perceptions of
information seeking orientation (Huneke, Cole & Web-based information seeking (Ford & Miller,
Levin, 2004). That is, whether an individual is 1996), and enduring psychological traits (e.g.,
engaging in attribute-based or alternative-based Ford et al., 2001; 2005a; 2005b).
processing of information affects the allocation of For example, Bhatnar and Ghose (2004) found
information seeking effort between general search that users with greater experience with the Internet
engines versus specific Websites (Toms & Trifts and more education utilized the Internet more
2006). Those who are engaged in alternative- frequently. Other demographic characteristics,
based processing are more focused on finding including age and sex, also have been associated
an appropriate source of information and thus with differences in Web information seeking pat-
allocate a greater amount of their information terns. Ford et al. (2001) established that informa-
seeking effort at a general search engine as op- tion retrieval effectiveness was associated with
posed to in-site search. males while retrieval failure was associated with
Other characteristics of the task shown in Fig- females. Women felt that they were not in control
ure 1 that can affect information seeking behavior of their information seeking; they were unable to
are whether the information seeking task is aided/ avoid irrelevant material and stay on target. Men,
unaided or ongoing/situational. The former refers however, were confident that they were in control
to the extent to which Web information seeking is and could bypass extraneous content. Educational
assisted by interactive decision aids. For example, research conducted with children also has found
in the domain of online shopping, Haubl and Trifts sex differences in information seeking such that
(2000) found that interactive tools that assisted boys searched differently from girls and were able
consumers in their initial screening of alternatives to acquire more target-specific and target-related
substantially reduced the amount of information information. Boys filtered information at an early
seeking undertaken and improved decision mak- stage but girls were linear and more thorough
ing. Despite the initial learning that is required, navigators (Roy & Chi, 2003).
in the long run the use of interactive decision aids As Figure 1 illustrates, there are several
should reduce task complexity such that users will psychological differences that may influence
be able to devote less effort to obtain the required information seeking behavior in terms of an
information than they would expend if unassisted. individual’s propensity to engage in elaborate,
The latter refers to whether information is needed effortful processing versus effort minimization
on an ongoing basis, such as when an individual and reliance on heuristics. Explained below, these
has an interest in a product category or topic but include: verbalizer /imager and holistic/analytic
does not intend to make a decision immediately, cognitive styles; deep, surface and strategic learn-
or whether information is needed for present use, ing approaches; cognitive complexity; Need for
such as a pre-purchase situation in which a deci- Cognition; and Self- and Other-Orientation.
sion is imminent.
As Figure 1 shows, individual differences play

93
Tips for Tracking Web Information Seeking Behavior

Cognitive Style Learning Style

Individuals differ in the strategies that they use Those with a surface approach describe learning
to seek and process information, and they tend to as knowledge reproduction achieved through
favor certain strategies, or cognitive styles, on a rote learning and memorization. They are pas-
consistent basis. Among these styles, verbalizer/ sive uncritical learners who devote relatively
imager and holist/analytic are the two dominant little effort to information seeking (Ford, 1986;
dimensions (Riding & Cheema, 1991). The verbal- Entwistle & Tait, 1995). Deep learners, on the
izer/imager dimension refers to a preference for other hand, view learning as a process that creates
and facility with tasks and information that are knowledge through the synthesis and assimilation
presented in a verbal versus visual format; verbal of new information. They seek a broad range of
and spatial ability are closely related measures information sources using a variety of information
(Ekstrom, French & Harman, 1976). Analytic seeking strategies. Strategic learners are able to
individuals perceive components of complex choose either deep or surface learning approaches
stimuli as discrete elements and are better able as appropriate to the task at hand. Analysis of self-
to analyze and impose structure than those who reported information seeking behavior has found
are holist, with their tendency to perceive stimuli that a surface approach to learning was associated
in a holistic or global manner. Where Internet with a fast surfing information seeking strategy
searching is concerned, Wang, Hawk and Tenopir in which users experienced problems with critical
(2000) found that holist searchers experienced analysis and had difficulty judging the relevance
more difficulty and confusion than analytic users. of retrieved documents (Heinström, 2005). They
Ford et al. (2001) found that poor retrieval was also demonstrated confirmatory bias and preferred
linked to a verbalizer cognitive style, as well as to access information using only a few documents.
perceptions that the Internet’s graphic elements In contrast, a deep diving approach characterized
were of little value. Similarly, Gugerty et al. (2006) those with either deep or strategic learning styles;
demonstrated that superior spatial, rather than these individuals were effortful information seek-
verbal, ability was associated with more favorable ers who sought high quality documents. In addi-
computer and Internet attitudes, and also had an tion, Ford et al. (2005a) have found that individual
indirect effect on information seeking perfor- items in the surface learning style (fear of failure
mance. Ford et al. (2001) found no relationships and poor time management) were linked to poor
between holist or analytic cognitive styles and retrieval. As with cognitive style, it appears that
retrieval effectiveness, but did find relationships learning style affects both the information seeking
among holists, imagers and Boolean searching process and its outcomes.
and among analytics, verbalizers and Best-match
searching (2005a; 2005b). It also appears that Cognitive Complexity
cognitive style effects are more important for
novice than for experienced Internet searchers Those who are cognitively simple tend to see the
(Palmquist & Kim, 2000). Such evidence suggests world in binary terms such as black and white or
that individual differences in cognitive style affect right and wrong, while those who are cognitively
not only the information seeking process but also complex are able to see shades of grey and rec-
its effectiveness. ognize that the validity of a given viewpoint may

94
Tips for Tracking Web Information Seeking Behavior

vary with circumstances. Higher levels of cogni- importance to an individual of Website character-
tive complexity appear to be associated with poor istics that imply an information-rich environment
retrieval (Ford et al., 2001). versus navigational aids that ease processing and
maximize efficiency (Hupfer & Detlor, 2007b).
Need for Cognition

Individuals with a high Need for Cognition (NFC) METHODS FOR COLLECTING
enjoy thinking and have a greater tendency to INDIVIDUAL DIFFERENCES, TASK,
elaborate upon, structure and evaluate information AND WEB TRACKING DATA
(Cacioppo, Petty & Kao, 1984). They engage in
more effortful decision making than those who are Recognizing the importance of both task and in-
low NFC and arrive at better information seeking dividual differences in Web information seeking
outcomes (Bailey, 1997). High NFC users also behavior, attention now turns to the methods that
have more favorable attitudes toward Websites allow for the collection of task, individual differ-
with complex verbal and simple visual elements ences, and information seeking behavior data in a
(Martin et al., 2005). NFC is positively correlated Web information seeking analysis. These methods
with Web information usage (Tuten & Bosnjak invariably involve the collection of Web tracking
2001) and has a direct impact on self-reported activity via the use of Web logs, but tracking alone
information seeking behavior (Das et al., 2003). is insufficient for a thorough information seek-
Finally, investigation of information seeking at ing analysis. Web tracking captures information
online grocery stores has found that high NFC seeking behavior with Web logfiles, but other data
shoppers, compared with low NFC consumers, collection instruments, such as questionnaires
investigated more URLs and spent more time and interviews, are required to collect individual
reading (Ho, 2005). differences and other task-related data.
In closed environments, like laboratories,
Self- and Other-Orientation researchers can control research participants’
information seeking tasks by giving them ex-
These characteristics describe differences in an plicit descriptions or instructions for their tasks
individual’s propensity to be concerned with one- and can ask them to complete surveys to collect
self versus others by tapping gender-related traits individual differences data. Closed environments
that pertain to an independent (Self-Orientation) also allow researchers to require that participants
versus interdependent (Other-Orientation) self- use specific software in the lab where their Web
concept orientation (Hupfer, 2001). Self- and activity will be tracked. Other advantages of the
Other-Orientation predict Internet use frequency closed environment include the opportunity to
and preferences that male-female indicators often modify the information seeking tools that are
fail to explain. The two scales interact to predict used (e.g., browsers, interfaces), control the avail-
how often individuals seek information online able functionality and even provide interactive
(Hupfer & Detlor, 2006) both for themselves (self- decision aids. However, closed environments are
relevant information) and for those close to them not without their drawbacks. Processing large
(other-relevant information). Other-Orientation numbers of participants through laboratory ses-
also is positively related to usage rates for Inter- sions requires considerable time and resources. In
net applications with relationship implications, addition, requiring subjects to conduct contrived
such as greeting cards (Hupfer & Detlor, 2007a). searches in an artificial setting or scenario may
Furthermore, the two scales interact to predict the compromise the validity and generalizability of

95
Tips for Tracking Web Information Seeking Behavior

any research results. and attitude data. To do this, we had to work closely
Richer Web information seeking analyses are with the portal development group to ensure that
more likely to be found in open environments the study’s data collection instruments (i.e., Web
where end-users can conduct real-life information tracking and online surveys) were incorporated
seeking tasks that are of relevance and importance directly within the portal’s design.
to them and that take place within natural environ- We believe that the use of a unique ID to link
ments and settings (e.g., the workplace, the home). data collected in the user surveys to the Web
In open environments, researchers are able to tracking metrics collected in the Web logs is a
observe natural behavior patterns and collect data key strength of our research project. By linking
that affords greater validity and generalizability. these data sources and triangulating results, we are
However, conducting Web information seeking able to arrive at a rich understanding of end-user
investigations in open environments presents online behavior. For example, regressions or path
its own challenges for the proper collection and analyses are being used to determine how well
analysis of task and individual differences data. individual differences predict task self-reports
For example, dynamic IP addressing prevents and actual usage behavior. Further, cluster or
the linking of a person’s Web tracking data with discriminant analysis techniques are being used
any individual differences data that is collected to establish the characteristics of low, medium
through user profiles or questionnaires. Knowing and high usage groups.
what task prompted a user to turn to the Web to As mentioned above, two types of Web surveys
seek information also is problematic in terms of were administered to people who consented to
understanding the type of task and its attributes. participate in the project. The first of these was an
There are various ways to go about capturing individual differences (user characteristics) survey
task, individual differences, and Web behavior that collected basic demographic information,
data, but a critical component is the ability to link personality traits, and technology background
all three types of data together for a specific indi- on each participant. Items were based on those
vidual. It probably is easiest to create this linkage found in a recent investigation by the authors
in a closed environment. However, if researchers (Hupfer & Detlor, 2006), the Georgia Institute of
and practitioners want to take advantage of the Technology’s annual GVU WWW User Surveys
benefits afforded by Web information seeking (cc.gatech.edu/gvu/user_surveys), and Ford and
analyses conducted in open environments, they Miller’s (1996) scales that measure perceptions
must devise a means of connecting these various of Web-based information seeking (see Appendix
data sources. We did this, quite successfully, in A for a list of the individual differences questions
our own research project at a municipal/commu- used in this project).
nity portal site called myhamilton.ca. The project The second type of survey was a brief user task
involved the use of two surveys (one pertaining to evaluation. Eliciting attitudes towards conducting
task; the other pertaining to individual differences) a specific online activity and perceptions of the
and the collection of Web tracking data to yield task’s importance, these surveys were adminis-
a robust understanding of Web user information tered after participants had completed a given
seeking behavior. Note that a selective subset of activity on the portal interface (see Appendix B for
task and individual differences variables from the the actual questions used to measure participants’
general model presented above were used in our attitudes toward and perceptions of tasks). These
myhamilton.ca research project. Importantly, a data were collected during short time windows
unique identification feature linked participants’ pertaining to a few specific portal activities, such
actual portal activity to demographic, personality as paying a fine or purchasing a dog license.

96
Tips for Tracking Web Information Seeking Behavior

In terms of Web tracking, like the two sur- on the client-side. These metrics included the fol-
veys, metrics were collected only from portal lowing: page attributes such as page views, page
users who had agreed to participate in the study. transitions, and HTTP referrer information; tem-
The portal was designed to facilitate metric col- poral attributes such as history time stamps, and
lection through third party applications hosted session times; and visitor attributes such as user
on the portal’s back-end Web servers. Specific identification tags, and remote host information.
detail concerning the type of Web metrics that Consequently, a composite of these primary Web
were collected and analyzed are discussed in the log metrics provided us with the desired analytics
following section. output related to information seeking behaviors
of end users. Table 1 shows the interrelationships
among the various Web log metrics that we used
WEB METRICS TO COLLECT and their sources, as well as the associations
AND TECHNIQUES FOR among the Web log metrics and their resulting
ANALYZING THEM composite analytics.
The advantage of using a customized server-
In general, the various sources of Web logs can side plug-in rather than other available logging
be classified as either server-level or client-level methods such as Web server logs, client-side track-
data sources (Srivastava et al., 2000). The primary ing utilities, and page tagging scripts is that the
metrics in our study were based on server-level overall process affords more control for updates
data generated through a custom programmed and modifications, and also entails less time and
server-side plug-in, and first-party cookies stored effort to clean the data and prepare it for further

Table 1. Summary of Web log sources, metrics, and composite analytics

Sources of Composite
Web Web Log Metrics Analytics
Log Information
Metrics Server Session Persistent Visitor Navigation
Seeking
Object Cookies Cookies Footprints Tracks
Trails
Page
• • • • •
Views
Page
• • •
Transitions
HTTP
Referrer Informa- • • •
tion
History
Time • • • • •
Stamps
Session
• • • •
Time
User
Identification • • • •
Tags
Remote
• • •
Host Information

97
Tips for Tracking Web Information Seeking Behavior

analysis (e-consultancy, 2003). Compared to Web (Sen et al., 2006). The formation of a Web data
server logs, custom logfiles suffer from fewer inac- warehouse enables simple decision support ser-
curacies and redundancies (e-consultancy, 2003; vices through channel traffic reports.
Murata & Saito, 2006). With respect to control, The second step in developing a Web analyt-
a server-side plug-in allows greater autonomy ics toolkit is to aid sophisticated visitor behavior
than a client-side remote tracking utility where tracking (Sen et al., 2006) which can only be
there is increased dependency on client platform enabled through additional structured statistical
capabilities and end user intentions (Winett, 1998). procedures and logic querying methods. In our
Similarly, vis-à-vis page tagging scripts where research study, we did this by following the three
the deployed solution is typically outsourced to phases suggested for Web usage mining studies,
an application service provider (Beasley, 2002; namely: 1) pre-processing, 2) pattern discovery,
e-consultancy, 2003), a server-side plug-in offers and 3) pattern analysis (Srivastava et al., 2000).
more control over the development and mainte- The pre-processing phase cleanses, sorts and
nance of an application interface. formats the raw data into organized segments
In our study, the server-side plug-in was of information (e.g., establishing sequence of
designed to poll several collections and proper- activities through sorting first by cookie-based
ties of the server-side objects including request, user identifiers and then by server session iden-
response, and session objects in order to retrieve tifiers). This information feeds into the pattern
values pertaining to attributes of visited site pages, discovery phase which converts raw logs into
times spent on each page and in each user session, data abstractions that are pivotal to the analysis
and unique visitor identification values stored in of usage patterns (e.g., deriving session length
client-side cookies. information from time stamps and page views).
In addition to using a customized server-side The pattern analysis phase calculates descriptive
plug-in for the collection of Web metrics data, our statistics and usage metrics that can help to identify
study utilized first-party cookies to track visitors. different user clusters based on their patterns of
This method is regarded as more reliable than information seeking. Overall, these three phases
using third-party cookies used by hosted analyt- enable the transformation of dimensions of Web
ics vendors as recent Internet statistics show that activities that are measurable into those that are
12%-17% of Internet users block third-party meaningful within the context of the analysis of
cookies while only 2%-5% block first-party cook- users’ information seeking behaviors.
ies (WebTrends, 2005). By storing automatically Unlike other Web analytics studies that under-
generated unique identifiers on users’ worksta- take the development of a self-contained proto-
tions, persistent cookies allow the identification typical Web usage mining system as part of the
of unique site visitors which can prove to be overall research project (Srivastava et al., 2000;
extremely valuable in determining the reach and Wu et al., 1998; Zaiane et al., 1998), the Web
audience penetration of a Website. analytics engine used in our study was based on a
To generate the composite metrics described selection of self-programmed application macros,
in Table 1, we created and utilized our own Web third party tools, and customized scripts.
analytics toolkit. Overall, the collection of singular Figure 2 shows a high level schematic of the
Web log metrics through server objects and cook- Web analytics engine depicting the various func-
ies, as shown in Table 1, facilitates the formation tional modules that were used to operationalize
of a Web data warehouse, which is regarded as the three phases of Web usage mining. While the
the first step in devising a Web analytics toolkit data cleanser module enabled the pre-processing

98
Tips for Tracking Web Information Seeking Behavior

Figure 2. Components of the Web analytics engine

phase, the URL filter and crawler, as well as the analysis widgets in the SPSS statistical applica-
metrics calculator, enabled the pattern discovery tion. The metrics calculator was used primarily
and pattern analysis phases. to report descriptive statistics and produce cluster
In terms of custom applications, server side analysis results that could be viewed in text, table
scripts using SQL (structured query language) or graphical formats through the metrics viewer
were deployed to extract data from tables in the data module.
warehouse and to export into a format suitable for The Web analytics engine allowed us to per-
spreadsheets and statistical analysis applications. form a composite analysis of Web log metrics from
In the data cleanser module, Excel macros were server-side and client-side sources. Specifically,
utilized to cleanse the data and organize it into as shown in Table 1, the Web analytics engine
meaningful segments that were to be used in met- allowed us to define a hierarchical view of user
rics calculations. The URL filter allowed us to form activities based on visitor footprints, navigation
clusters of Webpages based on their frequency and tracks, and information seeking trails.
mode of access. For instance, Website landing At the lowest level of user information seeking
pages were identified by noting external referrer pattern identification, a visitor footprint represents
Websites, and search tools were recognized by “a single clickstream record created by the interac-
query information such as keywords contained in tion of the visitor with a page on a Website” (Sen
URLs. Ultimately, a dictionary of landing pages et al., 2006). In establishing visitor footprints,
was compiled to facilitate page lookups during various clickstream metrics such as unique cookie
the analysis of online user activities. based user identification, server based session
The information derived from the URL filter identification, time stamp information, referrer
was further refined by mapping the URLs to the page URLs and destination page URLs can be
title of the Webpages. These titles were obtained compiled into simple records of online user ac-
by running the list of URLs through an Internet- tivities. In our study, the data cleanser and URL
based Web crawler utility which parses Webpages crawler modules in the analytics engine were
for various types of metadata. In this case, the only utilized to establish these visitor footprints. Figure
metadata that was of interest was the title tag per- 3 shows an extract of nine footprints pertaining
taining to the Webpages referenced by the URLs. to a sample user, User-34.
Finally, the metrics calculator module comprised An aggregation of visitor footprints enables
spreadsheet functions and macros in Excel and the formation of navigation tracks which provide

99
Tips for Tracking Web Information Seeking Behavior

Figure 3. Extract of visitor footprints

a chronological history of a user’s activities on the extract of navigation tracks pertaining to the
a Website (Sen et al., 2006). In our study, the visitor footprints shown in Figure 3.
configuration of each visitor navigation track Based on information from visitor footprints
comprised the entry point to the Website, the path and navigation tracks, information seeking trails
of Webpages that were traversed in each user ses- characterize the deepest level of pattern discovery
sion, along with the average time spent per page, employed in our study. Information seeking trails
content page requests per session, search queries can be discerned by using clustering algorithms
per session, and the exit point from the Website. which can group similar user beliefs, attitudes
Custom programmed spreadsheet macros in the and behaviors (Sen et al., 2006). In our study, the
metrics calculator module of the analytics engine information seeking trails were analyzed through
processed the visitor footprint information to composite analytics that acted as surrogate mea-
reveal these navigation tracks. Figure 4 shows sures to identify scanning, searching and browsing

Figure 4. Extract of visitor navigation tracks

100
Tips for Tracking Web Information Seeking Behavior

Figure 5. Extract of surrogate metrics and frequencies of archetypical information seeking modes

moves on Websites. These information seeking such as observing navigation tracks that consist of
moves were further assembled and classified into references to external search engines, high average
modes of information seeking such as undirected number of sessions with multiple search queries,
viewing, conditioned viewing, informal search, and low average time spent per Webpage.
and formal search (Aguilar, 1988; Choo et al., Formal search episodes are emblematic of us-
2000). Episodes of information seeking were ana- ers who make a purposeful and planned effort to
lyzed using recursive procedures in spreadsheet acquire specific information (Choo et al., 2000).
macros that hinged on multiple passes through Using information available in visitor navigation
visitor footprints and navigation tracks. Figure 5 tracks and other surrogate measures of information
shows an extract of the additional surrogate met- seeking modes, these episodes can be expected
rics that were used to formulate the archetypical to exhibit a high average number of sessions that
information seeking episodes. started with a search page, a low average number
Undirected viewing episodes can be identi- of page requests, and high average time spent
fied by observing the most common entry points per page.
in visitor navigation tracks and calculating the The next step in our research study will in-
average number of page requests and the average volve further examination of the four archetypical
time spent on Webpages. Since users engaged in episodes of information seeking outlined above.
undirected viewing modes demonstrate broad Specifically, we plan on linking the Web analyt-
scanning web moves (Choo et al., 2000), these ics information we collected in terms of visitor
modes typically consist of navigation tracks that footprints, navigation tracks and information
start at main landing pages such as the Website seeking trails to the data we collected through
homepages and sitemaps, and exhibit high average the individual differences questionnaire and
number of page requests and low average time task attitudes survey. Doing so will allow us to
spent per Webpage. better understand the associations between Web
Conditioned viewing episodes can be tallied behavior and the specific characteristics of users
by identifying similarities in entry and exit points or groups of users.
between visitor navigation tracks for the same
user. This can be accomplished by calculating
the dot product of page paths in visitor navigation ADDRESSING PRIVACY IN WEB
paths. Since users engaged in conditioned view- INFORMATION SEEKING ANALYSIS
ing episodes are interested in selected topics and
specific types of information (Choo et al., 2000), A final tip for any person wishing to conduct
these episodes also will typically exhibit a lower Web information seeking analysis in real-world
ratio of search queries to page requests. settings (i.e., non-laboratory environments) is
Informal search episodes show signs of un- the need to address privacy concerns during the
structured search efforts (Choo et al., 2000) and project’s initial stages and to design Web data
can be identified by using surrogate measures collection and analysis methods with privacy in

101
Tips for Tracking Web Information Seeking Behavior

mind. Ethics boards at academic institutions have restrict access. Academic opinion concerning the
always cautioned researchers about the need for data’s use has been divided; the very real privacy
anonymity and/or confidentiality. Academics, concerns have been acknowledged, but at the
industry analysts and privacy advocates also have same time, the availability of a very large and
raised concerns about the vast amount of data that current data set has immeasurable value for those
is collected using passive devices, such as adware, who investigate personalization and information
cookies, spyware and Web viruses, to record online retrieval (Hafner, 2006).
behavior (Marshall & Swartwout, 2006). Such With our own myhamilton.ca research project,
calls for privacy protection are warranted. we were very aware of the need to address privacy
For example, the need to preserve privacy with issues at the outset. In our initial discussions with
Web log analysis became a highly public debate the portal project team, we agreed that each user
in the summer of 2006 after America-On-Line would be assigned a unique ID that would allow us
(AOL) posted query log data (approximately to link tracking logs with survey data. Participant
20 million search inquiries obtained from over identities would not be released to us and the portal
650,000 users over a three month period) to a staff would handle the distribution of incentives
publicly accessible Website. These data, which to those who participated in our research. While
were intended for academic use, assigned each these measures ensured anonymity, we later were
user a unique ID and included the date and time confronted with the ramifications of municipal
of each query as well as addresses of Websites privacy legislation requirements. These concerns
that were visited after searching was concluded. were further exacerbated by a portal privacy breach
The AOL team that released the data intended to that resulted in the inadvertent broadcasting of
provide researchers with the opportunity to analyze people’s personal dog licensing information and
search patterns and strategies over time without the attendant negative press in the local media. The
having to disclose any personally identifiable ensuing privacy impact assessment conducted on
information. our research project required documentation of the
Unfortunately, it rapidly became apparent that flow and storage of both the survey and tracking
the specification of search parameters could in data. This documentation included the names of
some cases permit the identification of individuals all tables where data was stored, a description of
(Barbaro & Zeller, 2006; Hansell 2006a). These the programs that accessed and updated the data,
data were quickly removed from the site, but and the process by which the data would be sent
not before the data had been downloaded and to our research team. The city also planned to
circulated (Barbaro & Zeller, 2006). The ensuing delete all data once our research team confirmed
media furor informed readers about how much its safe receipt at the conclusion of the data col-
data is stored by major search engines (Zeller, Jr. lection period.
2006a), how advertisers used search history for The city’s privacy assessment identified one
segmentation and targeting purposes (Hansell, particularly important outstanding issue in that
2006b) and also advised them as to how they the city had no means to control or prevent the
could protect their identity online by using proxy possible misuse of the survey’s demographic
servers and by deleting browser cookies (Biers- information and Web tracking information, both
dorfer, 2006). In late August 2006, AOL dismissed of which resided in the city’s databases. This
a researcher and project manager, and their chief situation arose because participants’ unique IDs
technology officer resigned (Zeller, 2006b). AOL were stored within the city’s databases and thereby
also announced plans for the implementation of created the potential for unauthorized linking of
new technologies that would protect privacy and this information. Addressing this concern required

102
Tips for Tracking Web Information Seeking Behavior

extensive and costly program code and database launch for over six months. A better understand-
changes. Consequently, rather than storing par- ing of privacy legislation and its impact for our
ticipants’ unique IDs within the user profile table research would have allowed us to avoid both the
in the city’s database, unique IDs were embedded delay and the expense.
in session cookies that were transmitted between a
user’s computer and the City of Hamilton’s serv-
ers. Using session cookies that contained unique CONCLUSION
IDs allowed us to track user behavior individually
but eliminated the need to store a unique ID in To provide insight into methods for conducting a
a database resident on one of the city’s servers Web information seeking analysis, we presented
that possibly could have been used as a foreign several tips. First, we raised awareness of the
key to access and link together a person’s private importance of individual differences and tasks in
information. understanding information seeking behavior. We
Elimination of unique IDs within the user encourage practitioners and researchers to include
profile table also caused us to rethink the way the collection and analysis of task, individual dif-
we handled the distribution of participant incen- ferences, and behavioral data in any Web informa-
tives. Some mechanism was required to identify tion seeking analysis design. The second tip spoke
which participants had participated in the study to devising methods that collect task, individual
so that we could contact them and distribute gift differences, and Web tracking data and provide a
certificates. To facilitate this, a field was created means to link these data sets together. We provided
in the user profile table to simply act as a flag that our own research project at myhamilton.ca as an
would indicate whether or not a particular user example for others to follow in this regard. The
had agreed to participate in our research project. two surveys (see Appendices A and B) may help
This flag allowed us to identify those people who others structure similar research instruments to
participated in the study but it did not allow us collect data on tasks and individual differences.
to associate a person’s research data with their The third and fourth tips provided guidelines on
personally identifying (contact) information. which Web metrics to collect and how to go about
These changes to our data collection methods analyzing them. Though many alternative Web
required an additional review of our protocol metrics and methods of analysis exist, we antici-
by the McMaster research ethics board and the pate that our description of the Web metrics we
amendment of our call for participation and con- collected and our illustration of their analysis will
sent form in order to clarify privacy implications. assist others in their own investigations. Finally,
Specifically, potential participants were informed we discussed the need to address privacy concerns
that privacy had been built into the methods by right up-front in the collection of Web tracking
which data would be collected and stored. In terms information so as to avoid lengthy and costly
of databases, the unique ID would be stored only delays in conducting a Web information seeking
in three raw data tables that contained the Web analysis. By paying attention to these five tips,
tracking and survey data. No link would exist both academic researchers and practitioners can
between these tables and any other contained in ensure that their Website performance measure-
the myhamilton.ca database, and no personally ment initiatives run smoothly.
identifiable information would be sent to the Mc-
Master research team. All of these revisions, the
additional ethics review, and further testing of our
data collection instruments delayed the project’s

103
Tips for Tracking Web Information Seeking Behavior

REFERENCES Das, S., Echambadi, R., McCardle, M., &


Luckett, M. (2003). The effect of interpersonal
Aguilar, F. J. (1988). General managers in action. trust, need for cognition, and social loneliness
New York: Oxford University Press. on shopping, information seeking and surfing
Bailey, J. R. (1997). Need for cognition and on the Web. Marketing Letters, 14(3), 185–202.
response mode in the active construction of doi:10.1023/A:1027448801656
an information domain. Journal of Economic Dillon, A., & Watson, C. (1996). User analysis
Psychology, 18(1), 69–85. doi:10.1016/S0167- HCI. The historical lessons from individual differ-
4870(96)00047-5 ences research. International Journal of Human-
Barbaro, M., & Zeller, T., Jr. (2006). A face is Computer Studies, 45(6), 619–638. doi:10.1006/
exposed for AOL searcher No. 4417749. The New ijhc.1996.0071
York Times, August 9, 1. e-consultancy. (2003). Web measurement and
Beasley, C. (2002). It’s a hit! Gauging success analytics. London, UK: e-consultancy.
through traffic analysis, Retrieved May 7, 2007 Ekstrom, R.B, French, J.W., & Harman, H.H.
from http://www.sitepoint.com/article/success- (1976). Manual for kit of factor-referenced cogni-
traffic-analysis tive tests, 109-113, 173-177.
Bhatnagar, A., & Ghose, S. (2004). Online infor- Entwistle, H., & Tait, N. J. (1995). The revised
mation search termination patterns across product approaches to studying inventory. Edinburgh:
categories and consumer demographics. Journal Centre for Research on Learning and Instruction,
of Retailing, 80(3), 221–228. doi:10.1016/j.jre- University of Edinburgh.
tai.2003.10.001
Fisher, K. E., Erdelez, S., & McKechnie, L. E. F.
Biersdorffer, J. D. (2006). How to digitally hide (Eds.). (2005). Theories of information behavior.
(somewhat) in plain sight. The New York Times, Medford, NJ: Information Today.
August 12, 9.
Ford, N. (1986). Psychological determinants of in-
Byström, K., & Järvelin, K. (1995). Task complex- formation needs: A small-scale study of higher ed-
ity affects information seeking and use. Informa- ucation students. Journal of Librarianship, 18(1),
tion Processing & Management, 31(2), 191–213. 47–61. doi:10.1177/096100068601800103
doi:10.1016/0306-4573(94)00041-Z
Ford, N., & Miller, D. (1996). Gender differences
Cacioppo, J. T., Petty, R. E., & Kao, C. F. (1984). in internet perception and use. In Papers from the
The efficient assessment of need for cognition. 3rd Electronic Library and Visual Information
Journal of Personality Assessment, 48, 306–307. Research (ELVIRA) conference, April 30 (pp.
doi:10.1207/s15327752jpa4803_13 87-202). London: ASLIB.
Choo, C. W., Detlor, B., & Turnbull, D. (2000). Ford, N., Miller, D., & Moss, N. (2001). The role
Web work: Information seeking and knowledge of individual differences in internet searching: An
work on the world wide web. Netherlands: Kluwer empirical study. Journal of the American Society
Academic Publishers. for Information Science and Technology, 52(12),
1049–1066. doi:10.1002/asi.1165

104
Tips for Tracking Web Information Seeking Behavior

Ford, N., Miller, D., & Moss, N. (2003). Web Ho, S. Y. (2005). An exploratory study of using
search strategies and approaches to studying. a user remote tracker to examine web users’ per-
Journal of the American Society for Informa- sonality traits. ICEC, 05(August), 15–17.
tion Science and Technology, 54(6), 473–489.
Huneke, M. E., Cole, C., & Levin, I. P. (2004).
doi:10.1002/asi.10233
How varying levels of knowledge and motivation
Ford, N., Miller, D., & Moss, N. (2005a). Web affect search and confidence during consideration
search strategies and human individual differ- and choice. Marketing Letters, 15(2&3), 67–79.
ences: Cognitive and demographic factors, internet doi:10.1023/B:MARK.0000047385.01483.19
attitudes, and approaches. Journal of the American
Hupfer, M. E. (2001). Self-Concept orientation
Society for Information Science and Technology,
and response to agentic and communal adver-
56(7), 741–756. doi:10.1002/asi.20168
tising messages. Unpublished Doctoral Thesis,
Ford, N., Miller, D., & Moss, N. (2005b). Web University of Alberta, Edmonton.
search strategies and human individual differenc-
Hupfer, M. E., & Detlor, B. (2006). Gender and
es: A combined analysis. Journal of the American
web information seeking: A self-concept orien-
Society for Information Science and Technology,
tation model. Journal of the American Society
56(7), 757–764. doi:10.1002/asi.20173
for Information Science and Technology, 57(8),
Gugerty, L., Treadaway C., & Rubinstein, J.S. 1105–1115. doi:10.1002/asi.20379
(2006). Individual differences in internet search
Hupfer, M. E., & Detlor, B. (2007a). Beyond
outcomes and processes. CHI 2006, April, 815-
gender differences: Self-concept orientation and
820.
relationship-building applications on the Internet.
Hafner, K. (2006). Researchers yearn to use AOL Journal of Business Research, 60(6), 613–619.
logs, but they hesitate. The New York Times, doi:10.1016/j.jbusres.2006.06.010
August 23, 1.
Hupfer, M. E., & Detlor, B. (2007b). Sex, gender
Hansell, S. (2006a). AOL removes search data and self-concept: Predicting web shopping site
on vast group of web users. The New York Times, design preferences. In Detlor, B., Hassanein, K.,
August 8, 4. & Head M. (Eds.), Proceedings of the 8th World
Congress on the Management of E-Business (WC-
Hansell, S. (2006b). Advertisers trace paths users
MeB), July 11-13, Toronto, Canada. Los Alamitos,
leave on internet. The New York Times, August
California: IEEE Computer Society.
15, 1.
Marshall, K. P., & Swartwout, N. (2006). Market-
Häubl, G., & Trifts, V. (2000). Consumer decision
ing and internet professionals’ fiduciary respon-
making in online shopping environments: The ef-
sibility: A perspective on spyware. Journal of
fect of interactive decision aids. Marketing Science,
Internet Commerce, 5(3), 109–128. doi:10.1300/
19(1), 4–21. doi:10.1287/mksc.19.1.4.15178
J179v05n03_05
Heinström, J. (2005). Fast surfing, broad scanning
Martin, B. A. S., Sherrard, M. J., & Wentzel, D.
and deep diving: The influence of personality and
(2005). The role of sensation seeking and need
study approach on students’ information-seeking
for cognition on web-site evaluations: A resource-
behavior. The Journal of Documentation, 61(2),
matching perspective. Psychology and Marketing,
228–247. doi:10.1108/00220410510585205
22(2), 109–126. doi:10.1002/mar.20050

105
Tips for Tracking Web Information Seeking Behavior

Moe, W. W. (2003). Buying, searching or brows- Srivastava, J., Cooley, R., Deshpande, M., &
ing: differentiating between online shoppers us- Tan, P. N. (2000). Web usage mining: Discovery
ing in-store navigational clickstream. Journal of and applications of usage patterns from web
Consumer Psychology, 13(1&2), 29–40. data. ACM SIGKDD Explorations, 1(2), 12–23.
doi:10.1145/846183.846188
Moorthy, S., Ratchford, B. T., & Talukdar, D.
(1997). Consumer information search revis- Toms, E., & Trifts, V. (2006). When limited
ited: Theory and empirical analysis. The Jour- cognitive resources meet unlimited information
nal of Consumer Research, 23(4), 263–277. sources: Factors influencing consumers’allocation
doi:10.1086/209482 of web-based search effort. Proceedings of the 7th
World Congress on the Management of eBusiness,
Murata, T., & Saito, K. (2006). Extracting users’
July, Halifax, Canada.
interests from web log data. Proceedings of the
IEEE/WIC/ACM International Conference on Tuten, T. L., & Bosnjak, M. (2001). Understand-
Web Intelligence (WI’06). ing differences in web usage: The role of need for
cognition and the five-factor model of personality.
Palmquist, R. A., & Kim, K. S. (2000). Cogni-
Social Behavior and Personality, 29(4), 391–398.
tive style and on-line database search experi-
doi:10.2224/sbp.2001.29.4.391
ence as predictors of web search performance.
Journal of the American Society for Information Vakkari, P. (1999). Task complexity, problem
Science American Society for Information Sci- structure and information actions: Integrating
ence, 51(6), 558–566. doi:10.1002/(SICI)1097- studies on information seeking and retrieval.
4571(2000)51:6<558::AID-ASI7>3.0.CO;2-9 Information Processing & Management, 35(6),
819–837. doi:10.1016/S0306-4573(99)00028-X
Payne, J. W., Bettman, J. R., & Johnson, E. J.
(1993), The adaptive decision maker. Cambridge Wang, P., Hawk, W. B., & Tenopir, C. (2000). Us-
UK: Cambridge University Press. ers’ interaction with world wide web resources: An
exploratory study using a holistic approach. Infor-
Riding, R. J., & Cheema, I. (1991). Cog-
mation Processing & Management, 36, 229–251.
nitive styles – an overview and integra-
doi:10.1016/S0306-4573(99)00059-X
tion. Educational Psychology, 11, 193–215.
doi:10.1080/0144341910110301 WebTrends. (2005). Best practices for accurate
web analytics: Avoiding third-party cookie rejec-
Roy, M., & Chi, M. T. H. (2003). Gender differ-
tion and deletion. Portland, OR: WebTrends
ences in patterns of searching the web. Journal
of Educational Computing Research, 29(3), 335– Wildemuth, B., & Hughes, A. (2005). Perspec-
348. doi:10.2190/7BR8-VXA0-07A7-8AVN tives on the tasks in which information behaviors
are embedded. In Fisher, K.E., Erdelez, S. &
Saracevic, T. (1991). Individual differences in
McKechnie, L. (Eds.), Theories of information
organizing, searching and retrieving information.
behavior. Medford, NJ: Information Today, for
Proceedings of the American Society for Informa-
the American Society for Information Science &
tion Science, 28, 82–86.
Technology, 275-279.
Sen, A., Dacin, P. A., & Pattichis, C. (2006).
Wilson, T. D. (1999). Models in informa-
Current Trends in Web Data Analysis. Com-
tion behaviour research. The Journal of
munications of the ACM, 49(11), 85–91.
Documentation, 55(3), 249–270. doi:10.1108/
doi:10.1145/1167838.1167842
EUM0000000007145

106
Tips for Tracking Web Information Seeking Behavior

Wilson, T. D., & Walsh, C. (1996), Informa- Task: In this chapter, refers to the informa-
tion behavior: An interdisciplinary perspective. tion seeking task an individual user experiences
Sheffield: University of Sheffield Department of that instills a need for information and motivates
Information Studies. the user to satisfy this information need through
some sort of information seeking behavior. Task
Winett, B. (1998). Tracking your visitors. Re-
is the context surrounding a person’s informa-
trieved May 10, 2007, from WebMonkey Tracking
tion need.
Tutorial, http://www.webmonkey.com/e-business/
tracking/tutorials/tutorial2.html.
Web Information Seeking Behavior: Refers
Wu, K.-l., Yu, P. S., & Ballman, A. (1998). Speed- to information seeking behaviors that occur over
tracer: A web usage mining and analysis tool. IBM the Web. Choo, Detlor & Turnbull (2000) identify
Systems Journal, 37(1), 89–105. four main modes of information seeking on the
Web ranging from wayward browsing to goal-
Zaiane, O. R., Xin, M., & Han, J. (1998). Discov-
directed search (undirected viewing, conditioned
ering web access patterns and trends by applying
viewing, informal search, and formal search)
olap and data mining technology on web logs.
where each mode is characterized by predominant
Proceedings of the IEEE Forum on Reasearch
information seeking moves or activities (undi-
and Technology Advances in Digital Libraries,
rected viewing: starting and chaining; conditioned
Santa Barbara, CA.
viewing: browsing and differentiating; informal
Zeller, T., Jr. (2006a). Privacy vs. viewing the in- search: differentiating, monitoring, and extracting;
ternet user as a commodity. The New York Times, and formal search: monitoring and extracting).
August 12, 1.
Web Metrics: Pertains to the measures by
Zeller, T., Jr. (2006b). AOL acts on release of data.
which to assess a person’s Web information seek-
The New York Times, August 22, 1.
ing behavior or to assess and monitor activity
on a Website. Examples of commonly used Web
metrics include page views, page transitions, and
Key Terms session times.

Individual Differences: The demographic Web Tracking: Refers to the automated col-
and psychological characteristics of people that lection of Web information seeking behavioral
distinguish one person from another. data.

Information Seeking Behavior: Refers to


how people seek information in different contexts
(Fisher, Erdelez & McKechnie, 2005).

107
Tips for Tracking Web Information Seeking Behavior

APPENDIX A: INDIVIDUAL DIFFERENCES QUESTIONNAIRE

Note that some of these questions and their responses pertain to the City of Hamilton and a Canadian
context. Modifications would be necessary if administered to a different sample population.

Age

How old are you?

Under 18
18-24
25-29
30-34
35-39
40-44
44-49
50-54
55-59
60-64
65+
Prefer not to say

Sex

Are you?

Male
Female
Prefer not to say

Race

How would you classify yourself?

Aboriginal
Asian
Black
East Indian
Hispanic/Latino
Middle Eastern
Multi-Racial

108
Tips for Tracking Web Information Seeking Behavior

West Indian
White
Other
Prefer not to say

Kind of area you live in

Which of the following best describes the area you live in?

Urban
Suburban
Rural
Prefer not to say

Location in Hamilton

Which geographical area of Hamilton best categorizes the area you live in?

Hamilton East
Hamilton Central
Hamilton West
Dundas
Ancaster
Flamborough
Stoney Creek
Mountain West
Mountain East
Mountain Central
Other
Prefer not to say

Language

At a general conversational level, which of the following languages do you speak? Indicate all that
apply.

English
French
Arabic
Bengali
Cantonese
German
Hindi/Urdu

109
Tips for Tracking Web Information Seeking Behavior

Indonesian
Italian
Japanese
Mandarin
Polish
Portuguese
Russian
Spanish
Other
Prefer not to say

Marital Status

What is your current marital status?

Single (never married)


Co-habiting
Married
Separated/Divorced
Widow/Widower
Prefer not to say

Education

What is your highest education level obtained?

Primary School
Secondary (High) School
Vocational/Technical/Community College
University Undergraduate (bachelors degree)
University Graduate (masters/professional/doctoral degree)
Prefer not to say

Occupation

Your current primary employment status is

Employed
Unemployed
Student
Retired
Other
Prefer not to say

110
Tips for Tracking Web Information Seeking Behavior

Occupation (cont’d)

Which of the following categories best describes the industry you primarily work in (regardless of your
actual position)?

Agriculture
Arts, Entertainment, and Recreation
Broadcasting
Construction
Fishing & Hunting
Education
College, University, and Adult Education
Primary/Secondary (K-12) Education
Other Education Industry
Finance and Insurance
Forestry,
Government and Public Administration
Health Care and Social Assistance
Homemaker
Hotel and Food Service
Information Industry
Information Services and Data Processing
Other Information Industry
Legal Services
Manufacturing
Computer and Electronics Manufacturing
Other Manufacturing
Military
Mining
Publishing
Software
Telecommunications
Transportation and Warehousing
Real Estate, Rental and Leasing
Religious
Retail
Scientific or Technical Services
Utilities
Wholesale
Other Industry
Prefer not to say

111
Tips for Tracking Web Information Seeking Behavior

Which of the following best describes your role in industry?

Administrative Staff
Consultant
Junior Management
Middle Management
Researcher
Self-employed/Partner
Skilled Laborer
Student
Support Staff
Temporary Employee
Trained Professional
Upper Management
Other
Prefer not to say

The organization you work for is in which of the following:

Public sector (e.g., government)


Not-for-profit sector
Other
Private sector (e.g. most businesses and individuals)
Don’t know
Prefer not to say

Household Composition

What is the number of adults 18 and over living in your household?

0
1
2
3+
Prefer not not say

What is the number of children age 4 and under living in your household?:

0
1
2
3+
Prefer not not say

112
Tips for Tracking Web Information Seeking Behavior

What is the number of children ages 5 to 9 living in your household?:

0
1
2
3+
Prefer not not say

What is the number of children ages 10 to 13 living in your household?:

0
1
2
3+
Prefer not not say

What is the number of children ages 14 to 17 living in your household?:

0
1
2
3+
Prefer not not say

Household Income

What is your household income before taxes?

Less than $20,000


$20,000-$39,999
$40,000-$59,000
$60,000-$79,999
$80,000-$99,999
$100,000-$119,999
$120,000-$139,999
$140,000-$159,999
$160,000-$179,999
$180,000-$199,999
More than $200,000
Prefer not to say

113
Tips for Tracking Web Information Seeking Behavior

Years on the Internet

How long have you been using the Internet (including using email, gopher, ftp, etc.)?

Less than 6 months


6 to 12 months
1 to 3 years
4 to 6 years
7 years or more
Prefer not to say

Community Building

Complete the following sentence in the way that comes closest to your own views: ‘Since getting on
the Internet, I have...’

Become MORE connected with people like me


Become LESS connected with people like me
Become EQUALLY connected with people like me
Don’t know
Prefer not to say

Self- versus Other-Orientation(taken fromHupfer, 2001)

Rate each item below according to how well you think these statements describe you.

(Utilize a 9-point scale: “never true of me” 1 2 3 4 5 6 7 8 9 “always true of me”; 0 = “Prefer not to say”):

I am a nurturing person
I am a self-sufficient person
I am understanding
I make my own choices
I am a compassionate person
I am my own person
I am self-reliant
I am sympathetic
I am sensitive to the needs of others
I am an independent person

114
Tips for Tracking Web Information Seeking Behavior

Computer Proficiency

Indicate the extent to which you agree with the following statements

(Utilize a 7-point scale of “strongly disagree 1 2 3 4 5 6 7 strongly agree; 0 = “Prefer not to say”):

I am highly competent at...

... creating and editing documents in a word processor


... creating and maintaining electronic spreadsheets of data
... creating and maintaining data tables & records in a database progra
... sending and receiving email messages
... searching for information utilizing a Web search engine (like ‘Google’)

Web Skill Test

Which of the following have you done? (Check all that apply)

• Ordered a product / service from a business, government or educational entity by filling out a form
on the web
• Made a purchase online for more than $100
• Created a Webpage
• Customized a Webpage for yourself (e.g. MyYahoo, CNN Custom News)
• Changed your browser’s “startup” or “home” page
• Changed your “cookie” preferences
• Participated in an online chat or discussion (not including email)
• Listened to a radio broadcast online
• Made a telephone call online
• Used a nationwide online directory to find an address or telephone number
• Taken a seminar or class about the Web or Internet
• Bought a book to learn more about the Web or Internet
• Did Internet banking
• Prefer not to say

Technology Comfort

How comfortable do you feel using computers, in general?

Very comfortable
Somewhat comfortable
Neither comfortable nor uncomfortable
Somewhat uncomfortable
Very uncomfortable
Prefer not to say

115
Tips for Tracking Web Information Seeking Behavior

How comfortable do you feel using the Internet?

Very comfortable
Somewhat comfortable
Neither comfortable nor uncomfortable
Somewhat uncomfortable
Very uncomfortable
Prefer not to say

How satisfied are you with your current skills for using the Internet?

Very satisfied - I can do everything that I want to do


Somewhat satisfied - I can do most things I want to do
Neither satisfied nor unsatisfied
Unsatisfied - I can’t do many things I would like to do
Very unsatisfied - I can’t do most things I would like to do
Prefer not to say

Frequency of Accessing the Web from Different Locations

From home (including a home office)

Daily
Weekly
Monthly
Less than once a month
Never
Prefer not to say

From work:

Daily
Weekly
Monthly
Less than once a month
Never
Prefer not to say

From school:

Daily
Weekly
Monthly
Less than once a month
Never
Prefer not to say
116
Tips for Tracking Web Information Seeking Behavior

From public terminals (e.g., a library terminal, public kiosk):

Daily
Weekly
Monthly
Less than once a month
Never
Prefer not to say

From other places:

Daily
Weekly
Monthly
Less than once a month
Never
Prefer not to say

Connection to the Internet

Which of the following connection speeds do you primarily use to connect to the Internet? (Round to
the closest value if necessary.) If you access the Internet at home via a commercial provider, choose the
speed from you to your Internet provider:

Regular dial-up (through your phone company)


DSL low-speed (through your phone company)
DSL high-speed (through your phone company)
Cable (through your cable provider)
Do not know
Prefer not to say

Number of personal computers

How many personal computers are in your household (including laptops, but not including electronic
organizers)?:

0
1
2
3 or more
Prefer not not say

Perceptions of Web-based Information Seeking(taken fromFord & Miller, 1996)

Rate each item below according to how well you think these statements describe you.
117
Tips for Tracking Web Information Seeking Behavior

(Utilize a 5-point scale: “strongly agree” 1 2 3 4 5 “strongly disagree”; 0 = “Prefer not to say”)

• I usually only look at things on the Internet that have been suggested to me.
• Despite its complexity, I generally manage to find my way around the Internet fairly effectively.
• I rarely find anything useful on the Internet.
• I usually manage to keep ‘on target’ and avoid too much irrelevant material when using the
Internet.
• I’m prepared to plough through quite a lot of irrelevant information in case there’s something use-
ful I might otherwise miss on the Internet.
• If I had to choose only one, I’d prefer keyword searching to browsing (hypertext) on the
Internet.
• The Internet is too unstructured for my liking.
• I personally think that the graphical elements of the World Wide Web (i.e., pictures, icons, graph-
ics, etc. as opposed to just text) make me much more likely to use the Internet than if it were just
text-based.
• When I use the Internet, I feel as though I’m not as ‘in control’ as I would like.
• My advice to someone like me would be: The best way to learn to use the Internet is to explore
everything broadly to get a comparative ‘feel’ of the various aspects/tools before getting down to
mastering one in any depth.
• I tend to get lost when using the Internet.
• It’s best to use the Internet only when you have a well-defined plan (rather than just browsing
around).

APPENDIX B: THE TASK SURVEY

Perception of task importance

For you personally,


1. How important is it to do this task online?
“not important at all” 1 2 3 4 5 6 7 8 9 “very important”
2. How useful is it to do this task online?
“not useful at all” 1 2 3 4 5 6 7 8 9 “very useful”
3. How critical is it do this task online?
“not critical at all” 1 2 3 4 5 6 7 8 9 “very critical”

Attitude towards a particular task


4. I have positive feelings towards doing this task online.
“strongly disagree” 1 2 3 4 5 6 7 8 9 “strongly agree”
5. The thought of doing this task online is appealing to me.
“strongly disagree” 1 2 3 4 5 6 7 8 9 “strongly agree”
6. It is a good idea to be able to do that task online.
“strongly disagree” 1 2 3 4 5 6 7 8 9 “strongly agree”
This work was previously published in Handbook of Research on Web Log Analysis, edited by B. Jansen; A. Spink; I. Taksa,
pp. 256-283, copyright 2009 by Information Science Reference (an imprint of IGI Global).

118
119

Chapter 1.8
A Proposed Template for
the Evaluation of Web
Design Strategies
Dimitrios Xanthidis
Ciber, University College London, UK, DEI College, Greece, & NYC College, Greece

David Nicholas
Ciber, University College London, UK

Paris Argyrides
University College London, UK

Abstract seems that all these differences could boil down to a


systematic approach, here called evaluation template,
This chapter is the result of a two years effort to of 53 points that the design strategies of the Web sites
design a template aiming at standardizing, as much should be checked against. This template was tested
as such a task is feasible, the evaluation of Web sites. on a significant number (232) of Web sites of Greek
It is the product of a few publications in international companies and proved it can be used to evaluate the
conferences and journals. A thorough review of the quality of Web sites not only by technology experts
international literature on the subject led the authors but by non-experts alike. The evaluation template,
to conclude there is a very large number of opinions, suggested here, is by no means the solution to the
thoughts and criteria from different professionals problem of standardizing the process of evaluating
involved, directly or indirectly, with the process of a Web site but looking at other work done on the
designing a good Web site. To make matters even subject worldwide it is a step ahead.
more complicated there are a number of different
terms used by various scholars, scientists and profes-
sionals around the world that often refer to similar, INTRODUCTION
if not the same, attributes of a Web site. However, it
Despite the fact that in many developing countries
internet access and e-commerce was not introduced
DOI: 10.4018/978-1-60566-100-1.ch014

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
A Proposed Template for the Evaluation of Web Design Strategies

until very recently (Xanthidis and Nicholas, 2004), discussion that this template has be subject to in
the world is, clearly, moving towards the digital three international conferences leads the authors
platform (Figure 1 – Greece in the red circle) that it is, indeed, a small step ahead toward the
with a rapidly increasing number of companies solution to the problem.
in these countries hosting Web sites. The question
is whether these Web sites meet the international
standards concerning a site’s functionality. A BACKGROUND
preliminary research of large on- or off-line
universities’ libraries and governmental Web In a highly competitive environment, such as the
sites revealed that there are a number of different Web, with billions of sites online and thousands
professions involved in the process of designing a added every day (D-Lib Magazine, 2003) design-
good Web site, e.g. managers, marketing people, ers/developers consider important that Web sites
information technology experts, lawyers, ethnolo- are attractive and inspire trust to the users so as to
gists, all having different opinions regarding the cause them to revisit. Therefore, once decisions
functionality and appearance of a good Web site. have been made as of what features and function-
However, it also proved it is possible to design a alities the e-commerce solution should include
template that incorporates the main points of all the technology experts come into play in order to
these different views. realize the solution into a good Web site. Based
A critical parameter affecting the way people on extensive review of the international literature
interact with a Web site is its structure, overall four categories of features and functionality are
design and layout. A poorly designed Web site considered key for the successful implementation
might lead to accessibility problems as well as of an e-commerce strategy: user-interface design,
reduced interest to navigate through it causing globalization and customization, accessibility and
reluctance to visit the site again. Even though availability, security and privacy. This section
professional firms design many Web sites, there is, identifies, describes and analyzes the issues that
still, a substantial amount of work done by people comprise each category.
with limited knowledge on how a well-designed
Web site should be regardless of the fact these Stickiness
technology experts have the know-how to build
any e-commerce solution, from a simple to a very In the internet, billions of Web sites exist and even
sophisticated one. thousands are added every day (Trends…, 2003;
The main problem in most cases where an Faster…, 2002). Designers and developers con-
evaluation of a Web site is required is the lack of sider it important that their Web sites be attractive
certain systematic methodology to follow. This and inspire trust in users. This is reflected by the
is exactly what this chapter is all about. A simple amount of time repeatedly spent visiting a site;
yet comprehensive, straightforward yet seen from a practice known as stickiness, a combination of
four different angles, approach of how a Web site “content, usability and personalization” and other
could be evaluated based on a number of criteria issues, each one to be evaluated and measured on
gathered from many different sources, scholars, its own merit (Stickiness…, 2000).
scientists and academics, worldwide. Noone can Arguably the most important element of a
claim to have found the solution to a problem Web site’s attractiveness is its graphical user in-
so complex that it involves a variety of different terface (GUI). The user interface should contain
professions not limited to the information and a mixture of graphics and text that could make
telecommunications technology. However, the it “appropriate” and “appealing” to any visitor

120
A Proposed Template for the Evaluation of Web Design Strategies

Figure 1. Global Internet penetration 2006

(Sutcliffe, 2002). Furthermore, Nielsen (1999) hyperlinks difficult to be seen (Zhang et al., 2000;
supports the assumption that the actual Web site Iowa, 1999). Several techniques are available to
content is among its most important attributes. avoid such a negative aspect but two are the most
This can be accomplished by avoiding annoying popular among Web developers. The implementa-
and distracting elements i.e. banners, marquees, tion of floating hyperlinks, i.e. hyperlinks that are
graphics that “overshadow text”, also the “overuse programmed in such a way as to “follow” (floating
of animations”, or unusual designs which do not on screen) the active/visible part of the particular
follow certain accepted patterns of Web site design, Web page, is one of them. Another way is to limit
etc (California, 2002; Interface…, 2002). the content of a Web page to that which can fit in
Another important guideline to follow when one screen shot.
implementing a Web site is to avoid using scroll- Regarding hyperlinks the Web developers
ing mechanisms (Interface…, 2002). The reason should also be concerned with their placement
is it has been proven that the user, usually, ignores and style. They should be easily accessible at a

121
A Proposed Template for the Evaluation of Web Design Strategies

glance, and readable with their font preferences those that belong to large organization, incorporate
and their style not causing stress to the user when this feature as well. Its advantage is that it radically
trying to distinguish them from other content. In reduces the time needed to find the things that one
case icons are used instead of text as hyperlinks, is looking for in the site adding positive points to
these should be, intuitively identifiable at least by the company’s overall e-commerce strategy.
information and communications experts, business A critical element, often overlooked, of the
people and marketing professional, if not everyone quality of a Web site is the quality of information it
(Zhang et al., 2000; California, 2002). These are provides to its visitors. The information displayed
simple human-computer interaction guidelines should be extracted from the most accredited and
that apply in the case of Web sites as well. accurate source (Zhang et al., 2000). Addition-
One of the essential requirements for every ally, it should be updated regularly to eliminate
Web site is to avoid having undefined objects as the possibility of any outdated content, product
the targets of any of its hyperlinks, meaning that specifications or “past-due elements” (Katerat-
none of the hyperlinks should lead to a missing tanakul and Siau, 1999). Otherwise, failure to
or dead link. Also the content of the target object comply with the aforementioned criteria will cause
of the hyperlinks should be relevant to the Web a direct impact on the consumer’s behavior towards
site’s content (Helm, 2001; Internet…, 2002; the corporation i.e. misbelief of the information
California, 2002). The opposite could seriously presented (Katerattanakul and Siau, 1999).
damage the image of the company the Web site Finally, a very easily implemented feature
belongs as it would cause the visitor’s lack of trust with surprising positive influences on the visitors’
to the company severely reducing the chances to behavior is any mechanism that would allow for
succeed in the e-commerce arena. visitor feedback. A Web site should allow visi-
A very useful feature of every Web site is the tors to comment on its strengths and weaknesses
implementaion of a site map of some sort. The (Zhang et al., 2000) as well as provide a two-way
reason for that is every user would like to know communication channel (Katerattanakul and Siau,
their location at any depth in the Web site they are 1999), by means of online surveys, e-mail link,
surfing. For this purpose it would be useful if such feedback forms (Interface…, 2002; Evaluating…,
mechanisms as site tree diagrams, composed of 2004). It would help the visitor feel that it’s ideas,
text or icon hyperlinks, were included displaying comments, remarks are important and taken seri-
the user’s location (Roy et al., 2001; California, ously under consideration.
2002). Not doing so will most likely cause time
loss and tiredness from the part of visitor and, Customization and Globalization
consequently, reluctance to revisit the site.
Another quite handy element of a Web site A Web site’s success depends not only on features
implemented all too often recently is the internal such as the aforementioned, but also on its global
search engine. It is strongly recommended by (or local) perspective, meaning its attractive-
technology experts that any Web site that includes ness and usefulness to populations in different
articles or any type of information of significant geographic regions, a feature often referred to
size and/or value should either have this feature as globalization. The reason is people in differ-
implemented or at least contain hyperlinks to the ent parts of the globe behave in different ways
complete text if available (Nielsen, 1999). This to various stimulations triggered by a Web site
trend was followed in the past by online news- because they live in different cultures, practice
groups, journals and the like (Zhang et al., 2000) different religions and communicate using vari-
but the newly developed Web sites, especially ous languages and symbols (Hanrahan and Kwok,

122
A Proposed Template for the Evaluation of Web Design Strategies

2001). Furthermore, it would be quite helpful to a country’s government. Finally, online ordering
clarify, through the Web site, legal and/or customs forms should include a “universal address” format
particularities that might affect a visitor’s engage- i.e. instead of using the U.S. term “zip codes” use
ment in e-commerce activity when using the Web the universal “postal codes” (Housley, 2004), and
site of a company based in another country than the allow for the selection of various countries and
one the individual resides. This factor could cause their corresponding addresses.
positive first impressions to some populations or On the other hand, the developer must ad-
culture shock to others. Therefore, the strategy dress each individual’s preferences as well. This
followed in designing the site should address is called customization or “adaptive interfaces”
whole populations (Rutherford, 2000). (Ardissono et al., 2002) and refers to a Web site’s
In order to achieve the above a Web designer/ built-in facility to identify a user’s preferences
developer has to take under consideration certain even before any interaction takes place between
realities. First, it is a false assumption that English the user and the site and present information in
is the internet’s dominant language. Indeed several a way that is tailored by the users’ preferences
studies and surveys reveal other languages are (Svet, 2003) and knowledge. Some scholars,
used more by the internet users (Xanthidis and scientists, professionals and generally experts in
Nicholas, 2004; Communicate…, 2000). A com- the field use the term personalization instead of
pany that wishes to attract multilingual audiences customization. Most of the times they mean the
should include them in their e-commerce strategy same functionalities and just occasionaly they
as alternative target audience in addition to the distinguish the two referring to the former when
English speaking internet users (Hanrahan and focusing in Web site content and to the latter when
Kwok, 2001). Then, the color used for the back- focusing in Web site interface and design. These
ground is another issue that affects users’ positive variations in the way a Web site appears or as of
or negative reactions. Recent surveys revealed that the content presented to different user profiles can
different colors have different connotations for be created using forms, queries, cookies or other
different regional populations (Zhang et al., 2000; mechanisms and stored in databases.
Hanrahan and Kwok, 2001). One such example A few added features could yield some more
is the case of China where red is the dominant positive points towards achieving personalization
color (Anderson and Fell, 2003) and has several and/or customization. For example, the different
meanings including the symbolism that the person payment options accepted i.e. credit cards, money
whose name is written in red is dead, or about to orders, various types of checks, etc., should be
die, or on the other hand expresses such happy listed somewhere and a detailed description of the
moments of life as birthdays, weddings etc. and specific procedures followed both by the visitor
in general is considered good luck. and prospect customer of the company and by the
A third key point to seriously consider when representatives should be provided (Hanrahan and
developing a Web site with global prospects is Kwok, 2001; Housley, 2004). Also, a currency
that users should also be informed about the legal converter should be implemented to facilitate
framework related to taxation and import/export quick conversion between currencies (Housley,
procedures in case products are to move in or out of 2004; Hanrahan and Kwok, 2001). These are
a country (Rutherford, 2002). The reason for that is very easily implemented features and quite use-
it is not rare to have sanctions imposed to a country ful helping visitors’ decision towards engage in
from the part of the international community or e-commerce transaction but are, once again, quite
have the a ban on certain products from the part of often overlooked.

123
A Proposed Template for the Evaluation of Web Design Strategies

Requirements, Accessibility site designer/developer is to implement it in such


a way as to make it accessible for persons with
The introduction of new types of electronic devices disabilities. A Web site should be designed not only
such as personal digital assistants (PDAs), new with the media through which it is viewed in mind
generation mobile phones, also called cellular but its viewers as well. People with disabilities such
phones, with embedded internet capabilities, as limited vision, hearing or mobility could find it
etc., in addition to the personal computers and difficult to navigate in a site filled with “graphi-
laptops (notebooks) all having different abilities cally intense” content i.e. fairly large amounts
of presenting content to their user has lead the of different colors and graphics. Too much color
companies to find ways to make their Web sites could be a problem for a person who is “visually
available to different platforms and operating sys- impaired” person since it may make the content
tems. The advent of mobile commerce in several hard to read (Universal…, 2004). For that reason
digital economies worldwide has just stressed this it is suggested to host a second or even third ver-
need even further. sion of the Web site, customized for persons with
The main problem to be tackled is that a disabilities (McManis et al., 2001).
Web site display is primarily hardware/software There are a number of guidelines to follow
dependent. Imagine the enormous problems that for versions of Web sites for viewers with vari-
arise when a Web site designed in 1200x1600 in ous types of disabilities. The first is that if tables
32-bit color is viewed in the 120x240, 256-color are used in which text is contained, then a logical
display of a PDA or a mobile phone. It would, grouping must be followed since most screen
simply, be unreadable. In order to address this readers and magnification software may not read
issue professionals suggest the design of a Web the text correctly. Text in cells must be separated
site that is available and its content prenentable into paragraphs to assist reading software (Mak-
by any type of electronic device. Currently, it ing…, 2002; Pyatt, 2004). Additionally, include
appears that there are no tools available that the functionality of providing textual description
actually convert a Web site’s layout for different of images and “non-textual elements” in the case
device usage but there are tools that help create where browsers are configured not to display
accessible code [Macromedia…, 2004; Bohman, images, or the person who is using the browser
2003; Sullivan, 2004). The recommended way of uses a screen reader that cannot “read” images
dealing with platform diversity is to separate lay- (Making…, 2002).
out design and content by having different layout In the most likely event that a Web site in-
templates, or “style sheets” for the same content, cludes “motion and animation” this should be at
depending on the specific device (Nielsen, 1999; a frequency of less than 2Hz or more than 55Hz,
Accessibility…, 2004). as within this frequency range an animation may
Closely related to the type of electronic device, “trigger epileptic seizures” to individuals with
the operating system and the browser platform, but related health problems (Making…, 2002; 30;
mainly to the internet connection speed available Iowa, 1999). Finally, it would be quite useful to
is the time required for a Web page of a Web site to ensure that the site contains a mechanism that
be loaded. This is a crucial factor when evaluating may trace the utilization of a screen reader or
the Web site. Slow response speed, i.e. the time any other related tool and/or has also the ability
required to load a page inside a Web site, of more to identify a change in language and adjust ac-
than 7-10 seconds, could be annoying (California, cordingly (Making…, 2002).
2002) and discourage internet users to revisit. Some other issues related to accessibility and
No doubt the biggest challenge for every Web availability are the following:

124
A Proposed Template for the Evaluation of Web Design Strategies

• Text that is available through JavaScript Unfortunately, it is a proven fact that there is no
i.e. pop-up windows, should also be avail- bulletproof mechanism to ensure complete defense
able for users with JavaScript disabled against the various types of threats e.g. spyware,
and for users utilizing screen readers since viruses including Trojan horses among other mali-
screen readers may not read text contained cious programs/scripts, adware, even cookies used
in JavaScript (Pyatt, 2004). improperly by unauthorized people, etc. Then, it
• Frequently Asked Questions (FAQ) should is on the developer’s judgment to decide which
be implemented together with a Help of the available mechanisms should be used and
Topics feature. Accessibility covers mat- how to ensure the success of internet strategies
ters that deal with user problems, not just without compromising their visitors’ privacy and
hardware. Online guides, help topics and keeping security, legal and ethical issues properly
support reduce user stress to solve poten- addressed. This is a goal very difficult to achieve.
tial problems and increase the ease of navi- On one hand the more visitors’ data a company
gating the Web site (Roy et al., 2001). keeps recorded the more effectively it will present
a customized Web site tailored to their needs. On
The main problem with the implementation of the other hand this causes many compromises in
the features mentioned just above is the overall terms of the visitors’ personal privacy and quite
costs associated with it both in terms of time but possibly of their personal computer security from
most important in terms of money. To make matters malicious software. The following guidelines
worse the returns of investment are far from suf- address the majority of these issues except all
ficient enough as the internet users’ population that those that are not directly connected to the Web
would be addressed by such sophisticated sites is e.g. television/radio marketing.
a very small part of the overall online population. First, in order to ensure the visitor’s privacy
This is the main reason even large multinational Web sites should not collect sensitive information
corporations are reluctant in realizing such types from users’ PCs without their consent. In those
of Web sites despite the various directives towards cases where it is decided as necessary to collect
this goal from government organizations or na- such information the visitors should be clearly
tions like the European Commission and the U.S. informed as of the use of the information. Personal
department of Justice just to name a couple. data is most often collected through the use of
cookies, spyware, and other related mechanisms. It
Security, Privacy, Legal and Ethical is not always possible to directly confirm existence
Issues Involved in Internet Marketing of such mechanisms like spyware and computer
viruses. However, there are a number of anti-
Security, privacy, legalities and ethics are probably spyware and anti-virus software available which
the most discussed technologies issues, nowa- can be used by internet users to quite successfully
days. Several studies were conducted to clarify block any attempts to compromise their personal
how they affect large corporations’ successful computers’ security and their own privacy.
or failed strategies to attract digital consumers Second, aiming to further protect visitors’
(Privacy…, 2002). Currently, the most effective personal financial data e-commerce Web sites
way to tackle this problem is to apply available at the transaction level (Xanthidis and Nicholas,
mechanisms, in the form of software packages, 2007) should implement security protocols and
aiming to protecting and securing valuable and services, like SSL (Secure Socket Layer), SET
sensitive data and restricting access to vulnerable (Secure Electronic Transaction) to name the most
systems (Benjamin et al., 1998). widely used, to ensure safe transactions over the

125
A Proposed Template for the Evaluation of Web Design Strategies

Internet (Verisign, 2002). Furthermore, in those subjective judgement as much as possible. Also,
cases where e-mails are exchanged between the the authors wanted to design in a way to make the
company’s representatives and their Web site’s evaluation of any Web site, no matter how large or
visitors it could help if the latters’ e-mail addresses complicated, quite feasible as of the time required.
were protected or masked via some type of scripts, The results can be seen in the next four tables (see
forms, buttons, etc. to help defend against spam Table 1, Table 2, Table 3, Table 4).
bots’ attempts to identify e-mail addresses while In order to test the template the authors selected
crawling the Web. for evaluation a quite significant in size sample
Looking from the opposite site that of the of the Web sites of 232 medium-large companies
owner of the Web site things are a little more from the 15 sectors of the Greek economy, public
complicated in those cases where the site is and private, either “local” (Greek) or international
somehow connected with the company’s intranet. with subsidiaries in the country (see Appendix).
In that case, the company’s information and com- It took between 15’ and 20’ to evaluate each site
munication technology experts must ensure that and the evaluation period started on 11/5/2005 and
no unauthorized person is allowed to enter Web ended on 30/6/2005. The answers to the evaluation
pages restricted to authorized access. This could questions were quite straightforward, binary in
be achieved by implementing log-in procedures nature, a fact that simplified the process. In several
to authenticate a visit to such pages. cases during the evaluation, i.e. appropriate and
Finally, in every Web site like in every other appealing, we found it was somewhat subjective
software a message should be present to inform to determine/measure the Web site but in broad
the reader that the material is an official document lines our decision was in general agreement with
bound by relevant international copyright laws what any person would value as an appropriate and
(Fishman, 1994). appealing Web site. This holds true despite the fact
that people coming from different backgrounds
may view this issue under another perspective.
METHODOLOGY One aspect we found to be problematic was the
measurement of the time availability of a Web site.
The aim of this work was to develop a systematic This would be possible by monitoring the site’s
methodology for the evaluation of Web sites which online status on a 24/7 basis or alternatively by
would be as simple as possible yet comprehensive examining a detailed specification and possible the
enough to include all those elements discussed in logs files of the hosting server. Human resource
the previous section with straightforward answers and time constraints of this part of the study pro-
while the evaluation takes place. The authors, hibited us from doing the first. Furthermore, the
then, came up with an evaluation template which fact that the server and network specifications
comprises of 53 questions divided, not equally, and/or their log files were not available to us did
into four categories called dimensions each one not permit to follow the second path.
tackling the issues discussed earlier. It was also
decided to use only dichotomous questions and
avoid all other types, i.e. multiple answer, semantic ANALYSIS AND RESULTS
differential, likert scale valued, ranked ordered.
The rationale behind this decision is the idea to Web Site Design/Stickiness
have a template as binary in nature as possible,
one that any individual could use, even not hav- Initially, concerning Web site design issues, the
ing a significant technology background, limiting study showed that developers in Greece follow

126
A Proposed Template for the Evaluation of Web Design Strategies

Table 1. Evaluation template: Dimension I: Stickiness

1. Lack of tendency to use scrolling mechanisms. Yes = 1 No = 0


2. Hyperlink placement/style
i. Hyperlinks easily accessible (at a glance)? Yes = 1 No = 0
ii. Presence of floating hyperlinks (embedded in bars)? No = 0
iii. Font properties (name, size, bold/no bold, color) of the text hyperlinks distinguishing them from the
rest of the text? Yes = 1 No = 0
iv. Icons used in graphical type hyperlinks intuitively identifiable, i.e. do they represent the target object or
are they misleading? Yes = 1 No = 0
3. Hyperlink target/content
i. Tendency NOT to have dead hyperlinks in the site (use home page)? Yes = 1 No = 0
ii. Hyperlinks lead to relevant pages? Yes = 1 No = 0
4. Site maps
i. Presence of any type of site map, i.e. site tree diagram, drop-down menus, etc.? Yes = 1 No = 0
ii. Mapping mechanisms informative as to the actual depth in which the user navigates? Yes = 1 No = 0
5. Web site user interface attractiveness
i. Appropriate and appealing? Yes = 1 No = 0
ii. Lack of distracting and annoying elements? Yes = 1 No = 0
6. Information quality and completeness
i. Any “read more” hyperlinks available clarifying possibly broad, unclear or unknown topics to the reader? Yes = 1 No = 0
ii. Is the information provided in the Web site signed and, thus, credible? Yes = 1 No = 0
iii. Is the information provided updated on a reasonably expected timeliness? Yes = 1 No = 0
iv. Any internal search engine available? Yes = 1 No = 0
7. Visitor’s feedback enabled and online help available
i. E-mail links available to the visitors? Yes = 1 No = 0
ii. Online surveys available? Yes = 1 No = 0
iii. Feedback forms available? Yes = 1 No = 0
iv. On line help available (e.g. FAQs, etc.)? Yes = 1 No = 0

the internationally accepted standards of what were quite helpful in distinguishing it from the
a nice looking Web site should be like (Figure rest of the objects on the Web sites in 204/232
2). Indeed, the majority of developers (142/232; of the cases (87.9%). In general, the hyperlinks
61.21%) tend to avoid the use of scrolling mecha- could be found easily at a glance in 167/232 of
nisms which proved to have a negative impact in the cases (71.9%).
the Web site’s attractiveness. In all but 3 cases Concerning the presence of any sort of distract-
(229/232; 98.7%) the hyperlinks were found to ing and annoying elements, findings were quite
lead to relevant pages and in all but 6 (226/232; positive again as in 212/232 cases (91.3%) no
97.4%) the hyperlinks did not lead to a dead end. such elements were found and in general the user
If icons are used to represent the hyperlinks, the interface of the Web sites was found to be appropri-
selections were found to be intuitively identifi- ate and appealing in 215/232 cases (92.6%). The
able in 216/232 cases (93.1%). When text was only problem related to site interface was that in
the basis for the hyperlinks, the font properties the majority of the Web sites floating hyperlinks

127
A Proposed Template for the Evaluation of Web Design Strategies

Table 2. Evaluation template: Dimension II: Customization and globalization

1. Languages supported (cocacola.com is worldwide)


i. English Yes = 1 No = 0
ii. Spanish Yes = 1 No = 0
iii. Chinese Yes = 1 No = 0
iv. French Yes = 1 No = 0
v. German Yes = 1 No = 0
vi. Other (Greek, Turkish, Arabic, Hebrew, Japanese, etc.) Yes = 1 No = 0
2. Colors used: Is Web site color related with the cultural background (Western, Asian, etc) of the targeted
population? Yes = 1 No = 0
3. Issues related to globalization
i. Briefing/information provided concerning import/export and taxation issues? Yes = 1 No = 0
ii. Any restrictions applicable for a commodity to be exported/imported to/from certain countries? Yes = 1 No = 0
iii. Any list of countries to which import/export restrictions apply? Yes = 1 No = 0
iv. Any information provided about available shipping/ delivery options? Yes = 1 No = 0
4. Level of customization the Web site achieves
i. Level 0: No customization Yes = 1 No = 0
ii. Level 1: Content → display information based on previous user interaction and preferences stored in
log files. Yes = 1 No = 0
iii. Level 2: Suggestive → display information on relevant or competitive commodities/services. Yes = 1 No = 0
iv. Level 3: Informative → display further clarifications on issues not in the sphere of the user’s knowl-
edge. Yes = 1 No = 0
v. Level 4: Design format → lets the user permanently decide the layout of the Web site as it appears in
his/her browser. Yes = 1 No = 0
vi. Level 5: Language and Culture → identifies the user’s of language and culture preference based on the
IP address of the user’s system or on the user’s selection of a region/country from a map/list available. Yes = 1 No = 0
5. Payment – shipping/billing options
i. List of different payment options available? Yes = 1 No = 0
ii. Detailed description of each payment option available? Yes = 1 No = 0
iii. Currency converter available? Yes = 1 No = 0
iv. Use of the universal “postal code” instead of the regional “zip code”? Yes = 1 No = 0

were not present (152/232; 65.5%). 0.4%). The same can be said of the availability –
The problems started when more technical de- lack, rather – of online surveys with only 2 sites
tails were evaluated (Figure 3). One of the central implementing such feedback mechanisms for the
components of any Web site, the site map, was visitors (2/232; 0.8%).
only found in 85/232 (36.6%) of the Web sites. The most surprising element of the findings,
More technical mapping mechanisms that could however, was that only 156/232 sites (67.2%) in-
provide information as of the depth in which the corporated the very easy to implement and almost
visitor navigates were even more seldom utilized obvious to include e-mail link for the visitors and,
(10/232; 4.3%). As far as online help made avail- also, less than half the Web sites (106/232; 45.6%)
able to visitors is concerned, unfortunately, such made a feedback form available to visitors. The in-
a feature was only available in one site (1/232; formation was updated quite regularly in 227/232

128
A Proposed Template for the Evaluation of Web Design Strategies

Table 3. Evaluation template: Dimension III: Accessibility, availability, hardware/software requirements

1. Is the Web site accessible (Platform Compatibility) Yes = 1 No = 0


2. Is the Web site optimized for users with a mental or physical handicap? Yes = 1 No = 0
3. Hardware/ Software/ Network requirements
i. Time required loading the Web site’s home page? <10” = 1 No = 0
ii. Web site displayed properly, i.e. no horizontal scrolling mechanisms, no twisting of objects, etc., in
different display resolutions? Yes = 1 No = 0
iii. Option to download and install “third party” components required to view the Web site, e.g. activeX,
flash players, different fonts, etc. Yes = 1 No = 0

Table 4. Evaluation template: Dimension IV: Security, privacy

1. Security
i. Authentication required to login into possible intranet part of the site? Yes = 1 No = 0
ii. If transactional or interactive what protocol are they using (None, SSL, SET, Other)? Yes = 1 No = 0
iii. What is the cipher strength? Yes = 1 <128 = 0
iv. On-line anti-virus scanner available? Yes = 1 No = 0
v. Web site expires after a pre-defined amount of idle time? Yes = 1 No = 0
2. Privacy
i. Avoid the use of tracking/identifying mechanisms i.e. cookies, spyware, etc, without the consent of
the user? Yes = 1 No = 0
ii. Privacy statement? Yes = 1 No = 0
iii. Masked e-mail addresses through scripts, forms, buttons, etc? Yes = 1 No = 0

cases (97.8%), however, it is signed – and, hence, groups of people in parts of the world.
credible – in only 32/232 cases (13.8%). There In terms of the colors used, they were found to
is the projection of providing more information be appropriate (231/232; 99.57%) in connection
in 134/232 sites (57.8%) through the utilization with the populations of visitors targeted (based
of “read more” hyperlinks and 88/232 of those on the assumption that the language in which the
(38.0%) included an internal search engine. text was written reflected the mother tongue and
culture of the visitor targeted). However, it was
Customization and Globalization also realized that this last conclusion was not
based on enough evidence and it would be very
It was quite obvious that except Greek (211/232; interesting to see how things would change if,
90.95%), the only other language seriously con- say, a large proportion of the targeted population
sidered and used when developing Web sites was were Chinese-speaking people.
English (178/232; 76.72%). All the rest, i.e. Span- Finally, only the executives of a handful of
ish (7/232; 3.01%), French (7/232; 3.01%), Ger- companies seem to take things seriously as only
man (8/232; 3.44%) and Chinese (3/232; 1.29%), in 10/232 cases (4.31%) was there some kind of
included into our study, were seldom used mainly briefing about import/export and taxation matters,
in specific cases of companies which probably and only in 11/232 Web sites was information
had some business or other relations to certain provided on shipping/delivery options (4.34%).

129
A Proposed Template for the Evaluation of Web Design Strategies

Figure 2. Web site evaluation from a technical viewpoint. Dimension 1 - Stickiness

No discussion can be made about providing lists there were no Web sites found to allow changes
of countries or commodities for which certain in the format of the design, i.e. let the user per-
imports/exports restrictions apply (Figure 4). manently decide on the layout of the Web site as
Next, concerning customization features of- it appears on his/her client browser, neither did
fered by the Web sites, the study showed that any attempt to identify the user’s language and
89/232 (38.36%) Web sites provide no customi- culture preferences based on the IP address of
zation features whatsoever. Furthermore, no Web the user’s system or on the user’s selection of a
site was found displaying information based on region/country from a map/list available.
previous user interaction and preferences stored Last, as far as available payment and shipping/
in log files or in the form of cookies. A relatively billing options, the general picture of the results of
significant (and rather unexpected considering the study proved disappointing (Figure 5). Only
the previous results) number (108/232; 46.55%) 21/232 companies’ Web sites (9.05%) provided
displayed information on relevant or competi- a list of available payment options and just 3 of
tive commodities/services placing those sites in them (1.29%) described the steps to be followed
the suggestive level of customization. A small for each option in some detail. The currency con-
percentage (32/232; 13.79%) displayed further verter feature, often found in many international
clarifications on issues not in the sphere of the business-oriented Web sites, was present only in
knowledge of the user (informative level). Finally, 3/232 (1.29%). Universal terminology for pay-

130
A Proposed Template for the Evaluation of Web Design Strategies

Figure 3. Web site evaluation from a technical viewpoint. Dimension 2 – Mapping and communication
mechanisms – Informative structures

Figure 4. Information provided by the sites related to globalization issues

131
A Proposed Template for the Evaluation of Web Design Strategies

ment and shipping/delivery, e.g. the universal term time (as suggested for security purposes) was
“postal code” used instead of the regional term present and/or activated.
“zip code”, appeared in 34/232 cases (14.66%). On the other hand, concerning the visitors’
privacy, the study showed a tendency to avoid
Accessibility, Hard/ tracking/identifying mechanisms, like cookies,
Software Requirements spy ware, etc. (231/232 cases; 9.57%) but there
is serious doubt this was a conscious decision on
The results of the evaluation concerning acces- the part of the companies’ executives or lack of
sibility and hard/software requirements were radi- technical expertise required to implement it. Ad-
cally different (Figure 6). In most cases (162/232; ditionally, only 31/232 sites were found (13.36%)
69.83%) the sites were accessible from the com- with a privacy statement (an utterly simple feature
mon different platforms, i.e. Windows, Linux. to realize), and only 56/232 (24.14%) cases of
The results were also positive when the time to Web sites with masked e-mail addresses through
load the home page was evaluated (less than 10’’ scripts, forms, buttons, etc. for privacy-related
was considered a reasonable time) with 214/232 reasons.
(92.24%) succeeding, as well as the proper display
of the page contents without object distortion or
any other display anomalies under different display IMPLICATIONS
resolutions (226/232; 97.41%). On the negative
side, unfortunately as expected, only in 1 case It is quite possible after all, although very difficult,
were the Web sites designed with the physically to put together all bits and pieces that make up the
disabled people in mind. Also disappointing was puzzle of what suggests a good Web site, be it a
the fact that in case third party tools were needed simple one or a very sophisticated. The number
to run the Web pages of a Web site only in 15/232 of the elements to be evaluated is significant, of
cases (6.47%) were the options to download the course, but limited enough to make the whole
respective tools given to the visitors. idea of evaluating a Web site feasible as of time
and effort required. Furthermore, the nature of the
Security, Privacy questions of the template is such that allows even
non-experts of the field of e-commerce to use it
The results of evaluating the Web sites with for their evaluation of their Web sites.
security and privacy in mind were completely The template introduced in this chapter could
disappointing (Figure 7). be used as a “marking scheme” for the assessment
On the one hand, security concerns should be of all different types of Web sites by information
tackled but very little was found be undertaken and communication technology (ICT) experts in a
in this direction. Only 20/232 (8.62%) sites had systematic way. It could also be used by business
some kind of authentication process running when and other professionals as a suggestion, a rule of
a visitor requested to access to the companies’/ thumb indeed, of how could or should their com-
organizations’ intranet. Even less were the sites panies’ Web sites be implemented to meet their
(16/232; 6.90%) protected by a security protocol online strategies. The main reason the template
like SSL, SET, etc. The cipher strength for the can deliver both the aforementioned goals is that
sites protected by such protocols was less than the questions to be answered carry on them the
128 (very low indeed) in 218/232 cases (6.03%). “correct” approaches of how to address each of
No on-line anti-virus scanner was available and the issues related to the development of a Web
no expiration time after a specific amount of idle site. Therefore, it is quite straightforward that

132
A Proposed Template for the Evaluation of Web Design Strategies

Figure 5. More globalization issues faced

Figure 6. Accessibility and availability issues

every positive answer adds one positive point to performance of the rather large local information
the overall design of the site with any negative society in e-commerce activity it was possible to
answer pointing to its weaknesses. use this template to draw useful conclusions for
the quality of the Web sites and their effect in
e-commerce growth in the country. The authors
LIMITATIONS: FURTHER RESEARCH selected the Greek Web sites for reasons of physi-
cal proximity to the companies (Greek residence).
The authors used a number of Greek Web sites to However, it would be quite interesting to test the
test this systematic methodology that the suggested same methodology on the Web sites of companies
evaluation template represents. Despite the poor in the developed digital economies and those of

133
A Proposed Template for the Evaluation of Web Design Strategies

Figure 7. Security and privacy issues tacked

the underdeveloped countries as well just to see digital economy in the country of the company
if it is applicable in those cases as well. Also, that owns the site, and others.
the methodology was tested on medium-large
companies’ sites. It would help to see if and how
much applicable it would be in the case of small CONCLUSION
companies as well.
Apparently, like in every other field of informa- It should be underlined once again that the task
tion and communications technology this effort of designing a comprehensive template/formula
is not over. This is only a measurement template. that could be used to evaluate Web sites looking
The next step, parallel to updating the template from different viewpoints based on the various
with new ideas and insight from professionals professions involved is extremely difficult, but
all over the world, is to work towards clarifying proven feasible. The pieces of the puzzle are
the metric against which the various results of many and as the technology changes they change
measurement should be checked. There could be with it. However, there are a few facts that helped
certain numerical results that distinguish a poor achieve our goal of designing the proposed evalu-
designed Web site from a better one and then ation template.
from a very good one. Probably such a distinc- First, despite the quite significant, some could
tion is more complicated as it depends on certain consider it rather large, number of 53 elements of
variables including but not limited to the type a Web site to be evaluated they can be organized
of the Web site, e.g. information level up to full into categories each one addressing completely
e-commerce solution, the progress towards the different viewpoints. This is the reason the authors

134
A Proposed Template for the Evaluation of Web Design Strategies

called them dimension as it is like looking into Ardissono, L., Goy, A., Petrone, G., & Segnan,
the same structure from different angles. Second, M. (2002, May). Personalization in B2C in-
it is proven that with just a little difficulty it is teraction. Communications of the ACM, 45(5).
possible to answer the questions of the template doi:10.1145/506218.506245
in a straightforward binary way that leaves very
Benjamin, R., Gladman, B., & Randel, B. (1998).
little or no room to different and subjective inter-
Protecting IT systems from cyber crime. [ac-
pretations of the quality of features implementing
cessed: December 2004]. The Computer Journal,
while addressing each specific issue. Third, it is
41(7), 429–443. Available www.cs.ncl.ac.uk/
proven possible to have a numerical result of the
research/pubs/trs/papers/631.ps. doi:10.1093/
evaluation, although, by itself this cannot lead to
comjnl/41.7.429
safe conclusions as of the quality of the Web site
overall but only to relevant suggestions. Finally, Bohman, P. (2003). Using Opera to check for
the time required to use this template to evaluate accessibility. Available: www.Webaim.org/tech-
any Web site is very reasonable, less than 20’ per niques/articles/opera [accessed: 2004].
site, regardless of the quality, type or size of the
California State University. Sacramento (2002).
Web site.
Web Design Principles. Available: www.csus.edu/
The authors do not claim they found the ideal
uccs/training/online/design/d_principles.htm.
solution to the problem. What was proposed in
this chapter is the result of the ongoing process of Communicate in your customers’ language (2000).
reviewing, adjusting and incorporating different WorldLingo Press Releases [online], Available:
ideas and thoughts into a template as little technical www.worldlingo.com/company_information/
as possible in order for anyone to be able to use pr20000625_01.html [accessed: November
it even if that person is not an ICT expert. The 2004].
authors opened a path which they believe could
D-Lib Magazine. (2003). Trends in the Evolution
lead to a systematic way of assessing the quality
of the Public Web. Available: www.dlib.org/dlib/
of Web sites. The thing that encourages them to
april03/lavoie/04lavoie.html [accessed: 2004].
move further this type of research is the positive
and insightful comments received in the confer- Evaluating Web Pages. Techniques to apply and
ences where this methodology is presented. questions to ask (2004). UC Berkeley – Teaching
Library Internet Workshop [online], Available:
www.lib.berkeley.edu/TeachingLib/Guides/Inter-
REFERENCES net/Evaluate.html [accessed: December 2004].

Accessibility on the Web. (2004). International Faster Growth Heightens Web Class Divide.
Center for Disability Resources on the Internet (2002). TRN (Technology Research News) [on-
[online], Available: www.icdri.org [accessed: line], Available: www.trnmag.com [accessed:
December 2004]. December 2004].

Anderson, B., & Fell, J. M. (US Government Fishman, S. Software Development: A legal guide,
Export Portal) (2003). Taming the Red Dragon: Nolo Press, Berkeley, 1994, First Edition.
Getting in and Succeeding in China, Available: Hanrahan, M., Kwok, & Wei-Tai (2001). A New
www.export.gov/comm_svc/pdf/RedDragon.pdf Connectivity - How Technology is Changing the
[accessed: 2004]. Marketing World. Available: iaa2001.atalink.
co.uk/html/p026.htm [accessed: 2004].

135
A Proposed Template for the Evaluation of Web Design Strategies

Helm, J. C. (2001). Web based Application Qual- McManis, B. L., Ryker, R., & Cox, K. C. (2001).
ity Assurance. Available: sce.cl.uh.edu/helm/ An Examination of Web Usage in a global context.
Helm_html/papers/QW2001_PaperJune0101.pdf Industrial Management & Data Systems, 101(9),
[accessed: 2004]. 470–476. doi:10.1108/02635570110410654
Hirsh, L. (2002, February 25). Online Shopping Nielsen, J. (1999). User interface directions for
and the Human Touch, E-Commerce Times: The the Web. Communications of the ACM, 42(1),
E-Business and Technology Supersite [online], 65–72. doi:10.1145/291469.291470
Available: www.ecommercetimes.com/.../16483.
Novell (2004). Securing your Web site. Novel [on-
html [accessed: December 2004].
line], Available: www.novell.com/documentation/
Housley, S. (2004). Web site Globalization. nw6p/index.html [accessed: December 2004].
Available: www.onevision.co.uk/xq/ASP/id.1072
Privacy and Data Sharing. A Performance and
[accessed: 2004].
Innovation Report (2002). Cabinet Office (UK
Interface Design. (2002). Web Style Guide 2nd Government – Department of Constitutional Af-
Edition [online], Available: www.Webstyleguide. fairs) [online], Available: www.foi.gov.uk/shar-
com/interface/user-centered.hmtl [accessed: De- ing/pubs.htm [accessed: December 2004].
cember 2004].
Pyatt, E. J., Pennsylvania State University (2004).
Internet Guide. (2002). Canada Government Creating Accessible Web Sites, [accessed: Dec.
[online], Available: www.cio-dpi.gc.ca/ig-gi/m/ 2004].
wsm-msw/wsm-msw_e.asp [accessed: December
Roy, M. Christine, Dewit, O., Aubert, B. A. (2001).
2004].
The impact of interface usability on trust in Web
Iowa University. (1999). Designing accessible retailers. Internet Research: Electronic Network-
Web sites booklet, Available: www.uiowa.edu/ ing Applications and Policy, 11(5), 388–398.
infotech/WebAccess.htm [accessed: 2004]. doi:10.1108/10662240110410165
Katerattanakul, P., & Siau, K. (1999). Measuring Rutherford, E. (2002). How to Avoid Glob-
Information Quality of Web sites. In Proceedings al Website Disasters. Available: www.cio.
of the 20th International Conference of Informa- com/research/.../111400_disaster.html [ac-
tion Systems [online], Charlotte, North Carolina, cessed:2004].
United States, Available: portal.acm.org [ac-
Stickiness in an eTailing environment (2000).
cessed: December 2004].
Northwestern University [online], Available:
Macromedia – Accessibility. (2004). Macrome- www.medill.northwestern.edu/imc/studentwork/
dia [online], Available: www.macromedia.com/ projects/Sticky/index.htm [accessed: November
macromedia/accessibility [accessed: December 2004].
2004].
Sullivan, N. (2004). Why design accessible Web-
Making your Web sites Accessible with Microsoft sites. PostDiluvian Org. [online], Available: post-
FrontPage (2002). Microsoft Corporation [on- diluvian.org/~nicole/work/why.php [accessed:
line], Available: www.microsoft.com [accessed: December 2004].
December 2004].

136
A Proposed Template for the Evaluation of Web Design Strategies

Sutcliffe, A. (2002). Heuristic Evaluation of Verisign (2002). Guide to securing your Web site
Website Attractiveness and Usability. Pro- for business. Verisign [online], Available: www.
ceedings of the 35 th Hawaii International verisign.com/resources/gd/secureBusiness/se-
Conference on System Sciences [online], Avail- cureBusiness.html [accessed: December 2004].
able: csdl2.computer.org/comp/proceedings/
Xanthidis, D., & Nicholas, D. (2004). Evaluat-
hicss/2002/1435/05/14350137.pdf [accessed:
ing internet usage and ecommerce growth in
December 2004].
Greece. Aslib Proceedings, 56(6), 356–366.
Svet, B., & the State University of New York, doi:10.1108/00012530410570390
Buffalo. (2003). Personalization and Customi-
Xanthidis, D., & Nicholas, D. (2007). A pilot
zation in eCommerce. Available: www.cse.buf-
qualitative study of eCommerce practices and
falo.edu/~sbraynov/seninar2003/presenttations/
attitudes of medium-large companies’ executives
lecture.pdf [accessed: 2004].
in Greece. WSEAS Transactions on Information
Thelwall, M. (2000). Commercial Web sites: lost Science and Applications, 4(2), 354–361.
in cyberspace? Internet Research, 10(2), 150–159.
Zhang, X., Keeling, B. K., & Pavur, J. R. (2000).
doi:10.1108/10662240010322939
Information Quality of commercial Web sites
Trends in the Evolution of the Public Web. (2003). Home Pages: An explorative analysis. ACM
D-Lib Magazine [online], Available: www.dlib. [online], Available: portal.acm.org [accessed:
org/dlib/april03/lavoie/04lavoie.html [accessed: November 2004].
November 2004].
Universal Web Site Accessibility Policy for
State Web Sites – Version 4.0, (2004). State of
Connecticut Web site Accessibility Commit-
tee [online], Available: www.cmac.state.ct.us/
access/policies/accesspolicy40.html [accessed:
December 2004].

137
A Proposed Template for the Evaluation of Web Design Strategies

Appendix

Companies: http://www.presspoint.gr/sectors.asp

(Sources: ASE (Athens Stock Exchange): companies in blue background, Presspoint.gr: companies in
white background)
Table 5.

Sector 1: Food, beverages and tobacco industry Web Site


     1. Coca Cola Greece S.A. www.cocacola.gr/
     2. Chipita International Α.Β.&Ε.Ε. www.chipita.com
     3. Katseli & Sons Α.Β.Ε.Ε. www.katselis.gr
     4. Eurofarma Α.Β.Ε.Ε. www.evrofarma.gr
     5. CMA www.cma-greece.gr
     6. Diageo www.diageo.com
     7. Intercatering www.intercatering.gr/
     8. Kraft Foods www.kraftfoods.gr/
     9. Nestle Hellas www.nestle.gr/online
     10. Athens Beers S.A www.amstel.gr/
     11. Alatini S.A. www.allatini.com.gr
     12. Vasiliou Wines www.vassilioudomaine.gr/
     13. Delta Ice Creams S.A. www.delta.gr/
     14. Dodoni Ice-Creams A.B.E.E www.dodoni.com.gr/
     15. ELAIS S.A. Olive products businesses www.elais.gr/
     16. Pedestrians Union www.pezaunion.gr/
     17. Thraki S.A www.thraki-sa.gr
     18. ΙΝΟ S.A. www.inowines.gr/
     19. Kanakis ST. Α.Β.Ε.Ε. www.stelioskanakis.gr/
     20. Kri-Kri Milk Company Α.Β.Ε.Ε. www.krikri.gr/
     21. Lazaridis Wines S.A. www.domaine-lazaridi.gr/
     22. Mevgal S.A. www.mevgal.gr/
     23. Mega Farm www.megafarm.gr/
     24. Melisa - Kikizas www.melissa-kikizas.gr/
     25. Mparmpa Stathis Foods S.A. www.geniki-trofimon.gr/
     26. Louli Mills S.A www.loulisgroup.com
     27. Nostimo A.E.Β.Ε www.musses.gr/
     28. Xifias Fish S.A www.xifias.gr/
     29. Chatzikraniotis & Sons www.xatzikranioti.gr/
Sector 2: Chemical industries Web Site
     1. Veterin Α.Β.Ε.Ε. www.veterin.com/
     2. PLIAS Α.Β.&Ε.Ε. www.plias.gr/
     3. Druckfarben Hellas S.A.Β.Ε. www.druckfarben.gr/

138
A Proposed Template for the Evaluation of Web Design Strategies

Sector 1: Food, beverages and tobacco industry Web Site


     4. Ballis Chemicals A.E.Β.Ε. www.ballis.gr/profile
     5. Kerakoll Hellas www.kerakoll.com/
     6. Neochimiki L.V. Lavrentiadis Α.Β.Ε.Ε. www.neochimiki-lavrentiadis.gr/
     7. Famar S.A. www.famar.gr/
Sector 3: Transport equipment manufacturing Web Site
     1. Neorion Shipyards S.A www.neorion-shipyards.gr/
     2. Sfakianakis S.A.Β.Ε. www.suzuki.gr/
     3. Petropoulos, P., S.A.&Β.Ε. www.petropoulos.com/
     4. Hyundai Hellas www.hyundai.gr/hyundai/
Sector 4: Financial sector Web Site
     1. Alpha Leasing S.A. www.alpha.gr/introen.html
     2. Progress Funds S.A. www.progressfund.gr/
     3. New Millennium Investments A.E.Ε.Χ. www.newmillenniumaeex.gr/
     4. Dias fund A.E.Ε.Χ. www.diasfund.gr
     5. Altius Investments S.A.Ε.Χ. www.altius.gr/
     6. Credit Petropoulakis www.credit-sec.gr/
     7. Eurocapital Financial Services www.athenstock.com/
     8. EuroXX Finance www.euroxx.gr/
     9. Investor ΕΠΕ www.investor.gr/
     10. Aspis Bank www.aspissec.gr/
     11. Eurodynamics S.A.Ε.Χ. www.eudynamics.gr/
Sector 5: Insurance and pension funding services Web Site
     1. Ethniki Insurance .A.Ε.Γ.Α. www.ethniki-asfalistiki.gr/
     2. Agrotiki Insurances S.A. www.agroins.com/
     3. Phoenix Metrolife S.A www.phoenix-metrolife.com/
     4. Aspis Pronia S.A. General Insurances www.aspis.gr/
     5. Europisti S.A.Γ.Α. www.europisti.gr/
     6. Alico AIG Life www.alico.gr/
     7. ING Hellas www.ing.gr/
     8. International Life Group www.inlife.gr/
     9. Megaservice Ltd. www.megaservice.gr/
     10. Eurobrokers S.A. www.eurobrokers.gr/
     11. Interamerican Insurance S.A. www.interamerican.gr/
     12. Syneteristiki ΑΕΕΓΑ www.syneteristiki.gr/
Sector 6: IT services Web Site
     1. Logicdis S.A. www.logicdis.gr
     2. Ipirotiki Software & Publications S.A. www.ipirotiki.gr/
     3. Compucon Computer Applications A.B.E.E. www.compucon.gr/GR/
     4. Logismos S.A. www.logismos.gr/
     5. 01 Pliroforiki A.E. www.01p.gr/
     6. ABC Professional Services S.A. www.abc.gr/

139
A Proposed Template for the Evaluation of Web Design Strategies

Sector 1: Food, beverages and tobacco industry Web Site


     7. ACE Advanced Applications S.A. www.ace.gr/
     8. ACOM S.A.Β.Ε. www.acom.gr/
     9. Actis Information Systems S.A. www.actis.gr/
     10. Active Computer Systems Ε.Π.Ε www.active.gr
     11. Adacom S.A www.adacom.com
     12. Ahead Rm A.E www.aheadrm.com
     13. Algo Systems S.A. www.algo.com.gr
     14. Alpha Grissin INFOTECH S.A. www.alphagrissin.gr/
     15. Alpha IT S.A www.alphait.gr/
     16. Alphyra Hellas www.alphyra.gr/home/
     17. Altasoft www.altasoft.gr/
     18. American Computers & Engineers Hellas S.A. www.ace-hellas.gr/
     19. Anixter Greece Network Systems www.anixter.gr/
     20. Apollo www.apollo.gr/
     21. Areia S.A. www.areianet.gr/
     22. Arion Software www.arion.gr/
     23. Binary Logic Computers Ε.Π.Ε www.mmpi.net/
     24. Cardisoft S.A. www.cardisoft.gr/
     25. Datablue S.A. www.datablue.gr
     26. Hipac S.A.Β.Ε. www.hipac.gr/
     27. Infomap S.A. www.infomap.gr/
     28. Mantis IT S.A.Ε. www.mantis.gr/
     29. Gnomon IT S.A. www.gnomon.com.gr/
Sector: 7 Communications and Telecommunications Services Web Site
     1. Vodafone – Panafon S.A.Ε.Τ. www.vodafone.gr
     2. Forthent S.A. www.forthnet.gr
     3. Lanet S.A.Β.Ε.Τ. www.lannet.gr/
     4. Ιntersat S.A. www.intersat.gr/
     5. Algonet Telecommunications S.A www.algonet.gr
     6. Chorus Call Hellas S.A. www.choruscall.com/
     7. Com-Tonet S.A. www.com-tonet.gr
     8. Cosmoline S.A. www.cosmoline.com/
     9. Hellas Sat S.A. www.hellas-sat.net/
     10. OTEGlobe www.oteglobe.gr/
     11. Plural A.E.Τ.Β.Ε. www.plural.gr/
     12. Stet Hellas Telecommunications A.E.B.E www.tim.gr/
     13. Teledome www.teledome.gr/
     14. Telepassport Hellas www.telepassport.gr/
     15. Tellas Telecommunications S.A www.tellas.gr/
     16. Unitel Hellas S.A www.unitel.gr/
     17. Uunet Hellas www.gr.uu.net/

140
A Proposed Template for the Evaluation of Web Design Strategies

Sector 1: Food, beverages and tobacco industry Web Site


     18. Vivodi Telecommunications S.A www.vivodi.gr/
     19. VoiceWeb S.A. www.voiceWeb.gr/
     20. Winet www.winet.gr/
     21. Newsphone Hellas A.E.Ε. www.newsphone.gr/
     22. Mediatel Telephone Information S.A. www.mediatel.gr/
     23. Cosmote Mobile Telecommunications S.A. www.cosmote.gr/
     24. Q-Telecom www.myq.gr/
Sector 8: Health and social services Web Site
     1. IASO S.A. www.iaso.gr
     2. Euromedica S.A www.euromedica.com.gr/
     3. Ygeia Diagnostic & Therapeutic Center, Athens S.A www.hygeia.gr/
     4. Medicon Hellas A.E. www.mediconsa.com/
     5. Biorehab Hellas www.biorehab.gr/
     6. Gerolimatos Group of Companies www.gerolymatos.gr/
     7. Thessaloniki Psychiatric Hospital www.psychothes.gr/
Sector 9: Media and printing (newspapers) Web Site
     1. Lampraki Press Group S.A. www.dol.gr/
     2. Kathimerini S.A. www.kathimerini.gr
     3. Inform P. Lykos S.A www.lykos.gr/
     4. Ιmako Media Net Group S.A. www.imako.gr/
     5. Τechnical Press S.A www.technicalpress.gr/el/
     6. Alpha Satellite TV S.A. www.alphatv.gr/
     7. Filalthis www.filathlos.gr/
     8. MAD TV Productions www.mad.gr/
     9. Ellinika Grammata Publications www.ellinikagrammata.gr/
     10. Asfalisi Net www.asfalisinet.gr
     11. Αmalthia Publications S.A www.euro2day.gr/
     12. Alter TV S.A. www.alter.gr/
     13. Greek Radio Television S.A. www.ert.gr/
     14. Click FM ΕΠΕ www.klikfm.gr/
     15. OPAP S.A. www.opap.gr/
     16. Mega TV S.A www.megatv.com/
     17. Alupress S.A www.alupress.gr/
     18. Biokosmos News www.bioshop.gr/
     19. Compupress S.A. www.compupress.gr/
     20. Direction Publications S.A www.direction.gr/
     21. Europress Publications www.europress.gr/
     22. Knowledge Systems S.A. www.business2005.gr/
     23. Metamedia www.metohos.com
     24. Motor Press Hellas AEE www.chip.gr/
     25. Newspaper Direct Hellas www.newspaperdirect.gr

141
A Proposed Template for the Evaluation of Web Design Strategies

Sector 1: Food, beverages and tobacco industry Web Site


     26. Northmedia Publications www.city231.gr/
     27. Option Press S.A. www.optionpress.gr/
     28. Smartpress S.A. www.smartpress.gr/
     29. Travel Times Publishing ΕΠΕ www.traveltimes.gr/
Sector 10: Metal/machinery manufacturing – Mineral and Cement Web Site
     1. ELVAL S.A. www.elval.gr/
     2. Metal Company Arkadias Cr. Rokas Α.Β.Ε.Ε. www.rokasgroup.gr/
     3. Profil Pipe Company S.A. www.tzirakian.com/
     4. FITCO S.A. www.fitco.gr/
     5. S & B Industrial Minerals S.A www.s.andb.gr/
     6. Naxos Marbles ΑΕΒΕ www.naxos-marble.com/
     7. Pavlidis Marbles & Granites S.A www.pavlidismg.com/
     8. F.H.L. Η. Kyriakidis Marbles - Granites Α.Β.Ε.Ε. www.fhl.gr/
     9. Betanet Α.Β.Ε.Ε. www.betanet.gr/
     10. Heracles Cements Group www.aget.gr/
     11. Iktinos Hellas S.A. www.iktinos.gr/
     12. Mathios Refractories S.A www.mathios.gr/
     13. Sidma S.A. www.sidma.gr/
     14. Metka Metal Constructions Hellas S.A. www.metka.gr/
     15. Grecian Magnesite ΑΜΒΝΕΕ www.grecianmagnesite.com/
     16. Spider Metal Industry S.A. www.spidersa.com/
     17. MEVACO Α.Β.Ε.Ε. www.mevaco.gr/
Sector 11: Education Web Site
     1. Alba www.alba.edu.gr/
     2. Alexander www.alexanderinst.gr/
     3. Andim www.andim.gr/
     4. Business Training Center www.btc.com.gr/
     5. Centre of European Management Studies (CEMS) www.cems.gr/
     6. Compact S.A. www.compact.gr/
     7. Delta Singular Training S.A. www.ds-training.gr/
     8. Didacta Training Group www.didacta.gr/
     9. ECDL Hellas S.A. www.ecdl.gr/
     10. Icon International Training www.icon.gr/
     11. Infotest www.certification.gr/
     12. Inte*learn www.intelearn.gr/
     13. ITEC S.A. www.itec.edu/
     14. New York College www.nyc.gr/
     15. Proseed www.proseed.gr/
     16. Αkmi IEK www.iek-akmi.gr/
     17. Αmerican College, Hellas www.acg.edu/
     18. Xinis Educational Group www.xinis.com/

142
A Proposed Template for the Evaluation of Web Design Strategies

Sector 1: Food, beverages and tobacco industry Web Site


     19. Crete Educational Institutes www.sport-tourism.com/
     20. Aegean University www.aegean.gr/
     21. Omiros Training Group www.omiros.gr/
     22. Futurekids S.A S.A. www.futurekids.edu.gr/
     23. Piraeuos University, Department of Industrial Management and
Technology www.tex.unipi.gr/
Sector 12: Retail Web Site
     1. Atlantic Supermarket A.E.E www.atlantic.gr/
     2. Duty Free Shops, Greece S.A. www.dutyfreeshops.gr/
     3. Promota Hellas S.A www.promota.gr/
     4. AS Company S.A. www.ascompany.gr/index.jsp
     5. Expert Suppliers S.A. www.experthellas.gr/
     6. Glorybook-Economist Co.,Ε.Π.Ε. www.glorybook.gr/
     7. Metro ΑΕΒΕ www.metro.com.gr/
     8. Metropolis www.metropolis.gr/
     9. Moda Bagno N. Varveris S.A. www.modabagno.gr/
     10. Multirama Α.Β.Ε.Ε. www.multirama.gr/
     11. Oriflame www.oriflame.gr/
     12. Sara Lee Coffee and Tea Hellas S.A. www.bravo.gr/
     13. Tupperware Hellas S.A. www.tupperware.gr/
     14. Vadas A.E.Β.Ε.Ε. www.vardas.gr/
     15. Vassilias S.A. www.vassilias.gr/
     16. Eikona – Ixos A.E.E www.e-h.gr/
     17. Electornici Athens A.E.Ε. www.electroniki.gr/
     18. Interflora S.A www.interflora.gr/
     19. Kotsobolos A.E.Β.Ε. www.kotsovolos.gr
     20. Marinopoulos Group www.marinopoulos.gr/
     21. Plaisio Computers A.E.Β.Ε. www.plaisio.gr/
     22. Hatzigeorgiou S.A www.hatz.gr/
Sector 13: Tourism [Hotels] Web Site
     1. Louis Hotels www.louishotels.com/
     2. Loutraki Club Hotel Casino www.clubhotelloutraki.gr/
     3. Metropolitan Hotel www.chandris.gr/
     4. Park Hotel Athens www.athensparkhotel.gr/
     5. Rodos Park Suites Hotel www.rodospark.gr/
     6. Αstir Palace Vouliagmenis Α.Ξ.Ε. www.astir.gr/
     7. GEKE S.A. www.president.gr/
     8. Divans Hotels Group www.divanis.gr/
     9. Ιoniki Hotels S.A. www.ionianhe.gr/
     10. Capsis Tourist Group S.A. www.capsis.gr/
     11. Porto Carras www.portocarras.com/

143
A Proposed Template for the Evaluation of Web Design Strategies

Sector 1: Food, beverages and tobacco industry Web Site


Sector 14: Government Web Site
     1. Greek Parliament www.parliament.gr
     2. National Administration Center www.ekdd.gr
     3. Peiraias District www.nomarhiapeiraia.gr
     4. Ministry of development www.ypan.gr
     5. Ministry of foreign affairs www.mfa.gr
Sector 15: Business services Web Site
     1. Promaxon S.A www.procom.gr/
     2. Forever Print Recycling www.foreverprint.gr/
     3. Euroconsultants S.A www.euroconsultants.com.gr

This work was previously published in Emerging Markets and E-Commerce in Developing Economies, edited by K. Rouibah;
O. Khalil; A. Ella Hassanien, pp. 293-317, copyright 2009 by Information Science Reference (an imprint of IGI Global).

144
145

Chapter 1.9
A Review of Methodologies
for Analyzing Websites
Danielle Booth
Pennsylvania State University, USA

Bernard J. Jansen
Pennsylvania State University, USA

Abstract is available from Website visitors, and what can


be learned from studying such information? By
This chapter is an overview of the process of Web collecting various Web analytics metrics, such as
analytics for Websites. It outlines how visitor infor- number of visits,visitors, and visit duration, one can
mation such as number of visitors and visit duration develop key performance indicators (KPIs) – a ver-
can be collected using log files and page tagging. satile analytic model that measures several metrics
This information is then combined to create mean- against each other to define visitor trends. KPIs use
ingful key performance indicators that are tailored these dynamic numbers to get an in-depth picture of
not only to the business goals of the company run- visitor behavior on a site. This information allows
ning the Website but also to the goals and content businesses to align their Websites’ goals with their
of the Website. Finally, this chapter presents several business goals for the purpose of identifying areas
analytic tools and explains how to choose the right of improvement, promoting popular parts of the
tool for the needs of the Website. The ultimate goal site, testing new site functionality, and ultimately
of this chapter is to provide methods for increasing increasing revenue. This chapter covers the most
revenue and customer satisfaction through careful common metrics, different methods for gathering
analysis of visitor interaction with a Website. metrics, how to utilize key performance indicators,
best key practices, and choosing the right Web
analytics tool.
INTRODUCTION The first section addresses metrics, information
that can be collected from visitors on a Website. It
Web analytics is the measure of visitor behavior covers types of metrics based on what kind of data
on a Website. However, what kind of information is collected as well as specific metrics and how they
can be utilized. The following section discusses
DOI: 10.4018/978-1-59904-974-8.ch008

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
A Review of Methodologies for Analyzing Websites

the two main methods for gathering visitor in- METRICS


formation -- log files and page tagging. For each
method, this section covers the advantages and In order to understand the benefits of Website
disadvantages, types of supported information, analysis, one must first understand metrics – the
and examples for data format. Following this is different kinds of available user information.
a section on how to choose the key performance Although the metrics may seem basic, once col-
indicators (KPIs). This includes outlining several lected, they can be used to analyze Web traffic and
business strategies for integrating Web analytics improve a Website to better meet its overall goals.
with the rest of an organization as well as identify- According to Panalysis (http://www.panalysis.
ing the type of Website and listing several specific com/), an Australian Web analytics company, these
KPIs for each site type. The following section metrics generally fall into one of four categories:
provides the overall process and advice for Web site usage, referrers (or how visitors arrived at your
analytics integration, and the final section deals site), site content analysis, and quality assurance.
with what to look for when choosing analytics Table 1 shows examples of types of metrics that
tools as well as a comparison of several specific might be found in these categories.
tools. Finally, the conclusion discusses the future Although the type and overall number of met-
of Web analytics. rics varies with different analytics vendors, there
is still a common set of basic metrics common to
most. Table 2 outlines eight widespread types of

Table 1. Metrics categories (Jacka, n.d.)

Site Usage Referrers Site Content Analysis Quality Assurance


• Numbers of visitors and • Which websites are sending • Top entry pages • Broken pages or server errors
sessions visitors to your site • Most popular pages • Visitor response to errors
• How many people repeat- • The search terms people used • Top pages for single page view
edly visit the site to find your site sessions
• Geographic information • How many people place book- • Top exit pages
• Search Engine Activity marks to the site • Top paths through the site
• Effectiveness of key content

Table 2. Eight common metrics of Website analysis

Metric Description Category


Visitor Type Who is accessing the Website (returning, unique, etc.) Site Usage
Visit Length The total amount of time a visitor spends on the Website Site Usage
Demographics and System The physical location and information of the system used to Site Usage
Statistics access the Website
Internal Search Information Information on keywords and results pages viewed using a search Site Usage
engine embedded in the Website
Visitor Path The route a visitor uses to navigate through the Website Site Content Analysis
Top Pages The pages that receive the most traffic Site Content Analysis
Referrering URL and Keyword Which sites have directed traffic to the Website and which key- Referrers
Analysis words visitors are using to find the Website
Errors Any errors that occurred while attempting to retrieve the page Quality Assurance

146
A Review of Methodologies for Analyzing Websites

information that measure who is visiting a Website Visit Length


and what they do during their visits, relating each
of these metrics to specific categories. Also referred to as Visit Duration or Average Time
Each metric is discussed below. on Site (ATOS), visit length is the total amount of
time a visitor spends on a site during one session.
Visitor Type One possible area of confusion when using this
metric is handling missing data. This can be caused
Since analyzing Website traffic first became either by an error in data collection or by a ses-
popular in the 1990s with the Website counter, sion containing only one page visit or interaction.
the measure of Website traffic has been one of Since the visit length is calculated by subtracting
the most closely watched metrics. This metric, the time of the visitor’s first activity on the site
however, has evolved from merely counting the from the time of the visitor’s final activity, what
number of hits a page receives into counting the happens to the measurement when one of those
number of individuals who visit the Website. pieces of data is missing? According to the Web
There are two types of visitors: those who have Analytics Association, the visit length in such
been to the site before, and those who have not. cases is zero (Burby & Brown, 2007).
This difference is defined in terms of repeat and When analyzing the visit length, the measure-
new visitors. In order to track visitors in such a ments are often broken down into chunks of time.
way, a system must be able to determine individual StatCounter, for example, uses the following time
users who access a Website; each individual visitor categories:
is called a unique visitor. Ideally, a unique visitor
is just one visitor, but this is not always the case. • Less than 5 seconds
It is possible that multiple users access the site • 5 seconds to 30 seconds
from the same computer (perhaps on a shared • 30 seconds to 5 minutes
household computer or a public library). In ad- • 5 minutes to 20 minutes
dition, most analytic software relies on cookies • 20 minutes to 1 hour
to track unique users. If a user disables cookies • Greater than 1 hour (Jackson, 2007)
in their browser or if they clear their cache, the
visitor will be counted as new each time he or The goal of measuring the data in this way
she enters the site. is to keep the percentage of visitors who stay on
Because of this, some companies have instead the Website for less than five seconds as low as
begun to track unique visits, or sessions. A session possible. If visitors stay on a Website for such a
begins once a user enters the site and ends when a short amount of time it usually means they either
user exits the site or after a set amount of time of arrived at the site by accident or the site did not
inactivity (usually 30 minutes). The session data have relevant information. By combining this
does not rely on cookies and can be measured information with information from referrers and
easily. Since there is less uncertainty with visits, keyword analysis, one can tell which sites are
it is considered to be a more concrete and reli- referring well-targeted traffic and which sites are
able metric than unique visitors. This approach referring poor quality traffic.
is also more sales-oriented because it considers
each visit an opportunity to convert a visitor into Demographics and System Statistics
a customer instead of looking at overall customer
behavior (Belkin, 2006). The demographic metric refers to the physical
location of the system used to make a page request.

147
A Review of Methodologies for Analyzing Websites

This information can be useful for a Website that By analyzing internal search data, one can use
provides region-specific services. For example, the information to improve and personalize the
if an e-commerce site can only ship its goods to visitors’ experience.
people in Spain, any traffic to the site from outside
of Spain is irrelevant. In addition, region-specific Visitor Path
Websites also want to make sure they tailor their
content to the group they are targeting. Demo- A visitor path is the route a visitor uses to navigate
graphic information can also be combined with through a Website. Excluding visitors who leave
information on referrers to determine if a referral the site as soon as they enter, each visitor creates
site is directing traffic to a site from outside a a path of page views and actions while perusing
company’s regions of service. the site. By studying these paths, one can identify
System statistics are information about the any difficulties a user has viewing a specific area
hardware and software with which visitors access of the site or completing a certain action (such as
a Website. This can include information such as making a transaction or completing a form).
browser type, screen resolution, and operating According to the Web Analytics Association,
system. It is important that a Website be acces- there are two schools of thought regarding visi-
sible to all of its customers, and by using this tor path analysis. The first is that visitor actions
information, the Website can be tailored to meet are goal-driven and performed in a logical, linear
visitors’ technical needs. fashion. For example, if a visitor wants to pur-
chase an item, the visitor will first find the item,
Internal Search add it to the cart, and proceed to the checkout to
complete the process. Any break in that path (i.e.
If a Website includes a site-specific search utility, not completing the order) signifies user confusion
then it is also possible to measure internal search and is viewed as a problem.
information. This can include not only keywords The second school of thought is that visitor
but also information about which results pages actions are random and illogical and that the only
visitors found useful. The Patricia Seybold Group path that can provide accurate data on a visitor’s
(http://www.psgroup.com/) identifies the follow- behavior is the path from one page to the page
ing seven uses for internal search data: immediately following it. In other words, the only
page that influences visitor behavior on a Website
• Identify products and services for which is the one they are currently viewing. For example,
customers are looking, but that are not yet visitors on a news site may merely peruse the ar-
provided by the company. ticles with no particular goal in mind. This method
• Identify products that are offered, but of analysis is becoming increasingly popular
which customers have a hard time finding. because companies find it easier to examine path
• Identify customer trends. data in context without having to reference the
• Improve personalized messages by using entire site in order to study the visitors’ behavior
the customers’ own words. (Web Analytics Association, n. d.).
• Identify emerging customer service issues
• Determine if customers are provided with Top Pages
enough information to reach their goals.
• Make personalized offers. (Aldrich, 2006) Panalysis mentions three types of top pages: top
entry pages, top exit pages, and most popular

148
A Review of Methodologies for Analyzing Websites

pages. Top entry pages are important because the quality of keywords. Keyword quality is directly
first page a visitor views makes the greatest impres- proportional to revenue and can be determined by
sion about a Website. By knowing the top entry comparing keywords with visitor path and visit
page, one can make sure that page has relevant length (Marshall, n. d.). Good keywords will bring
information and provides adequate navigation to quality traffic and more income to your site.
important parts of the site. Similarly, identifying
popular exit pages makes it easier to pinpoint Errors
areas of confusion or missing content.
The most popular pages are the areas of a web- Errors are the final metric. Tracking errors has
site that receive the most traffic. This metric gives the obvious benefit of being able to identify and
insight into how visitors are utilizing the Website, fix any errors in the Website, but it is also useful
and which pages are providing the most useful to observe how visitors react to these errors. The
information. This is important because it shows fewer visitors who are confused by errors on a
whether the Website’s functionality matches up Website, the less likely visitors are to exit the site
with its business goals; if most of the Website’s because of an error.
traffic is being directed away from the main pages
of the site, the Website cannot function to its full
potential (Jacka, n. d.). GATHERING INFORMATION

Referrers and Keyword Analysis How does one gather these metrics? There are
two major methods for collecting data for Web
A referral page is the page a user visits immedi- analysis: log files and page tagging. Most current
ately before entering to a Website, or rather, a site Web analytic companies use a combination of the
that has directed traffic to the Website. A search two methods for collecting data. Therefore, it is
engine result page link, a blog entry mention- important to understand the strengths and weak-
ing the Website, and a personal bookmark are nesses of each.
examples of referrers. This metric is important
because it can be used to determine advertising Log Files
effectiveness and search engine popularity. As
always, it is important to look at this information The first method of metric gathering uses log files.
in context. If a certain referrer is doing worse than Every Web server keeps a log of page requests
expected, it could be caused by the referring link that can include (but is not limited to) visitor IP
text or placement. Conversely, an unexpected address, date and time of the request, request
spike in referrals from a certain page could be page, referrer, and information on the visitor’s
either good or bad depending on the content of Web browser and operating system. The same
the referring page. basic collected information can be displayed in
In the same way, keyword analysis deals a variety of ways. Although the format of the log
specifically with referring search engines and file is ultimately the decision of the company who
shows which keywords have brought in the most runs the Web server, the following four formats
traffic. By analyzing the keywords visitors use are a few of the most popular:
to find a page, one is able to determine what
visitors expect to gain from the Website and use • NCSA Common Log
that information to better tailor the Website to • NCSA Combined Log
their needs. It is also important to consider the • NCSA Separate Log

149
A Review of Methodologies for Analyzing Websites

• W3C Extended Log it does not require any changes to the Website
or any extra software installation to create the
The NCSA Common Log format (also known log files. Web servers automatically create these
as Access Log format) contains only basic informa- logs and store them on a company’s own servers
tion on the page request. This includes the client giving the company freedom to change their Web
IP address, client identifier, visitor username, analytics tools and strategies at will. This method
date and time, HTTP request, status code for the also does not require any extra bandwidth when
request, and the number of bytes transferred during loading a page, and since everything is recorded
the request. The Combined Log format contains server-side, it is possible to log both page request
the same information as the common log with successes and failures.
the following three additional fields: the referring Using log files also has some disadvantages.
URL, the visitor’s Web browser and operating One major disadvantage is that the collected
system information, and the cookie. The Separate data is limited to only transactions with the Web
Log format (or 3-Log format) contains the same server. This means that they cannot log informa-
information as the combined log, but it breaks tion independent from the servers such as the
it into three separate files – the access log, the physical location of the visitor. Similarly, while
referral log, and the agent log. The date and time it is possible to log cookies, the server must be
fields in each of the three logs are the same. Table specifically configured to assign cookies to visi-
3 shows examples of the common, combined, tors in order to do so. The final disadvantage is
and separate log file formats (notice that default that while it is useful to have all the information
values are represented by a dash “-“): stored on a company’s own servers, the log file
Similarly, W3C provides an outline for standard method is only available to those who own their
formatting procedures. This format differs from Web servers.
the first three in that it aims to provide for better
control and manipulation of data while still pro- Page Tagging
ducing a log file readable by most Web analytics
tools. The extended format contains user defined The second method for recording visitor activity
fields and identifiers followed by the actual en- is page tagging. Page tagging uses an invisible
tries, and default values are represented by a dash image to detect when a page has been success-
“-“ (Hallam-Baker & Behlendorf, 1999). Table 4 fully loaded and then uses JavaScript to send
shows an example of an extended log file. information about the page and the visitor back
There are several benefits of using system log to a remote server. According to Web Analytics
files to gather data for analysis. The first is that Demystified the variables used and amount of data

Table 3. NCSA Log comparison (IBM, 2004)

NCSA Common Log 125.125.125.125 - dsmith [10/Oct/1999:21:15:05 +0500] “GET /index.html HTTP/1.0” 200 1043
NCSA Combined Log 125.125.125.125 - dsmith [10/Oct/1999:21:15:05 +0500] “GET /index.html HTTP/1.0” 200 1043 “http://
www.ibm.com/” “Mozilla/4.05 [en] (WinNT; I)” “USERID=CustomerA;IMPID=01234”
NCSA Separate Log Common Log:
125.125.125.125 - dsmith [10/Oct/1999:21:15:05 +0500] “GET /index.html HTTP/1.0” 200 1043
Referral Log:
[10/Oct/1999:21:15:05 +0500] “http://www.ibm.com/index.html”
Agent Log:
[10/Oct/1999:21:15:05 +0500] “Microsoft Internet Explorer - 5.0”

150
A Review of Methodologies for Analyzing Websites

Table 4. W3C extended log file (Microsoft, 2005)

W3C Extended Log #Software: Microsoft Internet Information Services 6.0


#Version: 1.0
#Date: 2002-05-24 20:18:01
#Fields: date time c-ip cs-username s-ip s-port cs-method cs-uri-stem cs-uri-query sc-status sc-bytes cs-
bytes time-taken cs(User-Agent) cs(Referrer)
2002-05-24 20:18:01 172.224.24.114 - 206.73.118.24 80 GET /Default.htm - 200 7930 248 31 Mozilla/4.
0+(compatible;+MSIE+5.01;+Windows+2000+Server) http://64.224.24.114/

collected in page tagging are dependent on the Web Although log files and page tagging are two
analytics vendor. Some vendors stress short, easy distinct ways to collect information about the
to use page tags while others emphasize specific visitors to a Website, it is possible to use both
tags that require little post-processing. The best together, and many analytics companies provide
thing to look for with this method, however, is ways to use both methods to gather data. Even
flexibility – being able to use all, part, or none so, it is important to understand the strengths and
of the tag depending on the needs of the page weaknesses of both. Table 5 shows the advantages
(Peterson, 2004). and disadvantages of log file analysis and page
There are several benefits to using this method tagging.
of gathering visitor data. The first is speed of
reporting. Unlike a log file, the data received via The Problems with Data
page tagging is parsed as it comes in. This allows
for near real-time reporting. Another benefit is One of the most prevalent problems in Web ana-
flexibility of data collection. More specifically, lytics is the difficulty identifying unique users.
it is easier to record additional information about In order to determine repeat visitors, most Web
the visitor that does not involve a request to the analytic tools employ cookies that store unique
Web server. Examples of such information include identification information on the visitor’s personal
information about a visitor’s screen size, the price computer. Because of problems with users deleting
of purchased goods, and interactions within Flash or disabling cookies, however, some companies
animations. This is also a useful method of gather- have moved towards using Macromedia Flash
ing data for companies that do not run their own Local Shared Objects (LSOs). LSOs act like a
Web servers or do not have access to the raw log cookie, but standard browsers lack the tools re-
files for their site (such as blogs). quired to delete them, anti-spyware software does
There are also some disadvantages of page tag- not delete them because it does not see them as a
ging, most of which are centered on the extra code threat, and most users do not know how to delete
that must be added to the Website. This causes it to them manually. Awareness is growing, however,
use more bandwidth each time a page loads, and and Firefox and Macromedia are working against
it also makes it harder to change analytics tools LSOs and providing users with tools to delete
because the code embedded in the Website would them (Permadi, 2005).
have to be changed or deleted entirely. The final Sen, Dacin, and Pattichis (2006) cite various
disadvantage is that page tagging is only capable other problems with log data from Websites includ-
of recording page loads, not page failures. If a ing large data size and messy data. Problems with
page fails to load, it means that the tagging code large data size are caused by massive amounts of
also did not load, and there is therefore no way traffic to a Website and also the amount of informa-
to retrieve information in that instance. tion stored in each record. Records with missing

151
A Review of Methodologies for Analyzing Websites

Table 5. Log files vs. page tagging

Log Files Page Tagging


Advantages Disadvantages Advantages Disadvantages
Does not require changes to the Can only record interactions with Near real-time reporting Requires extra code added to the
Website or extra hardware instal- the Web server Website
lation
Does not require extra bandwidth Server must be configured to assign Easier to record additional Uses extra bandwidth each time
cookies to visitors information the page loads
Freedom to change tools with a Only available to companies who Able to capture visitor in- Can only record successful page
relatively small amount of hassle run their own Web servers teractions within Flash ani- loads, not failures
mations
Logs both page request successes Cannot log physical location Hard to switch analytic tools
and failures

IP addresses and changes to Website content cause is actionable. According to the Web Analytics
messy data. Even though the data may be hard to Association (McFadden, 2005), in order for a
work with at first, once it is cleaned up, it provides company to collect actionable data, it must meet
an excellent tool for Web analytics. these three criteria: “(1) the business goals must
be clear, (2) technology, analytics, and the busi-
ness must be aligned, and (3) the feedback loop
CHOOSING KEY PERFORMANCE must be complete” (Web Channel Performance
INDICATORS Management section, para. 3).
There are many possible methods for meet-
In order to get the most out of Web analytics, ing these criteria. One is Alignment-Centric
one must know how to choose effectively which Performance Management (Becher, 2005). This
metrics to analyze and combine them in mean- approach goes beyond merely reviewing past
ingful ways. This means knowing the Website’s customer trends to carefully selecting a few key
business goals and then determining which KPIs KPIs based on their future business objectives.
will provide the most insight. Even though a wealth of metrics is available from
a Website, this does not mean that all metrics
Knowing Your Business Goals are relevant to a company’s needs. Reporting
large quantities of data is overwhelming, so it is
Every company has specific business goals. Every important to look at metrics in context and use
part of the company works together to achieve them to create KPIs that focus on outcome and not
them, and the company Website is no exception. activity. For example, a customer service Website
In order for a Website to be beneficial, information might view the number of emails responded to on
gathered from its visitors must not merely show the same day they were sent as a measurement
what has happened in the past, but it must also be of customer satisfaction. A better way to measure
able to improve the site for future visitors. The customer satisfaction, however, might be to survey
company must have clearly defined goals for the the customers on their experience. Although this
future and use this information to support strate- measurement is subjective, it is a better repre-
gies that will help it achieve those goals. sentation of customer satisfaction because even
For a Website, the first step in achieving this if a customer receives a response the same day
is making sure the data collected from the site they send out an email, it does not mean that the

152
A Review of Methodologies for Analyzing Websites

experience was a good one (Becher, 2005). relevant performance metrics, the overall goals for
Choosing the most beneficial KPIs using the Website, and the performance measurements.
this method is achieved by following “The Four This method is similar to Alignment-Centric
M’s of Operational Management” as outlined by Performance Management in that it aims to aid
Becher (2005) which facilitate effective selection integration of the Website with the company’s
of KPIs: business objectives by involving major stakehold-
ers. The ultimate goals of OBPM are increased
• Motivate: Ensure that goals are relevant to confidence, organizational accountability, and
everyone involved. efficiency (Sapir 2004).
• Manage: Encourage collaboration and in-
volvement for achieving these goals. Identifying KPIs Based
• Monitor: Once selected, track the KPIs on Website Type
and quickly deal with any problems that
may arise. Unlike metrics, which are numerical representa-
• Measure: Identify the root causes of prob- tions of data collected from a Website, KPIs are
lems and test any assumptions associated tied to a business strategy and are usually mea-
with the strategy. sured by a ratio of two metrics. By choosing KPIs
based on the Website type, a business can save
By carefully choosing a few, quality KPIs to both time and money. Although Websites can have
monitor and making sure everyone is involved more than one function, each site belongs to at
with the strategy, it becomes easier to align a least one of the four main categories – commerce,
Website’s goals with the company’s goals because lead generation, content/media, and support/self
the information is targeted and stakeholders are service (McFadden, 2005). Table 6 shows com-
actively participating. mon KPIs for each Website type:
Another method for ensuring actionable data We discuss each Website type and related
is Online Business Performance Management KPIs below.
(OBPM) (Sapir, 2004). This approach integrates
business tools with Web analytics to help com- Commerce
panies make better decisions quickly in an ever-
changing online environment where customer data The goal of a commerce Website is to get visi-
is stored in a variety of different departments. The tors to purchase goods or services directly from
first step in this strategy is gathering all customer the site, with success gauged by the amount of
data in a central location and condensing it so that revenue the site brings in. According to Peter-
the result is all actionable data stored in the same son, “commerce analysis tools should provide
place. Once this information is in place, the next the ‘who, what, when, where, and how’ for your
step is choosing relevant KPIs that are aligned online purchasers (2004, p. 92).” In essence, the
with the company’s business strategy and then important information for a commerce Website is
analyzing expected versus actual results (Sapir who made (or failed to make) a purchase, what
2004). was purchased, when purchases were made, where
In order to choose the best KPIs and measure customers are coming from, and how customers are
the Website’s performance against the goals of a making their purchases. The most valuable KPIs
business, there must be effective communication used to answer these questions are conversion
between senior executives and online managers. rates, average order value, average visit value,
The two groups should work together to define the customer loyalty, and bounce rate (McFadden,

153
A Review of Methodologies for Analyzing Websites

Table 6. The four types of Websites and examples of associated KPIs (McFadden, 2005)

Website Type KPIs


Commerce      • Conversion rates
     • Average order value
     • Average visit value
     • Customer loyalty
     • Bounce rate
Lead Generation      • Conversion rates
     • Cost per lead
     • Bounce rate
     • Traffic concentration
Content/Media      • Visit depth
     • Returning visitor ratio
     • New visitor ratio
     • Page depth
Support/Self service      • Page depth
     • Bounce rate
     • Customer satisfaction
     • Top internal search phrases

2005). Other metrics to consider with a commerce It is common for commerce Websites to have
site are which products, categories, and brands are conversion rates around 0.5%, but generally good
sold on the site and internal site product search conversion rates will fall in the 2% range depend-
that could signal navigation confusion or a new ing on how a business structures its conversion
product niche (Peterson, 2004). rate (FoundPages, 2007). Again, the ultimate goal
A conversion rate is the number of users who is to increase total revenue. According to eVision,
perform a specified action divided by the total for each dollar a company spends on improving
of a certain type of visitor (i.e. repeat visitors, this KPI, there is $10 to $100 return (2007).
unique visitors, etc.) over a given period. Types The methods a business uses to improve their
of conversion rates will vary by the needs of the conversion rate (or rates), however, are different
businesses using them, but two common conver- depending on which target action that business
sion rates for commerce Websites are the order chooses to measure.
conversion rate (the percent of total visitors who Average order value (AOV) is a ratio of total
place an order on a Website) and the checkout order revenue to number of orders over a given
conversion rate (the percent of total visitors who period. This number is important because it allows
begin the checkout process). There are also many the analyzer to derive a cost for each transaction.
methods for choosing the group of visitors on There are several ways for a business to use this
which to base your conversion rate. For example, KPI to its advantage. One way is to break down
a business may want to filter visitors by exclud- the AOV by advertising campaigns (i.e. email,
ing visits from robots and Web crawlers (Ansari, keyword, banner ad etc.). This way, a business
Kohavi, Mason, & Zheng, 2001), or they may can see which campaigns are bringing in the best
want to exclude the traffic that “bounces” from customers and spend more effort refining their
the Website or (a slightly trickier measurement) strategies in those areas (Peterson, 2005). Overall,
the traffic that is determined not to have intent to however, if the cost of making a transaction is
purchase anything from the Website (Kaushik, greater than the amount of money customers spend
2006). for each transaction, the site is not fulfilling its goal.

154
A Review of Methodologies for Analyzing Websites

There are two main ways to correct this. The first Bounce rate is a measurement of how many
is to increase the number of products customers people arrive at a homepage and leave imme-
order per transaction, and the second is to increase diately. There are two scenarios that generally
the overall cost of purchased products. A good qualify as a bounce. In the first scenario, a visitor
technique for achieving this is through product views only one page on the Website. In the second
promotions (McFadden, 2005), but many factors scenario, a visitor navigates to a Website but only
influence how and why customers purchase what stays on the site for five seconds or less (Avinash,
they do on a Website. These factors are diverse 2007). This could be due to several factors, but in
and can range from displaying a certain security general, visitors who bounce from a Website are
image on the site (MarketingSherpa, 2007) to not interested in the content. Like average order
updating the site’s internal search (Young, 2007). value, this KPI helps show how much quality
Like many KPIs, improvement ultimately comes traffic a Website receives. A high bounce rate
from ongoing research and a small amount of may be a reflection of unintuitive site design or
trial and error. misdirected advertising.
Another KPI, average visit value, measures
the total number of visits to the total revenue. Lead Generation
This is a measurement of quality traffic important
to businesses. It is problematic for a commerce The goal for a lead generation Website is to obtain
site when, even though it may have many visi- user contact information in order to inform them
tors, each visit generates only a small amount of of a company’s new products and developments
revenue. In that case, even if the total number of and to gather data for market research; these sites
visits increased, it would not have a profound primarily focus on products or services that cannot
impact on overall profits. This KPI is also useful be purchased directly online. Examples of lead
for evaluating the effectiveness of promotional generation include requesting more information
campaigns. If the average visit value decreases by mail or email, applying online, signing up for
after a specific campaign, it is likely that the a newsletter, registering to download product
advertisement is not attracting quality traffic to information, and gathering referrals for a partner
the site. Another less common factor in this situ- site (Burby, 2004). The most important KPIs for
ation could be broken links or a confusing layout lead generation sites are conversion rates, cost
in a site’s “shopping cart” area. A business can per lead, bounce rate, and traffic concentration
improve the average visit value by using targeted (McFadden, 2005).
advertising and employing a layout that reduces Similar to commerce Website KPIs, a conver-
customer confusion. sion rate is the ratio of total visitors to the amount
Customer loyalty is the ratio of new to existing of visitors who perform a specific action. In the
customers. Many Web analytics tools measure this case of lead generation Websites, the most com-
using visit frequency and transactions, but there mon conversion rate is the ratio of total visitors
are several important factors in this measurement to leads generated. The same visitor filtering
including the time between visits (Mason, 2007). techniques mentioned in the previous section can
Customer loyalty can even be measured simply be applied to this measurement (i.e. filtering out
with customer satisfaction surveys (SearchCRM, robots and Web crawlers and excluding traffic that
2007). Loyal customers will not only increase bounces from the site). This KPI is an essential
revenue through purchases but also through tool in analyzing marketing strategies. Average
referrals, potentially limiting advertising costs lead generation sites have conversion rates rang-
(QuestionPro). ing from 5-6% and 17-19% conversion rates for

155
A Review of Methodologies for Analyzing Websites

exceptionally good sites (Greenfield, 2006). If Content/Media


the conversion rate of a site increases after the
implementation of a new marketing strategy, it Content/media Websites focus mainly on advertis-
indicates that the campaign was successful. If it ing, and the main goal of these sites is to increase
decreases, it indicates that the campaign was not revenue by keeping visitors on the Website longer
effective and might need to be reworked. and also to keep visitors coming back to the site.
Cost per lead (CPL) refers to the ratio of total In order for these types of sites to succeed, site
expenses to total number of leads, or how much content must be engaging and frequently updated.
it costs a company to generate a lead; a more If content is only part of a company’s Website, the
targeted measurement of this KPI would be the content used in conjunction with other types of
ratio of total marketing expenses to total number pages can be used to draw in visitors and provide
of leads. Like the conversion rate, CPL helps a a way to immerse them with the site. The main
business gain insight into the effectiveness of its KPIs are visit depth, returning visitors, new visitor
marketing campaigns. A good way to measure the percentage, and page depth (McFadden, 2005).
success of this KPI is to make sure that the CPL Visit depth (also referred to as depth of visit
for a specific marketing campaign is less than the or path length) is the measurement of the ratio
overall CPL (WebSideStory, 2004). Ideally, the between page views and unique visitors, or how
CPL should be low, and well-targeted advertising many pages a visitor accesses each visit. As a
is usually the best way to achieve this. general rule, visitors with a higher visit depth are
Lead generation bounce rate is the same mea- interacting more with the Website. If visitors are
surement as the bounce rate for commerce sites. only viewing a few pages per visit, it means that
This KPI is a measurement of visitor retention they are not engaged, and the effectiveness of the
based off total number of bounces to total number site is low. A way to increase a low average visit
of visitors; a bounce is a visit characterized by a depth is by creating more targeted content that
visitor entering the site and immediately leaving. would be more interesting to the Website’s target
Lead generation sites differ from commerce sites audience. Another strategy could be increasing
in that they may not require the same level of the site’s interactivity to encourage the users to
user interaction. For example, a lead generation become more involved with the site and keep
site could have a single page where users enter them coming back.
their contact information. Even though they only Unlike the metric of simply counting the
view one page, the visit is still successful if the number of returning visitors on a site, the return-
Website is able to collect the user’s information. ing visitor KPI is the ratio of unique visitors to
In these situations, it is best to base the bounce total visits. A factor in customer loyalty, this
rate solely off of time spent on the site. As with KPI measures the effectiveness of a Website to
commerce sites, the best way to decrease a site’s bring visitors back. A lower ratio for this KPI is
bounce rate is to increase advertising effectiveness best because a lower number means more repeat
and decrease visitor confusion. visitors and more visitors who are interested in
The final KPI is traffic concentration, or the and trust the content of the Website. If this KPI
ratio of the number of visitors to a certain area in is too low, however, it might signal problems in
a Website to total visitors. This KPI shows which other areas such as a high bounce rate or even
areas of a site have the most visitor interest. For click fraud. Click fraud occurs when a person
this type of Website, it is ideal to have a high or script is used to generate visits to a Website
traffic concentration on the page or pages where without having genuine interest in the site. Ac-
users enter their contact information. cording to a study by Blizzard Internet Marketing,

156
A Review of Methodologies for Analyzing Websites

the average for returning visitors to a Website is Support/Self Service


23.7% (White, 2006). As with many of the other
KPIs for content/media Websites, the best way Websites offering support or self-service are
to improve the returning visitor rate is by having interested in helping users find specialized an-
quality content and encouraging interaction with swers for specific problems. The goals for this
the Website. type of Website are increasing customer satisfac-
New visitor ratio is the measurement of new tion and decreasing call center costs; it is more
visitors to unique visitors and is used to determine cost-effective for a company to have visitors
if a site is attracting new people. When measuring find information through its Website than it is
this KPI, the age of the Website plays a role – newer to operate a call center. The KPIs of interest are
sites will want to attract new people. Similarly, visit length, content depth, and bounce rate. In
another factor to consider is if the Website is addition, other areas to examine are customer
concerned more about customer retention or satisfaction metrics and top internal search phrases
gaining new customers. As a rule, however, the (McFadden, 2005).
new visitor ratio should decrease over time as the Page depth for support/self service sites is the
returning visitor ratio increases. New visitors can same measurement as page depth content/media
be brought to the Website in a variety of different sites – the ratio of page views to unique visitors.
ways, so a good way to increase this KPI is to With support/self service sites, however, high page
try different marketing strategies and figure out depth is not always a good sign. For example, a
which campaigns bring the most (and the best) visitor viewing the same page multiple times may
traffic to the site. show that the visitor is having trouble finding
The final KPI for content/media sites is page helpful information on the Website or even that
depth. This is the ratio of page views for a spe- the information the visitor is looking for does
cific page and the number of unique visitors to not exist on the site. The goal of these types of
that page. This KPI is similar to visit depth, but sites is to help customers find what they need as
its measurements focus more on page popularity. quickly as possible and with the least amount of
Average page depth can be used to measure inter- navigation through the site (CCMedia, 2007).
est in specific areas of a Website over time and to The best way to keep page depth low is to keep
make sure that the interests of the visitors match visitor confusion low.
the goals of the Website. If one particular page on As with the bounce rate of other Website types,
a Website has a high page depth, it is an indication the bounce rate for support/self service sites re-
that that page is of particular interest to visitors. flects ease of use, advertising effectiveness, and
An example of a page in a Website expected to visitor interest. A low bounce rate means that qual-
have a higher page depth would be a news page. ity traffic is coming to the Website and deciding
Information on a news page is constantly updated that the site’s information is potentially useful.
so that, while the page is still always in the same Poor advertisement campaigns and poor Website
location, the content of that page is constantly layout will increase a site’s bounce rate.
changing. If a Website has high page depth in a Customer satisfaction deals with how the us-
relatively unimportant part of the site, it may signal ers rate their experience on a site and is usually
visitor confusion with navigation in the site or an collected directly from the visitors (not from log
incorrectly targeted advertising campaign. files), either through online surveys or through
satisfaction ratings. Although it is not a KPI
in the traditional sense, gathering data directly
from visitors to a Website is a valuable tool for

157
A Review of Methodologies for Analyzing Websites

figuring out exactly what visitors want. Customer KEY BEST PRACTICES
satisfaction measurements can deal with customer
ratings, concern reports, corrective actions, re- In this chapter, we have addressed which metrics
sponse time, and product delivery. Using these can be gathered from a Website, how to gather
numbers, one can compare the online experience them, and how to determine which information
of the Website’s customers to the industry average is important. But how can this help improve a
and make improvements according to visitors’ business? To answer this, the Web Analytics
expressed needs. Association provides nine key best practices to
Similarly, top internal search phrases applies follow when analyzing a Website (McFadden,
only to sites with internal search, but it can be used 2005). Figure 1 outlines this process.
to measure what information customers are most
interested in which can lead to improvement in Identify Key Stakeholders
site navigation. This information can be used to
direct support resources to the areas generating the A stakeholder is anyone who holds an interest
most user interest, as well as identify which parts in a Website. This includes management, site
of the Website users may have trouble accessing. developers, visitors, and anyone else who creates,
In addition, if many visitors are searching for a maintains, uses, or is affected by the site. In order
product not supported on the Website, it could be for the Website to be truly beneficial, it must inte-
a sign of ineffective marketing. grate input from all major stakeholders. Involving
Regardless of Website type, the KPIs listed people from different parts of the company also
above are not the only KPIs that can prove use- makes it more likely that they will embrace the
ful in analyzing a site’s traffic, but they provide a Website as a valuable tool.
good starting point. The main thing to remember
is that no matter what KPIs a company chooses, Define Primary Goals
they must be aligned with its business goals, and for Your Website
more KPIs do not necessarily mean better analysis
– quality is more important than quantity. To know the primary goals of a Website, one
must first understand the primary goals of its key

Figure 1. The best key practices of Web analytics

158
A Review of Methodologies for Analyzing Websites

stakeholders. This could include such goals as in- ing the ones that will provide little or no insight
creasing revenue, cutting expenses, and increasing into its goals. One can then use these metrics to
customer loyalty (McFadden, 2005). Once those determine which KPI you wish to monitor. As
goals have been defined, discuss each goal and mentioned in the previous section, the Website
prioritize them in terms of how the Website can type – commerce, lead generation, media/con-
most benefit the company. As always, beware of tent, or support/self service – plays a key role
political conflict between stakeholders and their in which KPIs are most effective for analyzing
individual goals as well as assumptions they may site traffic.
have made while determining their goals that may
not necessarily be true. By going through this pro- Identify and Implement
cess, a company can make sure that goals do not the Right Solution
conflict and that stakeholders are kept happy.
This step deals with finding the right Web analytics
Identify the Most Important technology to meet the business’s specific needs.
Site Visitors After the KPIs have been defined, this step should
be easy. The most important things to consider
According to Sterne, corporate executives are the budget, software flexibility and ease of
categorize their visitors differently in terms of use, and how well the technology will work with
importance. Most companies classify their most the needed metrics. McFadden suggests that it is
important visitors as ones who either visit the site also a good idea to run a pilot test of the top two
regularly, stay the longest on the site, view the vendor choices (McFadden, 2005). We will expand
most pages, purchase the most goods or services, on this topic further in the next section.
purchase goods most frequently, or spend the
most money (Sterne, n. d.). There are three types Use Multiple Technologies
of customers – (1) customers a company wants and Methods
to keep who have a high current value and high
future potential, (2) customers a company wants Web analytics is not the only method available for
to grow who can either have a high current value improving a Website. To achieve a more holistic
and low future potential or low current value and view of a site’s visitors, one can also use tools
high future potential, and (3) customers a company such as focus groups, online surveys, usability
wants to eliminate who have a low current value studies, and customer services contact analysis
and low future potential. The most important visi- (McFadden, 2005).
tor to a Website, however, is the one who ultimately
brings in the most revenue. Defining the different Make Improvements Iteratively
levels of customers will allow one to consider the
goals of these visitors. What improvements can When analyzing a Website’s data, it is helpful to
be made to the Website in order to improve their add gradual improvements to the Website instead
browsing experiences? of updating too many facets of the Website at
once. By doing this one can monitor if a singu-
Determine the Key lar change is an improvement or if it is actually
Performance Indicators hurting the site.

The next step is picking the metrics that will be


most beneficial in improving the site and eliminat-

159
A Review of Methodologies for Analyzing Websites

Hire and Empower a for answers about increased costs because


Full-Time Analyst of privacy threats or poor support offered
by free analytics tools.
It is important to put a person in charge of the data 2. Do you offer a software version of your tool?
once it is collected. According to the Web Analytics Generally, a business will want to look for
Association, a good analyst understands business a tool that is software based and can run on
needs (which means communicating well with the their own servers. If a tool does not have a
stakeholders), has knowledge of technology and software version but plans to make one in
marketing, has respect, credibility, and authority, the future, it shows insight into how prepared
and is already a company employee. Although they are to offer future products if there is
it may seem like hiring a full-time analyst is interest.
expensive, many experts agree that the return on 3. What methods do you use to capture data?
revenue should be more than enough compensation If you remember from the first section, there
to recoup the cost (McFadden, 2005). are two main ways to capture visitor data
from a Website – log files and page tagging.
Establish a Process of Ideally, one should look for a vendor that
Continuous Improvement offers both, but what they have used in the
past is also important. Because technology
Once the Web analysis process is decided upon, is constantly changing, look for a company
continuous evaluation is paramount. This means that has kept up with these changes in the
reviewing the goals and metrics and monitoring past by providing creative solutions.
new changes and features which are added to 4. Can you help me calculate the total cost
the Website. It is important that the improve- of ownership for your tool? The total cost
ments are adding value to the site and meeting of ownership for a Web analytics tool de-
expectations. pends on the specific company, the systems
they have in place, and the pricing of the
prospective Web analytics tool. In order to
SPECIFIC TOOLS make this calculation, one must consider the
following:
Choosing a Tool a. Cost per page view.
b. Incremental costs (i.e. charges for
Once the company decides what it wants out of overuse or advanced features).
the Web analysis, it is time to find the right tool. c. Annual support costs after the first
Kaushik outlines ten important questions to ask year.
Web analytics vendors (2007): d. Cost of professional services (i.e.
installation, troubleshooting, or
1. What is the difference between your tool customization).
and free Web analytics tools? Since the e. Cost of additional hardware you may
company who owns the Website will be need.
paying money for a service, it is important f. Administration costs (which includes
to know why that service is better than free the cost of an analyst and any additional
services (for example, Google Analytics). employees you may need to hire).
Look for an answer that outlines the features 5. What kind of support do you offer? Many
and functionality of the vendor. Do not look vendors advertise free support, but it is

160
A Review of Methodologies for Analyzing Websites

important to be aware of any limits that could Some examples of free and commercially
incur additional costs. It is also important available analytics tools are discussed below.
to note how extensive their support is and
how willing they are to help. Free Tools
6. What features do you provide that will al-
low me to segment my data? Segmentation One of the most popular free analytics tools on
allows companies to manipulate their data. the Web now is Google Analytics (previously Ur-
Look for the vendor’s ability to segment chin). Google Analytics (http://www.google.com/
your data after it is recorded. Many vendors analytics/) uses page tagging to collect information
use JavaScript tags on each page to segment from visitors to a site. In addition to expanding
the data as it is captured, meaning that the on the already highly regarded Urchin analytics
company has to know exactly what it wants tool, it also provides support for integrating other
from the data before having the data itself; analytic information (for example, WordPress
this approach is less flexible. and AdWords). Google Analytics reports many
7. What options do I have to export data into of the KPIs discussed in the previous sections
our system? It is important to know who including depth of visit, returning visitors, and
ultimately owns and stores the data and page depth.
whether it is possible to obtain both raw There is, however, concern about privacy is-
and processed data. Most vendors will not sues regarding Google Analytics because Google
provide companies with the data exactly as uses their default privacy policy for their analyt-
they need it, but it is a good idea to realize ics tools, but the company assures its Google
what kind of data is available before a final Analytics users that only account owners and
decision is made. people to whom the owners give permission will
8. Which features do you provide for integrat- have access to the data (Dodoo, 2006). Microsoft
ing data from other sources into your tool? also provides a free Web analytic software called
This question deals with the previous sec- Gatineau (Thomas, 2007).
tion’s Key Best Practice #6: Use Multiple
Technologies and Methods. If a company Paid Tools
has other data it wants to bring to the tool
(such as survey data or data from your ad InfoWorld provides an in-depth analysis compar-
agency), bring them up to the potential ing the top four Web analytic companies – Corem-
analytics vendor and see if it is possible to etrics, WebTrends, Omniture, and WebSideStory
integrate this information into their tool. HBX (Heck, 2005). They created a scoring chart
9. What new features are you developing that and measured each vendor on reporting, admin-
would keep you ahead of your competition? istration, performance, ease-of-use, support, and
Not only will the answer to this question value. Coremetrics received a score of 8.3 with its
tell how much the vendor has thought about highest ratings in administration and support. It is
future functionality, it will also show how a hosted service that offers special configurations
much they know about their competitors. for financial, retail, and travel services. WebTrends
10. Why did you lose your last two clients? Who also earned a score of 8.3 with its highest rating
are they using now? The benefits of this in reporting. This tool is expensive, but it offers
question are obvious -- by knowing how a wide range of performance statistics and both
they lost prior business, the business can be client and server hosting. Omniture is next in
confident that it has made the right choice. line with a score of 8.4 with its highest ratings in

161
A Review of Methodologies for Analyzing Websites

reporting and support. It is an ASP reporting ap- REFERENCES


plication that excels in providing relevant reports.
WebSideStory had the highest score of 8.7 with Aldrich, S. E. (2006, May 2). The Other Search:
its highest ratings in reporting, administration, Making the Most of Site Search to Optimize the
ease-of-use, and support. This tool is easy to use Total Customer Experience. Patricia Seybold
and is appropriate for many different types of Group. Retrieved March 7, 2007, from WebSide-
businesses. Story database.
Ansari, S., Kohavi, R., Mason, L., & Zheng, Z.
(2001). Integrating E-Commerce and Data Mining:
CONCLUSION Architecture and Challenges. IEEE International
Conference on Data Mining.
The first step in analyzing your Website and Web-
site visitors is understanding and analyzing your Avinash, A. (2007, June 26). Bounce Rate: Sexiest
business goals and then using that information Web Metric Ever? Retrieved December 2, 2007,
to carefully choose your metrics. In order to take from http://www.mpdailyfix.com/2007/06/
full advantage of the information gathered from bounce_rate_sexiest_web_metric.html.
your site’s visitors, you must consider alternative
Becher, J. D. (2005, March). Why Metrics-Centric
methods such as focus groups and online surveys,
Performance Management Solutions Fall Short.
make site improvements gradually, hire a full-time
DM Review Magazine. Retrieved March 7, 2007,
analyst, and realize that your site’s improvement is
from http://www.dmreview.com/article_sub.
a process and not a one-time activity. Using these
cfm?articleId=1021509.
key best practices and choosing the right analytics
vendor to fit your business will save your company Belkin, M. (2006, April 8). 15 Reasons why all
money and ultimately increase revenue. Unique Visitors are not created equal. Retrieved
As Web analytics continues to mature, the March 7, 2007, from http://www.omniture.com/
methods vendors use to collect information are blog/node/16.
becoming more refined. One article speculates
Burby, J. (2004, July 20). Build a Solid Foun-
that companies will find concrete answers to
dation With Key Performance Indicators, Part
the problems with cookies and unique visitors
1: Lead-Generation Sites. Retrieved March 7,
(Eisenberg, 2005). The Web analytics industry as
2007, from http://www.clickz.com/showPage.
a whole is also expanding. According to Eisenberg
html?page=3382981.
(2005), a recent Jupiter report predicts an increase
in the Web analytics industry – 20 percent annu- Burby, J., & Brown, A. (2007). Web Analytics
ally. More and more businesses are realizing the Definitions. Retrieved October 30, 2007, from
benefits of critically analyzing their Website traffic http://www.webanalyticsassociation.org/attach-
and are taking measures to improve their profits ments/committees/5/WAA-Standards-Analytics-
based off these numbers. Regardless of business Definitions-Volume-I-20070816.pdf.
size and objective, an effective Web analytics
CCMedia. (2007, August 30). How to Obtain a
strategy is becoming increasingly essential for
Cost-effective Operational Model for Support/
online success.
Self-service Websites? Retrieved December 5,
2007, from www.webnibbler.com/en/WhitePaper/
Online%20Support%20Website.pdf.

162
A Review of Methodologies for Analyzing Websites

Dodoo, M. (2006, March 3). Privacy & Google Jacka, R. Getting Results From Your Website.
Analytics. Retrieved March 7, 2007, from http:// Retrieved October 30, 2007, from http://www.
www.marteydodoo.com/2006/05/03/privacy- panalysis.com/downloads/gettingresults.pdf
google-analytics/.
Jackson, M. (2007, January 22). Analytics: Deci-
Eisenberg, B. (2005, April 1). Web Analytics: phering the Data. Retrieved March 7, 2007, from
Exciting Times Ahead. Retrieved March 7, http://www.ecommerce-guide.com/resources/
2007, from http://www.clickz.com/showPage. article.php/3655251
html?page=3493976
Kaushik, A. (2006, November 13). Excellent Ana-
eVision. (2007, September 27). Websites that lytics Tip #8: Measure the Real Conversion Rate &
convert visitors into customers: Improving the ‘Opportunity Pie.’ Retrieved November 3, 2007,
ability of your Website to convert visitors into from http://www.kaushik.net/avinash/2006/11/
inquiries, leads, and new business. Retrieved excellent-analytics-tip-8-measure-the-real-con-
March 7, 2007, from http://www.evisionsem.com/ version-rate-opportunity-pie.html
marketing/webanalytics.htm
Kaushik, A. (2007, January 23). Web Analytics Tool
FoundPages. (2007, October 25). Increasing Con- Selection: 10 Questions to ask Vendors. Retrieved
version Rates. Retrieved October 31, 2007, from March 7, 2007, from http://www.kaushik.net/
http://www.foundpages.com/calgary-internet- avinash/2007/01/web-analytics-tool-selection-
marketing/search-conversion.html 10-questions-to-ask-vendors.html
Greenfield, M. (2006, January 1). Use Web Ana- MarketingSherpa. (2007, October 20). Security
lytics to Improve Profits for New Year: Focus on Logo in Email Lifts Average Order Value 28.3%.
four key statistics. Retrieved March 7, 2007, Retrieved December 4, 2007, from https://www.
from http://www.practicalecommerce.com/ marketingsherpa.com/barrier.html?ident=30183
articles/132/Use-Web-Analytics-to-Improve-
Marshall, J. Seven Deadly Web Analytics Sins.
Profits-for-New-Year/
Retrieved March 7, 2007, from http://www.click-
Hallam-Baker, P. M., & Behlendorf, B. (1999,Feb- tracks.com/insidetrack/articles/7_deadly_weba-
ruary 4). Extended Log File Format. Retrieved nalytics_sins01.php
March 7, 2007, from http://www.w3.org/TR/
Mason, N. (2007, February 6). Customer Loy-
WD-logfile.html
alty Improves Retention. Retrieved March 7,
Heck, M. (2005, February 18). Chart 2007, from http://www.clickz.com/showPage.
Your Website’s Success. Retrieved March html?page=3624868
7, 2007, from http://www.infoworld.com/
McFadden, C. (2005, July 6). Optimizing the
Omniture_SiteCatalyst_11/product_56297.
Online Business Channel with Web Analytics.
html?view=1&curNodeId=0&index=0
Retrieved March 7, 2007, from http://www.We-
IBM. (2004, May 19). Log File Formats. Retrieved banalyticsassociation.org/en/art/?9
October 29, 2007, from http://publib.boulder.ibm.
Microsoft. (2005, August 22). W3C Extended
com/tividd/td/ITWSA/ITWSA_info45/en_US/
Log File Examples. Retrieved March 7,
HTML/guide/c-logs.html
2007, from http://technet2.microsoft.com/Win-
dowsServer/en/library/b5b8a519-8f9b-456b-
9040-018358f2c0c01033.mspx?mfr=true

163
A Review of Methodologies for Analyzing Websites

Permadi, F. (2005, June 19). Introduction to Web Analytics Association. Onsite Behavior - Path
Flash Local Shared-Object. Retrieved March Analysis. Retrieved March 7, 2007, from http://
7, 2007, from http://www.permadi.com/tutorial/ www.Webanalyticsassociation.org/attachments/
flashSharedObject/index.html contentmanagers/336/1%20Path%20AnAnalys.
doc
Peterson, E. T. (2004). Web Analytics Demystified.
Celilo Group Media. WebSideStory. (2004). Use of Key Performance
Indicators in Web Analytics. Retrieved December
Peterson, E. T. (2005, July 31). Average Order
2, 2007, from www.4everywhere.com/documents/
Value. Retrieved November 3, 2007, from Web
KPI.pdf
Analytics Demystified Blog Website: http://blog.
webanalyticsdemystified.com/weblog/2005/07/ White, K. (2006, May 10). Unique vs. Re-
average-order-value.html turning Visitors Analyzed. Retrieved March
7, 2007, from http://newsletter.blizzardin-
QuestionPro. Measuring Customer Loyalty and
ternet.com/unique-vs-returning-visitors-
Customer Satisfaction. Retrieved November 21,
analyzed/2006/05/10/#more-532
2007, from http://www.questionpro.com/akira/
showArticle.do?articleID=customerloyalty. Young, D. (2007, August 15). Site Search: In-
creases Conversion Rates, Average Order Value
Sapir, D. (2004, August). Online Analytics and
And Loyalty. Practical Ecommerce, Retrieved
Business Performance Management. BI Report.
November 15, 2007, from http://www.practi-
Retrieved March 7, 2007, from http://www.
calecommerce.com/articles/541/Site-Search-
dmreview.com/editorial/dmreview/print_action.
Increases-Conversion-Rates-Average-Order-
cfm?articleId=1008820
Value-And-Loyalty/
Search, C. R. M. (2007, May 9). Measuring
Customer Loyalty. Retrieved November 4,
2007, from http://searchcrm.techtarget.com/
KEY TERMS
general/0,295582,sid11_gci1253794,00.html
Sen, A., Dacin, P. A., & Pattichis, C. (2006, No- Abandonment Rate: KPI that measures the
vember). Current trends in Web data analysis. percentage of visitors who got to that point on the
Communications of the ACM, 49(11), 85–91. site but decided not to perform the target action.
doi:10.1145/1167838.1167842 Alignment-Centric Performance Man-
agement: Method of defining a site’s business
Sterne, J. 10 Steps to Measuring Website goals by choosing only a few key performance
Success. Retrieved March 7, 2007, from indicators.
http://www.marketingprofs.com/login/join. Average Order Value: KPI that measures the
asp?adref=rdblk&source=/4/sterne13.asp total revenue to the total number of orders.
Thomas, I. (2007, January 9). The rumors are Average Time on Site (ATOS): See visit
true: Microsoft ‘Gatineau’ exists. Retrieved length.
March 7, 2007, from http://www.liesdamnedlies. Checkout Conversion Rate: KPI that mea-
com/2007/01/the_rumors_are_.html sures the percent of total visitors who begin the
checkout process.
Commerce Website: A type of Website where
the goal is to get visitors to purchase goods or
services directly from the site.

164
A Review of Methodologies for Analyzing Websites

Committed Visitor Index: KPI that measures Metrics: Statistical data collected from a
the percentage of visitors that view more than Website such as number of unique visitors, most
one page or spend more than 1 minute on a site popular pages, etc.
(these measurements should be adjusted accord- New Visitor: A user who is accessing a Website
ing to site type). for the first time.
Content/Media Website: A type of Website New Visitor Percentage: KPI that measures
focused on advertising. the ratio of new visitors to unique visitors.
Conversion Rate: KPI that measures the per- Online Business Performance Manage-
centage of total visitors to a Website that perform ment (OBPM): Method of defining a site’s
a specific action. business goals that emphasizes the integration
Cost Per Lead (CPL): KPI that measures of business tools and Web analytics to make bet-
the ratio of marketing expenses to total leads and ter decisions quickly in an ever-changing online
shows how much it costs a company to generate environment.
a lead. Order Conversion Rate: KPI that measures
Customer Satisfaction Metrics: KPI that the percent of total visitors who place an order
measures how the users rate their experience on on a Website.
a site. Page Depth: KPI that measures the ratio of
Customer Loyalty: KPI that measures the page views for a specific page and the number of
ratio of new to existing customers. unique visitors to that page.
Demographics and System Statistics: A Page Tagging: Method of gathering metrics
metric that measures the physical location and that uses an invisible image to detect when a
information of the system used to access the page has been successfully loaded and then uses
Website. JavaScript to send information about the page and
Depth of Visit: KPI that measures the ratio the visitor back to a remote server.
between page views and visitors. Prospect Rate: KPI that measures the percent-
Internal Search: A metric that measures in- age of visitors who get to the point in a site where
formation on keywords and results pages viewed they can perform the target action (even if they
using a search engine embedded in the Website. do not actually complete it).
Key Performance Indicator (KPI): A com- Referrers and Keyword Analysis: A metric
bination of metrics tied to a business strategy. that measures which sites have directed traffic
Lead Generation Website: A type of Website to the Website and which keywords visitors are
that is used to obtain user contact information in using to find the Website.
order to inform them of a company’s new products Repeat Visitor: A user who has been to a
and developments, and to gather data for market Website before and is now returning.
research. Returning Visitor: KPI that measures the
Log File: Log kept by a Web server of informa- ratio of unique visitors to total visits.
tion about requests made to the Website including Search Engine Referrals: KPI that measures
(but not limited to) visitor IP address, date and the ratio of referrals to a site from specific search
time of the request, request page, referrer, and engines compared to the industry average.
information on the visitor’s Web browser and Single Access Ratio: KPI that measures the
operating system. ratio of total single access pages (or pages where
Log File Analysis: Method of gathering met- the visitor enters the site and exits immediately
rics that uses information gathered from a log file from the same page) to total entry pages.
to gather Website statistics. Stickiness: KPI that measures how many

165
A Review of Methodologies for Analyzing Websites

people arrive at a homepage and proceed to tra- Visit Length: A metric that measures total
verse the rest of the site. amount of time a visitor spends on the Website.
Support/Self Service Website: A type of Web- Visit Value: KPI that measures the total number
site that focuses on helping users find specialized of visits to total revenue.
answers for their particular problems. Visitor Path: A metric that measures the route
Top Pages: A metric that measures the pages a visitor uses to navigate through the Website.
in a Website that receive the most traffic. Visitor Type: A metric that measures users
Total Bounce Rate: KPI that measures the who access a Website. Each user who visits the
percentage of visitors who scan the site and then Website is a unique user. If it is a user’s first time
leave. to the Website, that visitor is a new visitor, and
Traffic Concentration: KPI that measures the if it is not the user’s first time, that visitor is a
ratio of number of visitors to a certain area in a repeat visitor.
Website to total visitors. Web Analytics: The measurement of visitor
Unique Visit: One visit to a Website (regard- behavior on a Website.
less of if the user has previously visited the site);
an alternative to unique visitors.
Unique Visitor: A specific user who accesses
a Website.

This work was previously published in Handbook of Research on Web Log Analysis, edited by B. Jansen; A. Spink; I. Taksa,
pp. 143-164, copyright 2009 by Information Science Reference (an imprint of IGI Global).

166
167

Chapter 1.10
Basics to Develop Web Services
for Human Resources
Roman Povalej
University of Karlsruhe (TH), Germany

Peter Weiß
University of Karlsruhe (TH), Germany

INTRODUCTION HR managers of a company or an organization


are challenged through the need to formalize skills
In this article certain pillars as basics are presented requirements and to continuously monitor the skills
being necessary to develop Web services (W3C, demand inside the company. Obtaining ICT skills
2007) supporting human resource (HR) processes are not a one-time event. Technological change ad-
like assessing, hiring, modeling information sys- vances at a high speed and requires that skills need
tems, staffing, and so forth; by the help of these continually to be kept up-to-date and relevant (The
Web services. Current HR information systems in European e-Skills Forum [ESF], 2005).
general do not adequately support tasks related to During the last years, new concepts have emerged
cross-organizational or global skills and competence which intend to empower learners and individuals
management. In the following, the topic is presented to steer learning processes to a large extent on their
which relates to knowledge management especially own. Learning objectives tend to be increasingly
to “communities of practice,” as well as related individual in character (ESF, 2005).
topics such as e-skills and ICT (information and In this context, providing an appropriate infra-
communication technologies) professionalism; the structure which supports the continuing profes-
latter currently being broadly discussed by experts sional development (CPD) of employees is today
in Europe. a key issue. CPD processes require a respective
infrastructure encompasses besides qualifications,
DOI: 10.4018/978-1-59904-883-3.ch013

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Basics to Develop Web Services for Human Resources

skills/competence frameworks and body of knowl- job profile, which mirrors ideally the companies
edge, as well required standards for competence, core competences; but how can this be efficiently
skills, and appropriate career and development done or supported by a concept method and/or
services. Standards encompass educational and tool, how can be potential employees be addressed
industry-oriented performance standards which independently of their national origin? Further
in turn are expressed preferably through a com- scenarios can be derived from Figure 1.
mon language as competence and skills standards. Noticeably, information about requirements
The governance and administration of the CPD is exchanged amongst various players along the
process require the availability of flexible and value chain. Hence, looking at ICT skills re-
personalized certification services which offer quirements necessitates looking at the belonging
the formal validation of individuals’ learning processes as a whole. This relates to a common
achievements independent of where and how they language which allows exchanging information
were acquired. amongst the various players involved and ex-
Recruitment and internal mobility are typical change of data stored in respective HR-systems.
use cases to identify the right people wanted, by The latter, in particular, motivates data services
matching their assessment outcomes with job which facilitate sharing of data required for the
positions requested. In case a company needs new execution of tasks along the HR processes (such
ICT competence, the HR manager in charge has as recruitment of ICT professionals). Typically,
to draw up a job advertisement, which describes systems do not have the required interoperability
the required knowledge, skills, competences because the applied data models differ in structure
(KSC), and qualifications for the specific job and semantics. In the following, we look at how
role. The information is generally structured in a to achieve better interoperability of systems and

Figure 1. Supply and demand of e-skills

168
Basics to Develop Web Services for Human Resources

an enhanced sharing of information about ICT ployees’ market value and often they rank a key
skills requirements typically stored in job or ICT role in the hiring process.
worker profiles. In the first place, ICT companies, but actually
Competence-based professional profiles are a all companies, employ people with ICT skills as
referred concept to define and exchange informa- well as people with a range of other relevant skills.
tion about respective professional standards. The Often the number of ICT practitioners in such
profile is the description of competence required to ICT user organizations (e.g., banks, manufactur-
operate on a process, on a service, or for a definite ing companies, and airlines) is greater than the
role, inside a direction or a function, a working number employed in ICT companies.
team, or a project. It shows as much explicit as New competency-based systems of certifica-
possible which type of know-how needs an or- tion of individual employees and/or others were,
ganization has in terms of competence. In a most are, and further on, will be established within the
diffused configuration, the competence based pro- information technology field and offered on the
file is defined with reference to an organizational market; this is mainly driven by ICT vendors, the
position or role. The competence based profile expected profit is highly valued.
of a role has to be built up without any reference Competence encompasses “[…] the demon-
to single persons that occupied this role or are stration of relevant, up-to-date skills and capabili-
candidates to cover the role in the future. ties appropriate to a particular task or role with
In the following, the article looks at a com- practical experience to complement theoretical
mon language and ICT worker profiles as a con- knowledge” (Hughes & Thompson, 2007). This
stituent part and central element of an adequate includes as well, necessary soft skills such as inter-
management of HR processes in general and CPD personal and communication skills as well as an
processes specifically. The article presents a con- understanding of the business domain. We argue
ceptual model which allows expanding existing that performance components are an adequate
barriers of HR systems. Web services are seen means to develop necessary skills and to avoid
as an appropriate concept to implement required that qualification measures coincide. Figure 1
functionality and processes. shows the supply and demand of e-skills.
E-skills present a mainly political driven joint
initiative of stakeholders in Europe such as the
BACKGROUND European Commission, ICT professionals asso-
ciations, and international ICT vendor industry
The competitiveness of European industry depends associations. Subsequently, we argue why e-skills
on both the effective use of ICT for industrial and require joint activities and concerted actions.
business processes and the KSC and qualifications Each stakeholder determines their own needed
of existing and new employees (ESF, 2005). and relevant e-skills requirements (profile A).
Today, nearly every area of economic activity Education institutions determine their curricula
is affected by ICT and the pace of technological (profile B) which consist of e-skills which should
change and short technology life cycles makes be taught to the students. Industry and employers
these the most dynamic of occupations (Cedefop, determine their e-skills requirements (profile A) to
2005); scilicet companies as well as ICT em- manage their tasks and challenges in the economy.
ployees (and also others) have an ongoing need Sometimes the requirements of these two groups
to keep their KSC and qualifications up-to-date do not exactly match, for example, parts of the
to prevail against competitors. In respect thereof, curricula are in general concepts, methods, and
ICT certifications are useful to validate ICT em- tools about databases; however, the industry

169
Basics to Develop Web Services for Human Resources

demands product-related knowledge about a common semantics to enable communication of


specific database, often they do not notice that the and automated data exchange between various
parts of the curricula will be actual on the long applications in HR systems.
run, whereas product-related knowledge about We see that electronic data interchange and
one specific database could be shortly unusable, integration has potential to make internal HR
because the belonging software product is no management processes more flexible and faster,
longer offered and used. After education, every as well as more efficient and effective.
person becomes part of the e-skills stock (profile
C) with their own portfolio (profile D), and with
their participation at the labor market they build CONCEPTUAL DATA MODEL
the e-skills supply (profile C+D). The relevant
job profiles are managed within companies’ skills During schooling and education, for (the first)
framework (profile A’) and are influenced by the letter of application, for a successful professional
e-skills requirements (profile A), e-skills stocks advancement, and more: the whole working life
(profile C and profile D), and e-skills supply (pro- belongs to the crucial points KSC and qualifica-
file C+D). It is a big challenge to compare and/ tions. There are standardized data models describ-
or to match these (job) profiles (profile A, B, C, ing competences in a re-usable way: IEEE Reus-
D, C+D, A’) among themselves because of lack able Competency Definition (RCD, 2007) and
of a common language. The degree of difficulty HR-XML (2004) (further information in Povalej
is much higher on the European level. and Weiß, 2007). When certain competences are
Web service adoption in the industry has al- once defined they can be re-used again and again,
lowed organizations to share information with their for example, in different job advertisements or
partners, providers, and customers in a standard- describing several job profiles. Further on, em-
ized manner. Web services provide a successful ployees can use the already defined competences
way to communicate distributed applications in a to describe their own ones.
platform-independent and loosely coupled man- Here are some clues based on the results of the
ner. The integration of applications is one of the analyses of these two standardized data models
main challenges when building ICT solutions. which could be extracted for our conceptual data
Integration is often achieved using costly cus- model (Povalej & Weiß, 2007): the re-usability and
tomized solutions for every pair of applications. the flexibility are strengthened through an object
Web service technologies are a set of standards oriented approach; one competency consists of the
that allow software interfaces to be defined using components skills, context, and proficiency level;
XML (extensible markup language) as the mes- context and proficiency level has to be clearly
sage format and the Internet infrastructure for definable which is a precondition for profile match-
message transport. While lowering costs, Web ing and domain-wide re-usability; free description
service technologies by themselves do not ensure area out of RCD is needed and it represents the
that two businesses use the same data structures smallest skills unit; additional, the component
or business protocols; neither do they provide grade is introduced. Figure 2 displays the connec-
the means to resolve potential conflicts (Lausen, tions and dependencies between relevant objects
Brujin, Keller, & Lara, 2006). In order to deal or classes of a job profile model.
with existing interoperability issues, ontology- A job profile describes among others, KSC and
based models for describing Web services are qualifications which employees have or should
proposed. Hence, Web services are in the wider have to cope with work tasks successfully within
context of analysis, we will look at issues of their working areas. Hence, there are two kinds

170
Basics to Develop Web Services for Human Resources

Figure 2. Example of a job profile model

of profiles: (1) the required profile contains all CONCEPTUAL META-FRAMEWORK


information about KSC and qualification which
employees should have to manage their tasks in a However, in order to achieve transparency and
company and (2) the acquired profile contains all comparability of ICT worker profiles, data struc-
information about KSC and qualification which tures need to be analyzed and their semantics or
employees gained in their life. Every (job) profile meta data needs to be described. Subsequently,
belongs to at least one working area. Further, on we look at the modeling of profile structure and
the (job) profile exists of one or more profile semantics. Meta data is preferably described and
components—each profile component represents stored in the form of ontologies. Preparatory
experience, proficiency, talent, or qualification. work to create ontologies requires analyzing the
Every profile component exists of none, one, or underlying data structure in a formal way, which
many KSC objects, whereby every component then determines our structure of data sets stored in
contains none or one level. profiles. We present the results of our analysis of
The code in Figure 3 is an extract from a HR- meta data applied in major ICT skills/competence
XML schema describing competence and level frameworks.
belonging to a job profile. The smallest building block or unit of qualifica-
Figure 4 displays exemplarily the component tions has to be defined (CEN/ISSS, 2005; Weiß,
profile of our conceptual model (Povalej & Weiß, Povalej, & Stucky, 2005; Weiß, Stucky, Dolan,
2007). & Bumann, 2005).

171
Basics to Develop Web Services for Human Resources

Figure 3. Exemplary extract from a HR-XML schema

Figure 4. Exemplary component profile (Source: Povalej & Weiß, 2007)

In the following section, the developed matrix with the job role q as follows (notation: Figure 5)
of profiles P is described (further information: (our approach corresponds to the central concepts
Weiß, Povalej, & Stucky, 2005; Povalej & Weiß, or terms suggested by Cedefop, 2003):
2007). In our approach, a profile constitutes on
Pq = Lq = (lqij ) = (wai , qc j )
i =1...m i =1...m
three sub-systems to be defined and specified by
a respective modeler/developer: 1) the general j =1...n j =1...n
(1)
description of qualification component (QC) (e.g.,
constructs described as learning outcome (LOC)
by KSC), 2) the specific description of the working
area (WA), job role, function, business process, and FUTURE TRENDS
so forth, and 3) the performance level required.
These three elements combined, define the quali- CEN/ISSS (2007) is working on a European e-
fication unit (QU) as the smallest describable and competence framework. A team of experts has
measurable unit within a qualification or profile been appointed and will produce guidelines for
P. These results are summarized in the definition the developments of a European e-competence
of ICT worker sector profile P. An ICT worker framework for consensus building. The team
profile is defined as the selection of the tupel involves experts from stakeholders of leading
(WA, QC, L) and can be described as matrix P frameworks such as SFIA in UK, AITTS in

172
Basics to Develop Web Services for Human Resources

Figure 5. Overview of applied notation (Source: Weiß, Povalej, & Stucky, 2005)

Germany, CIGREF in France, and ECCO in of information between various systems can be
Italy. The European e-competence framework seen as particular challenges.
is addressed to HR managers, ICT managers, Portfolios as well as e-portfolios are closely
and ICT practitioners/ professionals as well as linked to (job) profiles. Hence, consistently, there
training institutions and educational bodies. The should be a more detailed review about the con-
framework intends to contribute to the daily work nection between these areas. Butler (2006) made
of HR managers as an international planning and the following recommendation for the success
development tool. of e-portfolios: “Institutions need to be aware of
The results will be a basic structure for the de- the impact that an electronic portfolio develop-
velopment of vocabularies and skills/competence ment will have. Electronic portfolios need to be
dictionaries. Building vocabularies and dictionar- an integral part of a programme of study, not an
ies requires in depth domain knowledge of ICT ‘added-on’ assessment, which may necessitate the
competences in industry. Hence, the European review and restructuring of courses. The type of
e-competence framework delivers the content portfolio required, its purpose and its audience
which populates profiles and instantiates described need to be clearly articulated.”
meta data structure.
KSC described in coherence with the frame-
work structure allows exchanging and reusing this CONCLUSION
information in various HR and CPD processes.
The framework builds a grid which is spanned A detailed consideration of the supply and de-
mainly by the dimensions competence areas and mand of e-skills (or further knowledge, skills,
respective descriptors to specify attitudes, skills, competences, and qualifications) on the labor
and knowledge. Furthermore, levels and the ori- market is crucial for the development of a com-
entation towards ICT processes are considered. mon European framework. It has been pointed out
The next steps require to populate the grid and to that a common language and a shared common
build dictionaries and catalogues of competence framework are needed.
descriptions. The reuse of data and the exchange Further, this article accentuated the formal

173
Basics to Develop Web Services for Human Resources

description of the smallest unit as a building block Cedefop (The European Centre for the Devel-
of qualifications and for analyzing ICT worker pro- opment of Vocational Training). (2003). Credit
files. Qualification units are the smallest element Transfer In Vet NOVEMBER 2002-OCTOBER
to be scrutinized in respective competence assess- 2003. Author: Isabelle le Mouillour, University
ments and in the certification process. By the help of Kassel, Germany. Retrieved October 15, 2007,
of our developed and presented conceptual data from http://communities.trainingvillage.gr/
model (derived from standardized data models) as credittransfer-eqf
well as our conceptual framework (derived from
Cedefop (The European Centre for the Develop-
existing frameworks using the meta-level), ICT
ment of Vocational Training). (2005). European
competence profiles, or in general, profiles can be
reference levels for education and training promot-
modeled at the meta-, reference, and application
ing credit transfer and mutual trust. Study com-
level (Weiß, Povalej, & Stucky, 2005).
missioned to the Qualifications and Curriculum
Hence, there are now the necessary basics to
Authority, England. Authors: Mike Coles, Tim
develop further Web services for human resources.
Oates. Cedefop Panorama series; 109, Luxem-
Central aims are to create the “right” conditions to
bourg: Office for Official Publications of the
attract talented people by respective career guid-
European Communities, 2005.
ance, information services, and by highlighting
flexible entry points to the ICT profession—these CEN/ISSS ICT-Skills Workshop Phase 2. (2005).
all offered as Web services. CEN Workshop Agreement (CWA). Retrieved
The next steps are to develop further the meta- October 15, 2007, from http://www.uninfo.polito.
framework for KSC and qualifications. Further on, it/WS_ICT-Skills
our up to now work is also the basic for further
CEN/ISSS ICT-Skills Workshop Phase 3. (2007).
development of KSC and qualification catalogues
Retrieved October 15, 2007, from http://www.
and repositories. At the same time the conceptual
cen.eu/CENORM/BusinessDomains/business-
models will be implemented to real world systems,
domains/isss/activity/wsict-skills.asp
considering these catalogues and repositories,
and also in the future, Web services for a higher ESF (The European E-skills Forum). (2005).
transparency, comparability, transferability, and E-skills for Europe: Towards 2010 and Beyond.
interchangeability. Luxembourg: Office for Official Publications of
the European Communities, Synthesis Report.
2005.
REFERENCES
HARMONISE. (2007). Survey of certification
Anderson, J. E., Barret, K. S., & Schwager, P. schemes for ICT professionals across Europe
H. (2005). Informing the HR hiring decision of towards harmonisation. Project of CEPIS Council
IT personnel: The HR professional’s view of IT of European Professional Informatics Societies,
certification, education, & experience. Informing technical report, confidential. Retrieved October
Science Journal, 8, 281–302. 15, 2007, from http://www.cepis-harmonise.org

Butler, P. (2006, October). A review of the lit- HR-XML Consortium. (2004). Competencies.
erature on portfolios and electronic portfolios. Retrieved October 15, 2007, from http://www.
Creative Commons Attribution-NonCommercial- hr-xml.org
ShareAlike 2.5 License.

174
Basics to Develop Web Services for Human Resources

Hughes, C., &, Thompson, C. (2007, June). The Data Modeling: The process of analyzing and
international IT professional practice programme: identifying data objects and their relationships
ICT certifications for informatics professionals. amongst each other within certain contexts like
UPGRADE, VII(3). business or others where they are used. It is a first
step into object-oriented programming,
Lausen, H., de Bruijn, J., Keller, U., & Lara, R.
E-Portfolio: A Web-based information man-
(2006, October). Semantic web services with
agement system that uses electronic media and
WSMO. UPGRADE, VII(5), 34–37.
services. In the first line, it is used for self-expres-
Povalej, R., & Weiß, P. (2007, September). Con- sion, for example, trying to establish contacts for
ceptual data model and meta framework: Mod- private or professional matters.
eling ICT competence profiles. I-KNOW 2007. Framework: Set of assumptions, concepts,
Retrieved from http://www.i-know.at/ conditions, methods, preconditions, values, prac-
tices, and so forth, that constitutes a way of viewing
RCD. IEEE Learning Technology Standards Com-
reality respectively, that supports through making
mittee RCD. (2007). Draft standard for reusable
available blueprints within certain contexts, like
competency definitions (RCD). Retrieved May 12,
ICT development or building projects.
2007, from http://ieeeltsc.org/wg20Comp/Public/
Human Resources: A modern conceptuality
IEEE_1484.20.1.D3.pdf
in the economy, it describes persons as manpower
Weiß, P., Povalej, R., & Stucky, W. (2005, Decem- and/or resources. Through this, the importance of
ber). Learning across boundaries: Developments employees should be emphasized.
towards a European framework for ICT skills. Information and Communication Technolo-
Paper presented at the International Conference gies (ICT): Technologies which provides for the
on Internet & Information Technology in Modern electronic input, storage, retrieval, processing,
Organizations. transmission, and dissemination of information.
Often, ICT are linked with a particular context,
Weiß, P., Stucky, W., Dolan, D., & Bumann, P.
for example, ICT in education, health care, or
(2005): E-skills certification in Europe towards
libraries.
harmonisation. In P. Cunningham & M. (Eds),
Job Profiles: Written definitions of primary
Innovation and the knowledge economy, issues, ap-
job duties, key responsibilities, necessary KSC,
plications, case studies. Amsterdam: IOS Press.
and qualification, education, experiences, profi-
World Wide Web Consortium. (W3C). (2007). ciencies, talent, personal characteristics sought in
Web services activity statement. Retrieved Oc- a candidate, and so forth. They belong to certain
tober 15, 2007, from http://www.w3.org/2002/ working areas and are subdivided into components
ws/Activity which exist of objects like KSC, including the
belonging level. There are required and acquired
ones.
Meta Modeling: The analysis, building up,
KEY TERMS and development of constraints, concepts, frames,
methods models, rules, theories, and so forth,
Certification: One process that formally vali-
practicable and useful for modeling within certain
dates the knowledge, skills, and competences (and/
domains to cope with predefined challenges. In
or know-how, know-what, know-why) acquired
general, a model is an abstraction of objects in
by an individual, following a standard assessment
the real world; a meta-model is an abstraction of
procedure.
model itself.

175
Basics to Develop Web Services for Human Resources

Qualification: Official record (certificate, Web Services: Enterprise applications that


diploma) of achievement which recognizes suc- exchange data, share tasks, and automate processes
cessful completion of education or training, or sat- over the Internet.
isfactory performance in a test or examination.

This work was previously published in Encyclopedia of Human Resources Information Systems: Challenges in e-HRM, edited by
T. Torres-Coronas; M. Arias-Oliva, pp. 84-91, copyright 2009 by Information Science Reference (an imprint of IGI Global).

176
177

Chapter 1.11
Web 2.0 and E-Discovery
Bryan Kimes
Altria Client Services, Inc., USA

Abstract of electronically stored information as a result of


new communication technologies, including Web
Companies today face an overwhelming amount of 2.0 applications.
digital information, and many of them are involved
at some point in civil litigation. When a company is
in the discovery (pretrial) phase of civil litigation, Introduction
it usually exchanges information, including docu-
ments, with the opposing party in the litigation. The Companies today are dealing with new technologies
Federal Rules of Civil Procedure, which govern and a growing amount of digital information. Each
civil litigation in federal courts, were amended in year, the global information growth is larger than
2006 to provide additional guidance to parties with any previous year, and that trend is not showing
regard to electronically stored information. The any signs of slowing down. In 2002, an estimated 5
management teams of many U.S. corporations are exabytes of digital information were created world-
working with their IT departments and lawyers in wide. For reference, 5 exabytes is the equivalent of
order to understand the sources of electronically 37,000 libraries the size of the Library of Congress
stored information that may be potentially relevant (University of California at Berkley, 2003). The an-
to their litigation. Over the last 20 years, technol- nual growth is roughly doubling each year, and U.S.
ogy has grown increasingly more complex, from corporations are contributing to this trend. Electronic
the early mainframe and personal computers to communication technologies, including e-mail as
sophisticated e-mail and instant messaging applica- well as next-generation Web 2.0 social technologies
tions that enable users to send and receive millions (e.g., instant messaging [IM], blogs, and wikis), are
of messages every day. This chapter addresses the also contributing to this rapid growth. For example,
issues companies may face related to the discovery Cellular-News reported that Gartner predicts that
an estimated 1.9 trillion instant messages were sent
DOI: 10.4018/978-1-60566-122-3.ch006 worldwide in 2007 (“SMSs to Surpass 2 Trillion

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Web 2.0 and E-Discovery

Messages in Major Markets in 2008,” 2007). States Department of Justice, 1999).


These new technologies are transforming modern Civil litigation follows specific procedures.
organizations and their employees. Corporate Cases in United States federal courts follow the
technology managers are assessing the potential Federal Rules of Civil Procedure. The Federal
benefits of these new technologies relative to the Rules define the rules into different categories,
burden and risk they present to the organization, including Commencement of Action, Pleadings
including that from litigation. and Motions, Parties, Deposition and Discovery,
This chapter is not intended to be a compre- Trial, and Judgment. The remainder of this chapter
hensive review of the phases of civil litigation or will focus on the discovery phase of civil litigation.
the Federal Rules of Civil Procedure (FRCP). In During the discovery phase of a case, thousands
addition, the author of this chapter is not a lawyer or even millions of documents may be collected,
and the information in this chapter is generalized reviewed, and produced from the files of either
and not intended for specific situations, and should party in the lawsuit.
not be substituted for the advice of a lawyer. The Let us use a hypothetical example to illustrate
following is an overview of the business issues the process. The Acme Corporation, a fictional
related to managing electronically stored infor- product manufacturer, is involved in civil litiga-
mation that may be required in litigation. By way tion. The case involves a former employee, John
of background, many companies will face civil Doe. John filed a lawsuit against Acme alleging he
litigation, which is when one individual or orga- was terminated as a result of age discrimination.
nization sues another seeking relief, oftentimes John Doe, who is the plaintiff, hired a lawyer to
for money damages. represent him in his suit against Acme, the defen-
dant. Acme also hired a law firm to represent it
in the case. When the case reached the discovery
E-Discovery Background phase, John’s lawyer asked Acme for documents
that related to John’s termination. Acme’s man-
The United States v. Microsoft case, which was agement, working with its lawyers, determined
tried in 1998, is a famous civil case in which the which employees had documents that related to
United States Department of Justice (DOJ) filed his termination. It instructed specific employees
an antitrust claim against the software giant alleg- to not discard the documents that related to the
ing it used its monopoly power of the Windows issues in the case. Acme’s IT director worked
operating system to bundle its Internet Explorer with his management to determine the locations
Web browser. The government claimed that other where electronic documents that related to the
Web browsers, such as Netscape, were not able to case were stored.
compete in the Web browser market due to Mi- Now, let us walk through three variations of
crosoft’s positioning of Internet Explorer within the Acme example at different points in time. In
the Windows operating system. The DOJ won in the first variation, Acme terminated John in 1990.
the trial court, but an appeals court overturned the In 1990, most companies did not have company-
decision and the case later settled. The Microsoft wide e-mail accounts or network servers. Wireless
case was full of headlines and was an example cellular telephones were just getting started and
of a case where e-mail played a critical role. The smart phones were still a science fiction concept.
DOJ trial lawyers used internal Microsoft e-mails, The computer systems at Acme, including those
many authored by senior executives, including used by John and his management, were probably
Bill Gates, to demonstrate the company’s alleged running MS-DOS or Windows 2.0 (the classic
strategy to “cut off Netscape’s air supply” (United 286 PC). If Acme was on the cutting edge of

178
Web 2.0 and E-Discovery

computer technology in 1990, it may have had more productive and communicate faster. In 2007,
Windows 3.0, which ran on the 386 powerhouse. Acme used Microsoft Vista with SharePoint. It
In order to respond to the John Doe lawsuit, had Microsoft Exchange 2007 with unified mes-
Acme’s management contacted the employees saging (voice mail integrated into users’ e-mail
that may have had records that related to John’s account). Many senior employees have wireless
termination, including the personnel files from handheld devices that integrate into their e-mail
John’s manager, the human resources (HR) files accounts as well as instant text messaging. Acme
that related to John and Acme’s employment poli- employees use these new technologies and create
cies. In order to find the electronic documents, more documents on their desktops and rely on
Acme’s management went to the IT director who electronic communication tools, including e-mail
managed all of Acme’s computer systems. The IT and instant messaging. In 2007, the IT director
director managed a relatively small system that has to address a more complicated network of
included a mainframe computer system. The IT computer systems to help management identify
director worked with management to find material those systems that contain documents that relate
regarding John’s case. to John’s lawsuit. As a result of the increase in the
Now let us assume Acme terminated John in number of documents, Acme’s costs associated
2000. In 2000, Acme had been using e-mail for with the discovery phase of the case is likely to
several years now and employees regularly use be even more expensive and time consuming in
e-mail to correspond with each other to get their 2007 than in 2000 due to the higher volume of
work done, including personnel matters. Acme electronically stored information.
had Windows 98 running on employees’ desktops, Acme is certainly not alone. Companies all
which allowed employees to save documents, for over the world are facing a new generation of
example, in Microsoft Word, to their computers. technology as new employees, who have been
Employees saved documents to the computers’ teething on instant messaging, MySpace, and
desktops, local drives, and some network file Facebook, join the workforce. For example, many
servers. Acme’s management followed a similar companies convene meetings using Web-based
process for collecting the documents as it did in electronic rooms, for example, WebEx, with shared
1990. However, because the electronically stored desktops. These next-generation meeting rooms
information at Acme has multiplied since 1990, allow participants to attend meetings without ever
Acme’s job to identify the relevant documents has leaving their desks, and many participants may
become more complicated. Acme management never meet each other face to face.
asked various employees that worked with John, Technology managers, like the IT director at
including his manager, to identify the documents Acme, may have a difficult task when they are
that related to the case. It also worked with the IT asked to identify systems and electronically stored
director to understand how the computer systems information that may contain documents relevant
were managed and how employees used those to a lawsuit. Additionally, the technology genera-
systems. Acme’s cost associated with the dis- tion gap is often short relative to the time it takes
covery phase of the case has likely grown due to for courts and litigators to deal with these new is-
the increased volume of documents that Acme’s sues in actual cases. It can take years for a dispute
lawyers needed to review and potentially produce to make its way into a courtroom. Technology
to John’s lawyers. managers are faced with making decisions about
Finally, let us assume that Acme terminated how to implement new technology tools without
John in 2007. Since 2000, Acme implemented knowing what, how, or when they will have to
several new technologies to help its employees be produce documents in the future.

179
Web 2.0 and E-Discovery

Federal Rules of result of the explosion of electronic documents,


Civil Procedure specifically e-mail and other documents stored
on Acme’s computer systems.
Companies that find themselves involved in civil Below is a summary of the electronic-discovery
litigation follow certain rules. If the case is in a amendments to the Federal Rules of Civil Proce-
federal court, they use the FRCP; if the case is in dure that this chapter will discuss.
a state court, the parties follow the state’s rules
of civil procedure, which often follow, if not • Rule 26(a)(1)(B): General Provisions
verbatim, the FRCP. Again, this chapter will not Governing Discovery; Duty of
be an exhaustive review of the FRCP, but it will Disclosure; Required Disclosures;
touch on a few rules that relate to electronically Methods to Discover Additional Matter
stored information.
On December 1, 2006, amendments to the A copy of, or a description by category and
Federal Rules of Civil Procedure were submitted location of, all documents, electronically stored
by Congress and approved by the Supreme Court. information, and tangible things that are in the
The amended rules specifically address electroni- possession, custody, or control of the party and
cally stored information. Let us start with a rule that the disclosing party may use to support its
that was not amended, FRCP 26(b)(1), which claims or defenses, unless solely for impeachment.
defines the scope and limits of Discovery. (“Law Library/Court Rules,” n.d.)
What Might this Mean for Acme? When Acme’s
• Rule 26(b)(1): Discovery Scope and IT director reviewed this amended rule with
LimitsIn General. Parties may obtain management and outside lawyers, they discussed,
discovery regarding any matter, not privi- among other things, the computer systems at Acme
leged, that is relevant to the claim or de- and thought about a summary of the electronically
fense of any party, including the existence, stored information, such as in databases, servers,
description, nature, custody, condition, and and electronic archives, that may be relevant to
location of any books, documents, or other John’s case against Acme.
tangible things and the identity and loca-
tion of persons having knowledge of any • Rule 26(b)(2)(B): General Provisions
discoverable matter. For good cause, the Governing Discovery; Duty of
court may order discovery of any matter Disclosure; Discovery Scope and Limits;
relevant to the subject matter involved in Limitations
A party need not provide dis-
the action. Relevant information need not covery of electronically stored information
be admissible at the trial if the discovery from sources that the party identifies as not
appears reasonably calculated to lead to reasonably accessible because of undue
the discovery of admissible evidence. All burden or cost. On motion to compel dis-
discovery is subject to the limitations im- covery or for a protective order, the party
posed by Rule 26(b)(2)(i), (ii), and (iii). from whom discovery is sought must show
(U.S. House of Representatives Committee that the information is not reasonably ac-
on the Judiciary, 2006) cessible because of undue burden or cost. If
that showing is made, the court may none-
Now let us return to the Acme example. Why theless order discovery from such sources
is Acme’s IT director’s job more complicated in if the requesting party shows good cause,
2007 than in 1990? I suggest to you that it is the considering the limitations of Rule 26(b)

180
Web 2.0 and E-Discovery

(2)(C). The court may specify conditions request (1) to produce and permit the party
for the discovery. (“Law Library/Court making the request, or someone acting on
Rules,” n.d.) the requestor’s behalf, to inspect, copy,
test, or sample any designated documents
What Might this Mean for Acme? When Acme’s or electronically stored information—in-
IT director reviewed this rule with management cluding writings, drawings, graphs, charts,
and lawyers, they discussed, among other things, photographs, sound recordings, images,
what sources of potentially relevant electronic and other data or data compilations stored
information that they thought were not reasonably in any medium from which information
accessible and that they did not intend to search. can be obtained—translated, if necessary,
If John’s lawyer wanted Acme to provide docu- by the respondent into reasonably usable
ments from one of the sources that Acme disclosed form, or to inspect, copy, test, or sample
as not reasonably accessible, then John’s lawyer any designated tangible things which con-
could serve a motion to compel the production stitute or contain matters within the scope
of information from the sources identified as not of Rule 26(b) and which are in the posses-
reasonably accessible. Acme will then have to sion, custody or control of the party upon
explain why the information that John’s lawyer whom the request is served;
is seeking would require undue burden or cost
to obtain. (b) PROCEDURE: The request shall set
Now let us look at another hypothetical ex- forth, either by individual item or by cat-
change. John’s lawyer wants Acme to provide egory, the items to be inspected, and de-
information from an old HR computer system scribe each with reasonable particularity.
that is no longer active, and Acme considers the The request shall specify a reasonable time,
old system not reasonably accessible because it place, and manner of making the inspection
would require the IT director to hire a specialized and performing the related acts. The request
vendor to restore the old system. John’s lawyer may specify the form or forms in which
files a motion to compel in order to obtain the electronically stored information is to be
information contained within the computer system. produced. Without leave of court or writ-
Acme’s IT director explained to the court that ten stipulation, a request may not be served
the request is unduly burdensome; specifically, it before the time specified in Rule 26(d). The
will cost Acme US$500,000 to restore the legacy party upon whom the request is served shall
computer system. Acme feels that the burden is serve a written response within 30 days
excessive relative to the amount John’s lawyer is after the service of the request. A shorter
seeking in the case, which is only US$100,000 or longer time may be directed by the court
in money damages. The court will have to weigh or, in the absence of such an order, agreed
these factors when determining whether to permit to in writing by the parties, subject to Rule
discovery. 29. The response shall state, with respect
to each item or category, that inspection
• Rule 34(a) & (b): Production of and related activities will be permitted as
Documents, Electronically Stored requested, unless the request is objected to,
Information, and Things and Entry including an objection to the requested form
Upon Land for Inspection and other or forms for producing electronically stored
Purposes; Procedure
(a) SCOPE: Any information, stating the reasons for the objec-
party may serve on any other party a tion. If objection is made to part of an item

181
Web 2.0 and E-Discovery

or category, the part shall be specified and IT director understands that John’s lawyer may
inspection permitted of the remaining parts. specify a form of production and if he does not,
If objection is made to the requested form then Acme will have to produce the documents in
or forms for producing electronically stored the form they are ordinarily maintained at Acme
information—or if no form was specified or in a reasonably usable format. Let us assume
in the request—the responding party must John’s lawyer agreed to accept a production of the
state the form or forms it intends to use. The relevant documents in Adobe’s Portable Document
party submitting the request may move for Format (PDF). In addition, Acme provided an
an order under Rule 37(a) with respect to accompanying index of metadata including the
any objection to or other failure to respond date, title, author, recipient, document type, and
to the request or any part thereof, or any extracted full text.
failure to permit inspection as requested. The amended federal rules have provided or-
Unless the parties otherwise agree, or the ganizations and their IT departments with more
court otherwise orders: instruction regarding electronically stored infor-
mation. However, complying with the obligations
(i) a party who produces documents for associated with electronic discovery can be very
inspection shall produce them as they difficult for organizations, even those with a mod-
are kept in the usual course of business est technology portfolio. Organizations with newer
or shall organize and label them to generation technologies may find themselves in
correspond with the categories in the a much more complicated situation when faced
request; with discovery in civil litigation.
(ii) if a request does not specify the form In a recent Richmond Journal of Law & Tech-
or forms for producing electronically nology article, Doug Rogers, citing the Sedona
stored information, a responding party Principals,1 stated “it is unreasonable to expect
must produce the information in a parties to take every conceivable step to preserve
form or forms in which it is ordinarily all potentially relevant electronically stored in-
maintained or in a form or forms that formation”:
are reasonably usable; and For instance, a party should not have to preserve
(iii) a party need not produce the same all possible sources of e-mail simply because it is
electronically stored information in possible that an existing or former employee, who
more than one form. (“Law Library/ claims she was harassed, may have forwarded an
Court Rules,” n.d.) e-mail from her supervisor to other employees, or
because one of the e-mails may have been exported
What Might this Mean for Acme? When the into a .pst file outside of the e-mail folders in the
Acme IT director reviewed this rule with manage- cache of a computer.
ment and lawyers, they likely discussed, among Comment 5a to Sedona Principle 5 explains:
other things, the options regarding how they The obligation to preserve relevant evidence is
would produce documents for John’s lawyers. generally understood to require that the producing
They discussed an upcoming meeting between party make reasonable and good faith efforts to
their lawyers and John’s lawyers, during which identify and manage the information that it has
they would discuss the form of the document identified as reasonably likely to be relevant. Satis-
production. The Acme IT director has already fying this obligation must be balanced against the
met with some vendors that provide specialized right of a party to continue to manage its electronic
tools to assist with document productions. The information in the best interest of the enterprise,

182
Web 2.0 and E-Discovery

even though some electronic information is neces- computers’ temporary RAM to be excluded from
sarily overwritten on a routine basis by various discovery because they are so transitory (Jacoby,
computer systems. (Rogers, 2008, p. 25-26) 2007). Some commentators have expressed that
Companies like Acme are not alone when Rule 34, which describes “other data or data
working through these issues. Over the past few compilations stored in any medium from which
years, many specialized litigation support vendors information can be obtained,” would not apply
have emerged that offer a variety of technology because the RAM data were not easily obtainable.
and process solutions to assist organizations and In this case, however, the judge stated that the
their IT departments in managing the discovery data stored in this computer system’s RAM was
process. In July 2007, one such vendor, Autonomy, “extremely relevant and may be key,” and that it
acquired another company, Zantaz, which is an was not available anywhere else (Jacoby).
e-mail archiving firm, for US$375 million. Soon A notable distinction of the temporary, or
after the Zantaz acquisition, Autonomy reported ephemeral, data at issue in the Columbia Pictures
a US$70 million deal with an unspecified cus- case is the way the specified computer system was
tomer. The customer is believed to be a large configured. Many computer systems use RAM as
global bank that is preparing for a wave of civil a memory buffer that will randomly (hence the
litigation in the wake of the subprime mortgage name random access) write some data to help
lending crisis. The value of this transaction, which support the application. For example, take an
is Autonomy’s largest, is reported to be almost ordinary desktop computer. When a user turns
double Autonomy’s average transaction with a the computer on, the computer will allocate some
single customer. If this type of transaction is any RAM to support the operating system. If the user
indication of future business, Autonomy could then opens Microsoft Word, the computer will
see a return on the Zantaz acquisition much faster allocate more RAM for the application and the
than many anticipated (Mellor, 2008). associated documents that are open. If the user then
Companies and organizations around the world opens Excel and closes Word, the RAM assigned
are evolving their technology footprints. Many to Word is sometimes overwritten in order for
commentators believe the volume of documents the computer to run Excel. The computer, during
that employees create on new applications are also normal operation, will recycle RAM memory to
increasing at a rapid rate, some say exponentially. whatever application(s) needs it. The information
The legal environment is also evolving and new stored in RAM, if viewed by a human, would
cases are emerging that deal with new issues most likely appear as gibberish as it is bits and
regarding electronically stored information. One bytes of computer code that the computer uses to
case that has gained some notoriety is the Co- function properly.
lumbia Pictures case (Columbia Pictures Indus. Nevertheless, the computer systems at Co-
v. Bunnell et al.). In this case, Columbia Pictures lumbia Pictures, based on its own testimony, had
was ordered to capture and preserve data stored a logging feature that was turned off. The court
in temporary random access memory (RAM), ordered Columbia to prospectively activate the
which is very volatile and can be overwritten by logging feature and that the prior deletion of RAM
the computer system when shutting down, starting was not a violation of the preservation order in
up, or simply opening and closing applications. the case (Jacoby, 2007).
Some commentators have stated that the judge’s Instant Messaging is now a ubiquitous tech-
holding is limited to the specific facts at issue nology across grade schools, college campuses,
in this case and generally considered data in shopping malls, and movie theaters (much to my

183
Web 2.0 and E-Discovery

chagrin). Instant messaging took off in the mid- Web 2.0 technologies, including IM, offer
1990s with the popular ICQ (“I seek you”). This employees at large organizations instant access
was followed by AOL’s Instant Messenger in the to colleagues across the organization. These
late 1990s. Today, there is a variety of open-source technologies flatten out the hierarchical structure
instant messaging, or text messaging, applications of many organizations. Andrew McAfee, associ-
that run on desktop computers, mobile phones, and ate professor at Harvard Business School, states
other wireless devices. Many instant messaging employees “increasingly react to situations and
applications, especially those on mobile phones, problems on the fly, not solely by hierarchy”
are similar to RAM in that many of them do not (Mamberto, 2007, p. 2).
keep a permanent log of past messages. Many The American Bar Association (ABA; Ameri-
mobile messaging devices keep messages for a can Bar Association Legal Technology Resource
short period of time, sometimes only minutes or Center, 2008) cited a January 2008 Detroit Free
hours, before they are overwritten or deleted. Press report (Wendland, 2008) that IM text mes-
IM is slowly invading the workplace. Instant sages obtained from the offices of Detroit Mayor
messaging allows employees to communicate Kwame Kilpatrick suggested he had a 2-year affair
faster than e-mail and slightly slower than a real- with his chief of staff Christine Beatty. Over 14,000
time phone call. Users can carry on several con- text messages between Kilpatrick and Beatty, who
versations at the same time. They can create buddy were both married at the time, were exchanged
lists, which allow for quicker communication as that allegedly illustrated a sexual affair including
well as the ability to see if someone is online or during official business trips. The lawsuit, which
not. In a Wall Street Journal online article, Mam- was filed by a former police officer, alleged that
berto (2007) cites a 2006 study of the American he and another officer were fired in retaliation
Management Association and the ePolicy Institute because of their role in an internal investigation
that found roughly one third of employees are us- that would potentially expose the mayor’s affair.
ing IM, many of them without their employers’ Mayor Kilpatrick testified that he did not have an
knowledge. Many corporations and organization affair with his chief of staff and that he did not
are reluctant to deploy IM systems due to security retaliate by firing two police officers. The City of
concerns, loss of productivity, and inappropriate Detroit tried to avoid producing the damaging IM
use. Despite these reservations, Gartner, an IT text messages, but eventually they were disclosed.
research firm, predicts IM will dominate 95% of The case eventually settled for US$8 million.
the workforce of large companies within the next The ABA article (2008) further reported that
5 years (Mamberto). an estimated 20 billion IM text messages are sent
Desktop-computer-based IM applications each month in the United States alone. The article
are slightly different than mobile-phone-based cited a Gartner report, which estimated that 1.9
systems, notably with regard to storage. Many trillion text messages were sent worldwide in
desktop-based IM applications provide options 2007. The report predicted a 19% increase to
for longer term storage, similar to conventional 2.3 trillion text messages in 2008. When Acme’s
e-mail systems. When Acme’s IT director imple- IT director implemented desktop IM, he or she
mented desktop IM, he or she may have worked worked with the records, compliance, and law
with the records, compliance, and law department department to consider how to train employees
to determine whether Acme needed to include a on the proper use of IM, similar to the corporate
storage component in the event it would need to e-mail system.
save some messages for a longer term purpose, Blogs (Web logs), wikis (collaborative Web
either for business or litigation purposes. sites), and RSS (really simple syndication) are also

184
Web 2.0 and E-Discovery

on the rise within companies and organizations. Journal, citing a 2006 survey of the Employment
These new technologies can offer significant ad- Law Alliance, found that about 5% of Americans
vantages to organizations. Most of the advantages have a personal blog, while only 15% of their
are in response to the limitation of e-mail. For ex- employers have a blogging policy. Additionally,
ample, a user can subscribe to relevant work sites Browning stated that an estimated 27% of adults
using RSS feeds as opposed to relying on e-mails read blogs according to the Pew Internet and
from colleagues. E-mail can be a very inefficient American Life Project. Employees may make
way to keep employees updated on a given topic seemingly innocent statements on a public blog
because many employees will overuse the carbon that can put a company, and the employee, at great
copy (CC) function as a way to inform people risk. For example, if one of the Acme employees
on the distribution list. There are two problems working on the new product was at home spend-
with this technique. First, the person sending the ing time on a personal blog, he or she may post
message may not know all the people that really an entry about how great Acme is and that it has a
need to be informed of the topic. Second, many great future—referencing the new product without
people on the distribution list may not even care actually mentioning it by name. The employee is a
about the message. hardworking and dedicated individual who really
One of the advantages of RSS as a commu- believes in Acme and is proud to work there. The
nication tool is that users can decide what they employee may not even have to mention the new
are interested in and will subscribe to those topic product by name but could indirectly disclose
areas. The RSS technology will notify users of trade secrets or forward-looking financial infor-
updates and direct them to the source of the in- mation that could be damaging to Acme. Before
formation. Many of the RSS sites will keep track the lightening fast World Wide Web, an innocent
of the pages viewed so users do not need to worry comment to a neighbor in the driveway presented
about reading articles twice because it is tracked some risk to the company, but in the days of on-
by the system (“The Logic of Blogs and Wikis in line blogs, where users from all over the world
the Enterprise,” 2005). can easily search and find just about anything
Now, let us return to the Acme example, spe- on anything, a seemingly innocent comment is
cifically, a project to develop a new product. The now within infinitely greater reach. Again, the
Acme project team created an electronic workroom employee probably had no intention of making a
where project team members could save and up- public statement about Acme’s financial future.
date important documents related to the product. Many companies have existing policies that govern
The workroom had several blogs and wikis for employee communication, which would generally
employees to discuss current issues related to cover employee blogging. However, the Society
the project, including materials, sales strategies, for Human Resource Management reports that a
and budgets. Managers who were not working on small number (3%) of firms have implemented
the project day to day could use the RSS tool to blogging policies specifically (Browning).
subscribe to the workroom. The manager would
be notified when new content had been added to
the workroom on a specific topic. Conclusion
Blogging is still a relatively new technology
and companies are still learning and assessing Employees are faced with new technologies
the security risks related to employee blogging. every year, and the current wave of Web 2.0
John Browning (2007) of the Houston Business technologies will soon be replaced with a new

185
Web 2.0 and E-Discovery

generation of tools that will enable employees to Jacoby, C. J. (2007). E-discovery update: Discov-
share information faster and easier. Information ery of ephemeral digital information. Law and
technology managers must make decisions regard- Technology Resources for Legal Professionals.
ing these changing technologies and must work Retrieved February 25, 2008, from http://www.
with their management to assess the pros and cons llrx.com/columns/fios19.htm
of implementing them. Many IT managers realize
Law library/court rules. (n.d.). LexisNexis Applied
that employees will create and store documents
Discovery. Retrieved February 25, 2008 from
on these new systems and those documents may
http://www.lexisnexis.com/applieddiscovery/
be needed for future litigation. Furthermore, as
lawLibrary/courtRules.asp
the overall volume of documents increases as a
result of newer technologies, the overall cost to Mamberto, C. (2007). Instant messaging in-
organizations will also likely increase due to the vades the office: Companies say it spurs broader
time and expense for lawyers to collect, review, collaboration—and scares some bosses. Wall
and produce these documents in litigation. As Street Journal Online. Retrieved February
the volume of discoverable electronically stored 25, 2008, from http://online.wsj.com/article/
information grows, the real winners may be the SB118523443717075546.html
specialized vendors, like Autonomy, that offer
Mellor, C. (2008). Sub-prime crisis god news for
specialized document management and litigation
legal discovery software. Techworld. Retrieved
support services to companies dealing with huge
February 25, 2008, from http://www.techworld.
volumes of data. If the current data-growth trends
com/storage/news/index.cfm?newsid=11042
continue, coupled with increased user adoption of
Web 2.0 technologies, then companies, including Rogers, D. (2008). A search for balance in the
their management, IT managers, and lawyers, will discovery of ESI since December 1 2006. Rich-
have a lot of work ahead of them. mond Journal of Law & Technology. Retrieved
May 5, 2008, from http://law.richmond.edu/jolt/
v14i3/article8.pdf
References
SMSs to surpass 2 trillion messages in major
American Bar Association Legal Technology markets in 2008. (2007). Cellular-News. Retrieved
Resource Center. (2008). Text messaging and the February 25, 2008, from http://www.cellular-
importance of a records retention policy. Retrieved news.com/story/28126.php
February 25, 2008, from http://meetings.abanet.org/ The logic of blogs and wikis in the enterprise.
ltrc/index.cfm?&data=20080204#E6AADE50- (2005). The Gilbane Report, 12(10). Retrieved
02C7 February 25, 2008, from http://gilbane.com/
Browning, J. (2007). Adopting a blogging policy gilbane_report.pl/104/Blogs__Wikis_Technolo-
could limit your exposure to legal risk. Houston gies_for_Enterprise_Applications.html
Business Journal. Retrieved February 25, 2008, United States Department of Justice. (1999). U.S.
from http://houston.bizjournals.com/houston/ Department of Justice proposed findings of fact.
stories/2007/05/28/smallb2.html Retrieved February 25, 2008, from http://www.
justice.gov/atr/cases/f2600/2613-1.htm

186
Web 2.0 and E-Discovery

University of California at Berkeley’s School of In- ENDNOTE


formation Management and Systems. (2003). How
1
much information? Retrieved January 2, 2008, The Sedona Principals: Best Practices, Rec-
from http://www2.sims.berkeley.edu/research/ ommendation & Principles for Addressing
projects/how-much-info-2003/execsum.htm Electronic Document Production is a project
of The Sedona Conference Working Group
U.S. House of Representatives Committee on the on Best Practices for Electronic Document
Judiciary. (2006). Federal rules of civil procedure. Retention and Production. The Sedona
Retrieved February 25, 2008, from http://judiciary. Conference is a non-profit organization of
house.gov/media/pdfs/printers/109th/31308.pdf lawyers, consultants, academics and jurist.
Wendland, M. (2008). Most text messages just
vanish. Detroit Free Press Online. Retrieved
February 25, 2008, from http://www.freep.
com/apps/pbcs.dll/article?AID=/20080125/
COL11/801250422/1081/COL

This work was previously published in Social Software and Web 2.0 Technology Trends, edited by P. Deans, pp. 68-78, copyright
2009 by Information Science Reference (an imprint of IGI Global).

187
188

Chapter 1.12
The Power and Promise
of Web 2.0 Tools
G. Andrew Page
The University of Alaska Anchorage, USA

Radwan Ali
Kennesaw State University, USA

Abstract What is Web 2.0?

The key idea that sets constructivism apart from The Internet has presented great opportunity for
other theories of cognition was launched about global human participation by transcending geo-
60 years ago by Jean Piaget. It was the idea that graphical, cultural, religious, social classification,
what is called knowledge does not and cannot and political barriers. As the proliferation of learning
have the purpose of producing representations of with technology increases, there is also amplification
an independent reality, but instead has an adap- in the array of technological possibilities for a vari-
tive function (Von Glasersfeld, 1996, p.3). In this ety of asynchronous and synchronous interactions.
chapter, a variety of Web 2.0 applications and their Therefore, it becomes necessary to provide insight
affordances are presented and discussed in relation into the effective use of these technologies and the
to constructivism in higher education. The aim is facilitation of e-learning. According to Schrum and
to explain how these applications can be used in Hong (2002) the goals of teaching with technology
higher education to promote interactive and engag- should include facilitating higher-level, thinking
ing learning environments. Recommendations for skills, such as analysis, synthesis, and evaluation.
harnessing the potential of these tools along with Alexander describes this phenomenon:
practical examples will assist facilitators of higher
education with creative means to design their courses Web 2.0 is defined as a way of creating webpages
and thus promote Learning 2.0. focusing on microcontent and social connections
between people. It also exemplifies that digital
content can be copied, moved, altered, remixed, and
linked, based on the needs, interests, and abilities
of users—quite against the grain of both traditional
and recently expanded copyright (Alexander, 2008,
DOI: 10.4018/978-1-60566-654-9.ch010 p.151).

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
The Power and Promise of Web 2.0 Tools

There are many creative Web 2.0 applications, “Constructivism affords a knowledge build-
tools, and services available online (See Appen- ing process that engages active learners with the
dix A for a list of digital resources and links). physical and social world” (Twomey-Fosnot,
These tools or web-based learning objects have 1996, p.30). These interactive online tools include
the potential to engage and involve the learner portals, blogs and video blogs, widgets, plugins,
with technology as opposed to having a student wikis, conferencing, games, survey instruments
learn from a computer module or digital lecture. and games. Using these Web 2.0 tools in an edu-
There are many Web 2.0 tools that can be found cational setting builds on Vygotsky’s (1978) view
at different sites on the web. For the purpose of of interactive learning within the zone of proximal
this chapter, we will discuss blogs, wikis, widgets, development (ZPD) and how the social process
nings, plugins, social networking (MySpace and is crucial to the development of thought and be-
Facebook), and virtual environments (Second havior patterns. The application and integration
Life). What issues will Web 2.0 solve? What are of quality educational content is an important
the potential learning outcomes from using Web part of the constructivist framework. This is the
2.0? It is the promises of positive educational adaptive function of constructivism alluded to by
outcomes that web-based tools can potentially Von Glasersfeld (1996).
produce through quality interaction at a meta-level When interactive technology is applied in
that are intriguing many educators and trainers. accordance with constructivist principles, it
This chapter has implications for instructors, has been called “learning 2.0” (Brown & Adler,
students, instructional designers, and administra- 2008). Learning 2.0 is about actively using tech-
tors involved with e-learning in higher education. nology. The multiliteracies of education (e.g.,
This chapter also provides a synthesis of e-learning critical thinking, creating content, collaboration)
issues and an overview of Web 2.0 tools for pro- are fostered due to the constructivist nature and
moting a constructivist environment. exchange of knowledge that is part of the Web
The 5th edition of The Principles of Instruc- 2.0 phenomenon (Alexander, 2008). Learners
tional Design (Gagne, Wager, Golas, Keller, taking the initiative to create content and learning
2005), states a very important instructional by being engaged and involved is at the heart of
design question which is often overlooked. The constructivist practice. Brown and Adler (2008)
question is “For what problem is instruction the acknowledge the impact of Web 2.0 and how it
solution?”(p.23). This question is especially rel- can add a new dimension to communication and
evant considering the numerous challenges and participation:
limitless potential of Web 2.0 tools. By using
Web 2.0 tools the theory of constructivism can The latest evolution of the Internet, the so-called
be applied to reach new levels of digital creation Web 2.0, has blurred the line between producers
and e-learning. and consumers of content and has shifted attention
Constructivism is the process of linking new from access to information toward access to other
understanding to old, modifying and enriching people. New kinds of online resources—such as
existing knowledge, and enhancing depth of social networking sites, blogs, wikis, and virtual
comprehension about a topic. McFedries refers communities— have allowed people with common
to Web 2.0 as functioning as a platform (p.68) interests to meet, share ideas, and collaborate in
because of the ability to delete, edit, and add innovative ways. Indeed, the Web 2.0 is creating a
content and work collaboratively with others in new kind of participatory medium that is ideal for
a synchronous approach. supporting multiple modes of learning. (p.18)

189
The Power and Promise of Web 2.0 Tools

The Use of Web 2.0 learning outcomes. Basically, blogs are web-based
Applications to Promote journals that allow participants to contribute to or
Constructivist Learning create discussion threads. The threads can discuss
different subjects separately or concurrently. There
This section will consider the benefits of construc- are possibly millions of blogs on the web whose
tivism and match them with the characteristics of subjects include religion, politics, medicine, and
a few of Web 2.0 applications. Constructivism, as car repairs in addition to others. Blogs are char-
a learning theory, provides an appropriate fit for acterized by their asynchronous nature and ease
the use of Web 2.0 applications. Constructivism of use. However, they also provide opportunities
as a learning framework can be served immensely for audio and video files. Blogs can be used for
by using the tools and affordances (Jonassen, constructivist learning in many ways. First, a
1999) presented by the Web 2.0 applications. blog can serve as an individual learning journal.
Constructivism suggests that learners can enjoy A learner may start a blog to keep inventory of
their experience by being more involved in their his/her learning. This inventory will help learners
learning than by being listeners (Franklin, 2000; build from one phase of their learning to another.
Ge & Tok, 2003). The idleness or boredom aspects By doing so, learners use previous knowledge
of passive listening prevent learners from realizing to generate new insight. Another example of a
the potential of the subject matter being discussed. blog being used for educational purposes is a
However, by engaging in relevant learning activi- class group project. Learners can be engaged
ties, the learner will have the opportunity to apply with discussion topics and debates in a flexible
the concept learned. Web 2.0 applications can be learning space that allows for the incorporation
very engaging for different reasons: the novelty of videos, graphics, and audio and the enhanced
of the technologies, the aesthetic appeal, and their creativity. Blogs are user-generated web-based
interactive modes. journals that offer opinions and information and
Web 2.0 applications present many affordances that may include text, images, and links to other
for constructivist learning. Some of these applica- blogs and web pages. Some blogs are confined to
tions include social networking, blogs, podcasting, personal expressions, but others make provision
media sharing, and virtual worlds. The common for reactions and comments from readers. In higher
trait of these applications is shareability. These education, blogs have been used as a means by
applications and similar others allow for global which students can collaborate asynchronously. A
interaction where participants contribute and use blog can also serve as a group knowledge builder.
others’ contributions. The contribution can be as A group of learners can join forces in order to
simple as a few paragraphs posted to a blog or as complete tasks or projects. Each user is usually
complex as creating avatars on a Second Life (SL) assigned a section of the task to complete and
island. The main premise is that these applications contribute to the blog. Through this collaboration,
represent great potential for learning. In this sec- useful content is usually produced as a solution
tion, a sample of five Web 2.0 applications is used for the task or project. The produced content is
to illustrate the use of Web 2.0 applications as presented as the fruition of the learning experi-
contexts for constructivist learning. This sample ence. By collaborating with others, a learner can
includes social networking and bookmarking, be benefit by adding to his/her knowledge through
blogs and wikis, podcasts, and media sharing, the different members’ contributions.
and virtual worlds A wiki is a collaborative website that allows
Blogs have been around for several years and visitors to add, remove, edit, and change content,
we are seeing the impact of their use for positive typically without the need for registration. It also

190
The Power and Promise of Web 2.0 Tools

allows for linking among any number of pages. as tools to learn from and teach others and pro-
This ease of interaction and operation makes vide follow-up on previous discourse. Dearstyne
a wiki an effective tool for mass collaborative (2007) notes the trends in collaboration such as the
authoring. (e.g., Wikipedia.org). Similar to blogs increase in blogs, mashups, and wikis. Mashups
in many aspects, Wikis can serve numerous con- are websites, or other applications, that integrate
structivist learning opportunities such as content and aggregate content from more than one source
production. Wikis are made of many web pages into an integrated application (e.g., combining
that allow contributors to add and modify content, data on a topic of interest with geographical data
theirs and others’. The modification opportunities such as Google Maps).
make it possible for all wiki members to share Virtual worlds applications such Second Life
their knowledge in addition to assessing that of are interactive environments where participants as-
other members. By assuming the editor (modifier) sume personalities (avatars) to interact with other
role, a wiki contributor is improving his/her writ- individuals and objects such as rooms, documents,
ing and critical thinking. Through this practice, presentations, and videos. By interacting with
new knowledge may be found and added to the the different objects, a learner will gain a bit of
learner’s previous knowledge. The aggregation knowledge from each object and then connect these
of faculty publications in a wiki is described by different bits from the different objects to form
Conner (2007) as a means to socially construct the new experience. Similarly, the virtual world
an online library resource and promote research setting provides group collaboration contexts,
productivity. and consequently, opportunities for constructing
An example of an educational wiki in higher content. The application of collaborative and
education would be Ask Dr. Wiki (http://www. social constructivist-oriented activities, such as
askdrwiki.com/). This is a medical wiki devoted virtual spaces and a “user-centric approach” that
to creating a free source of medical information. promotes higher level learning are described by
Individuals can publish clinical notes, pearls, Sommerville and Nino (2007). The social syn-
X-ray images, angiograms and more on the site. chronicity achieved through these media allows
Using this wiki, anyone with a medical background distant learners to connect in real-time and promote
can contribute or edit medical articles. connectedness and community.
The group experience and the community
Media Sharing and Construction membership provide positive confines that em-
brace collaboration and individual creativity.
Podcasts and media sharing include video, audio, While the participants are making products, they
photos, slides, and others to present other con- will have opportunities to acquire and gain feed-
structivist learning opportunities through produc- back from other community members (Jonassen,
ing and sharing digital content. Media sharing is 1999). That will provide the different participants
appealing because of its interactive and personal with new skills and knowledge as benefits of the
nature. The content can be audio or video files. feedback. These applications present opportunities
The main premise here is that learners can gener- where a learner can physically actively engage
ate new knowledge by creating these podcasts and through the human computer interface (HCI)
learn from the creation process. Through podcasts, technology, virtually, and mentally active. An ex-
learners can share information and broadcast many ample of how a learner can be physically engaged
useful materials. Learners, then, can use podcasts in a virtual environment (and arguably a web 3.0

191
The Power and Promise of Web 2.0 Tools

phenomenon) is in the work of Mitch Kapor and designing and implementing authentic activities
Philippe Bossut (2008) who have developed a to which the learners can relate can be a great
hands-free navigation device for SecondLife. A motivation for the learner to participate (Zaulk-
video of this prototype is available at: http://www. ernan, 2006). That authenticity can pique the
youtube.com/watch?v=2t52gkAwJq8 learners’ interest and curiosity. For example, the
Consequently, by being mentally active, use of podcasting and vodcasting (video casting)
learners will expand their cognitive abilities to by music students enhances the learning experi-
move beyond memorization. Within that practice, ence and enables the transmission of content to
long-term critical thinking becomes the norm for mobile devices (m-Learning) such as iPods and
these learners. Furthermore, as the learners’ criti- cell phones.
cal thinking evolves, a few positive habits such How will a facilitator use the different Web 2.0
as planning and organization are gained. These applications as contexts for constructivist learning
habits become second nature for these learners experiences? This question should be considered
in their potential careers. Web 2.0 activities are in relation to learning objectives, activities, and
similar in those regard. For example, learners who outcomes.
use blogs to track their learning (writing) and to Implementing Web 2.0 technologies, while
collaborate with others (building camaraderie) will establishing learning objectives, is the beginning in
develop better writing and interpersonal skills to creating a constructivist approach to learning. To
help in their respective careers. The main point accommodate for the potentially different levels
is that many of these learning experiences are of motor skills among the learner, the facilitator
transferrable, and can be potentially helpful in must include learning the technology as one of the
many different contexts. learning objectives of the learning experience. By
Perhaps the most important aspect of con- doing so, the learner will understand the impor-
structivism is that learners are charged with tance of these technologies for completing their
their learning. That charge will breed a sense of tasks. For example, a facilitator may create a list
ownership and accordingly, learners become more of learning objective as shown below in Table 1
interested in the process. Similarly, Web 2.0 ap- to promote Learning 2.0.
plications encourage all participants to create their Accordingly, the activities in the same learning
own work that can be seen by millions around the experience should be planned to help the par-
world and have the potential for cross-hemispheric ticipants use the respective Web 2.0 technologies
collaboration. Because they own that work and while completing their tasks toward achieving the
the fact that it is in the public spotlight, those learning outcomes. In order to apply an activity
participants become more motivated to do work to achieve Learning Objective 5, a facilitator may
of high quality. There is an interesting dynamic require the learners to find a blog and monitor the
and digital responsibility that is inherent to content interaction or to set up an RSS feed to manage the
creation on the web. The participants will seek and flow of created content. Furthermore, the learn-
explore new venues and resources from which ers are asked to explore how to participate in the
to learn to make better products. These results respective blogs they found. Since participation is
contribute to personal involvement. Furthermore, multi-layered the teacher could also encourage the
the experience will generate a long-term memory participants to use other Web 2.0 technologies on
of the new knowledge and skills that are gained the blog such as widgets or avatar creation (meez.
(Doolittle, 1999). com). Assessments of the participating learners
Another significant aspect which relates Web should include measure to assess whether they
2.0 applications to constructivist settings is that learned and used the respective Web 2.0 technolo-

192
The Power and Promise of Web 2.0 Tools

Table 1. An example of implementing Web 2.0

The learners will be able to


     1. Identify what a learning strategy is.
     2. Explain what a lesson plan is.
     3. Create a lesson plan.
     4. Explain what a blog is.
     5. Use a blog to post his/her lesson plan.
     6. Provide and receive feedback.

gies. Using survey instruments such as Zoomerang course objectives are accomplished, and the levels
or SurveyMonkey can provide feedback to faculty of interaction with content, peers and the instruc-
about the level of learning. tor (Legon, 2007). Can Web 2.0 tools increase the
Zone of Proximal Development (ZPD)? Research
centered on making the transition to the Web 2.0
Why Adopt Web 2.0 Tools? world and the use of podcasting audio technolo-
gies to promote collaborative learning uncovered
There are numerous benefits of using Web 2.0 some interesting findings. Lee, Mcloughlin and
tools: access to social networking tools, knowledge Chan (2008) found that students were empowered
sharing, digital publishing, and cost effectiveness, by the creation of rich content, as was evident in
are just a few. Computerworld magazine (2008) their own reflections and analysis of constructivist
reports that Web 2.0 tools such as wikis, blogs, thought. However, Sandars and Schroter (2007)
and social networks are being used by many surveyed 3000 medical students and found that
corporations (e.g., Intel Corporation) to solve while there was a high familiarity and interest
complex problems, to promote training, and to with a wide variety of Web 2.0 technologies, there
build an online community. Higher education is was hesitancy in how to use these technologies.
also using these emerging tools in creative ways The authors conclude that there needs to be ad-
to promote a higher quality of interactivity and ditional training in the instructional approaches for
collaboration among learners. Palloff and Pratt achieving desired educational outcomes through
(2005) discuss the idea that teaching and learning the power of technology.
involve more than taking old teaching models The increase of online interactivity through
of lecture and transferring them to a different these tools offers solutions to communicating
medium. They describe an engaged learner who ideas collaboratively not only effectively and
collects, create, and recreates knowledge. The efficiently, but also with a high degree of creativ-
definition of dialogue is enhanced to included ity. Interactivity is complex and important to the
Web 2.0 tools such as blogs, synchronous audio learning process. There is interactivity between the
and video conferencing. Social constructivist learner and the teacher, the learner and peers, and
theory emphasizes the negotiation of meaning the learner and the technology. Woo and Reeves
and construction of shared understandings through (2007) deconstruct the notion of interaction in
dialogue (Jonassen, Davidson, Collins, Campbell, online learning through the social constructivist
& Haag, 1995; Bonk & Kim, 1998). lens and show how it can promote meaningful
There are numerous issues with the quality of learning. They conclude:
online or blended learning courses, including how

193
The Power and Promise of Web 2.0 Tools

The bottom line is that to increase the learning content, and (3) the lack of perceived need. Clearly
effects of online interaction, we should, first of there needs to be more of an acculturation of Web
all, understand clearly the nature of interaction 2.0 tools so that current and future educators can
within the framework of social constructivist be more confident in effectively using them in
learning theory. Once we gain such an in-depth their practice. In some cases there will be a need
understanding, we should be able to engage in for an epistemic instructional transformation for
productive research and development to identify faculty who are in various stages of adopting new
the necessary design principles for implementing technologies.
more effective interaction activities within web- Campus Technology magazine published an
based learning environments. (p.23) article in April 2008 about creative ways that
college instructors have integrated these tools
Likewise, according to Jonassen (1995), con- into their practice. Examples include the use of
structivists use technologies for purposes that are a wiki that allows students to submit exam ques-
(1) applied to real-world situations, (2) problems tions for consideration. A portal is a system that
and constructs, and (3) authentic and appropriate has many functions including serving as an online
social context. There are numerous benefits to community and a means to catalog or organize
integrating the Web 2.0 tools as learners are em- different digital creations and provide them to
powered through the creation of content and the members in an effective manner.
many forms of communication. What are some of Alexander (2006) describes Web 2.0 tools
these Web 2.0 tools and how are they being used and how this “heterogeneous mix of relatively
in higher education? familiar and emergent technologies” (p.33) is used
in a social networking context. Another similar
example is the University of Alaska Anchorage’s
Web 2.0 and Higher Education Web 2.0 wiki which not only defines what a Web
2.0 tool is but how this online learning instrument
Thompson (2007) reasoned that “Today’s institu- can be used in an educational setting. Graduate
tions of higher education (IHEs) need to consider students enrolled in an instructional design course
how they can move from being Education 1.0 collaborated to create the content on the wiki.
institutions because their competitors are revising Each learner was given editor status so that they
how they provide services and coursework using could post their content defining and describing
Web 2.0 applications” (p.40). However, Eijkman the technology but also sharing their creativity
(2008) believes that “. . . the effective educational by sharing how they used the tool for a positive
use of Web 2.0 will in due time lead to a radical learning outcome.
reframing of our educational thinking and practices In addition to the most popular (blogs and
and a redesign of digital learning spaces around wikis), other Web 2.0 tools include social book-
interdependent acculturation . . .” (p. 102). From marking (de.li.ci.ous), collaborative webpage
the standpoint of the learner, Karpinski (2008) design (jot.com), and (Gnosh.org). Wix.com is a
found that a majority of learners surveyed were free online tool that helps the web designer with
unfamiliar with Web 2.0 resources. While students the development of a Flash interface website.
may be familiar with the Web 2.0 tools there was This intuitive tool offers “what you see is what
hesitancy on their part to adopt them because of (1) you get” (WYSIWYG) design that is dynamic
difficulty of use (2) skepticism about the quality of and fluid.

194
The Power and Promise of Web 2.0 Tools

Conferencing software is a tool that allows for ber of an online learning community” (p.7). Lave
the synchronous sharing of ideas. Commercial and Wenger (1991) state: “A learning curriculum
products include HorizonLive, Elluminate, and is not something that can be considered in isola-
Adobe Breeze. WizIQ is a free online conferenc- tion, manipulated in arbitrary didactic terms, or
ing software which provides the interactivity of a analyzed apart from the social relations that shape
whiteboard, webcam, direct messaging, and more, legitimate peripheral participation” (p. 97). But
(http://wiziq.com). Understandably, Alexander what is meant by “actively participate” and which
concludes: “Meanwhile, academic implementa- technologies and approaches support positive
tions are bubbling” (2006, p.44) with the advent learning experiences and promote quality interac-
of Web 2.0. tion or legitimate peripheral participation?
While there have been studies on interactiv-
ity and collaborative learning through commu- Social Networking
nity building (e.g. Davies, Ramsay, Linfield,
& Couperthwaite, 2005; Palloff & Pratt, 2005; Social networking Sites (SNS) applications allow
Sherry, 2000), Hung (2001) contends there still participants to belong to “friends” communi-
remains much research on how the social inter- ties where individuals are invited to be friends.
actions that are an inherent Web 2.0 component Two well-known applications are MySpace and
support the Vygotskian claim that cognition begins Facebook. On the other hand, social bookmark-
at the social level. What does this mean in terms ing applications are used to help users organize
of utilizing Web 2.0 technologies for facilitation information by saving different web sites in one
of Learning 2.0? What is the educational utility location because the user deems these web sites
of the Web 2.0 tools? important for learning and research. Tufekci
(2008) did a study on social networking sites
(SNS) by college students and draws an impor-
The Opportunities of Web 2.0 tant distinction between the “expressive Internet”
with an emphasis on the social interactions and
The purpose of this section is to convey how Web networking and the “instrumental” Internet that
2.0 applications represent and provide opportuni- focuses on e-commerce and news. Users of SNS
ties for constructivist learning. The capacity to are also much more frequent users of the expres-
apply and construct new knowledge, to encour- sive Internet.
age critical thinking, promote collaboration, and Global access and democracy can easily in-
to use guided scaffolding to support higher level tertwine to create new possibilities. For example,
thinking and learning is at the very center of the what if classrooms in Tasmania were paired with
constructivist paradigm. It has been argued that their counterparts in Alaska to study polar science
higher education needs to move “beyond the from an Antarctic and Arctic perspective? The
individual mind to include learning that is built learners could collaborate to write and illustrate
up by mediated conversations among members a report that could involve topics such as global
of peer groups, local learning communities, and warming, environmental, or ecological science.
broader cultural systems” (Sherry, 2000, p. 21). The learners could communicate through web-
Conrad and Donaldson (2004) also take this point based synchronous and asynchronous mediums.
of view: “Engaged learning stimulates learners to The Internet has made digital space and the
actively participate in the learning situation, and ZPD much larger and the world much smaller.
thus gain the most knowledge from being a mem- Two video game players can enjoy the same

195
The Power and Promise of Web 2.0 Tools

game via separate individual web access points development it has been widely adopted by higher
without regard to their location or time zone; that education departments, workgroups, ESL classes,
is the larger space. These games can be embed- significant interest groups, and so on.
ded within a blog, webpage, or wikis (See, for Web 2.0 tools support many forms of social
example, http://onlinegames.net for the extract- networking and collaboration, including those
able and embeddable code). With the possibility which a more revolutionary sense of emancipatory
that one of these players may be in the USA and Freirean education, such as Flash mobs, groups that
other may be in China, the world reach becomes get together physically to demonstrate or protest
smaller. The world of Web 2.0 applications such using digital technologies (e.g. LiveJournal.com).
as MySpace (http://www.myspace.com/) and There are countless possibilities.
FaceBook is also small; through them, a person
may have hundreds, if not thousands, of friends Widgets and Plugins
around the world. These and similar social net-
working applications make global interaction Other tools include widgets or interactive add-
and cross-hemispheric collaboration easier and ons to supplement the learning experience. These
provide settings for global communities. widgets can be embedded into course blogs to
Another example of a Web 2.0 tool that achieve a variety of results. There are numerous
promotes interactivity and community is the widgets to help with the learning of language
Ning. A Ning aggregates Web 2.0 tools such as and testing of skills on a variety of topics, from
a webpage discussion forum, file sharing, within tracking the financial markets, to creating artwork
its multi-functional framework. While MySpace like Matisse. A good example of a widget used in
and Facebook have been used for several years, education is to have it deliver course content that
a Ning allows the user to create their own net- the students can embed in their individual web
work and not just be a participant in a network sites. Various widgets include embedded videos,
that has already been created. It is a means by photos, podcasts, maps, all aggregated on one
which private or public groups with a shared page (pageflakes.com) (The Chronicle of Higher
interest can construct digital space and interact Education, 2008).
through a variety of mediums online. The Ning Web 2.0 tools may fill a specific need or ad-
at http://Classroom20.ning.com is an example of dress a certain problem, such as the management
constructed digital space for educators who can of the enormous amount of digital information.
share how they are integrating technology into For example, a pragmatic plug-in tool for brows-
their practice. It is a dynamic community that ers is Real Simple Syndication (RSS). RSS feeds
is changing because of the content constructed allow the user to aggregate various resources (e.g.
by participants. Nings have been created for a blogs, wikis libraries, periodicals, websites) when
myriad of groups including those focused on action new content has been added. Another example
research, Scholarship of Teaching and Learning is Zotero, a free plug-in for anyone who needs
(SoTL), computer enthusiasts, history buffs, etc. assistance when managing reference data from
It takes only a few minutes to construct a Ning online database searches. It helps organize the
and create a viable network that aggregates the citations of research articles and makes the writ-
virtual social presence of learners, their thoughts ing experience flow easier because the knowledge
and discussions, and digital artifacts. The Ning can be arranged, classified, and shared. It also
has enormous potential for providing structured facilitates the importing and exporting of infor-
community and while it is in its early stages of mation. For example, you can capture citations

196
The Power and Promise of Web 2.0 Tools

from webpages, store documents, images, and aspects of their lives”(Julyan & Duckworth, 1996,
other indexed citations from large databases. The p.58). We can use these Web 2.0 tools to transform
user has the ability to co-create digital libraries our practice and enhance the reality of learning.
and be notified via Really Simple Syndication The dialectic between the individual and society
(RSS) when a relevant research publication is through the use of Web 2.0 tools is expanding
tagged for retrieval from similar articles. Filters dramatically as these new possibilities suggest
can also be set up for advanced searches. Users that learning is a constructive building process
can also write annotations that will be saved with of meaning-making. Faculty should adopt these
each reference. constructivist tools or run the risk of “engage or
enrage” (Prensky, 2005). Faculty cannot continue
to teach as they were taught but rather they must
Future Trends employ the interactive Web 2.0 tools to engage
learners. This is the power and promise of Web
The question “What issues will Web 2.0 ad- 2.0 tools.
dress?” was asked earlier in this chapter. The
new approaches to learning with digital media
have the potential to enhance the higher educa- References
tion learning experience. We are at the nexus
of a historic transformation in how learning is Addison, C. (2006). A new chapter in web develop-
shared, transmitted, collected, and constructed. ment? Development in Practice, 16(6), 623–627.
The social web has seen a profound shift in higher doi:10.1080/09614520600958348
education and the building of community through Alexander, B. (2006). A new wave of innovation
virtual social networking sites. As the facilitation for teaching and learning. Educause (March/April)
of online learning evolves (artificial intelligence 33-44. Retrieved April 5, 2008 from http://www.
and non-human facilitation, web 3.0?) it becomes educause.edu/ir/library/pdf/ERM0621.pdf
necessary to adopt Web 2.0 tools in the design of
effective and efficient e-learning. Alexander, B. (2008). Web 2.0 and emergent mul-
tiliteracies. Theory into Practice, 47(2), 150–160.
doi:10.1080/00405840801992371
Conclusion Bonk, C. J., & Kim, K. A. (1998). Extending
sociocultural theory to adult learning. In C. M.
This chapter discussed the promise of using Web
Smith & T. Pourchot (Eds.), Adult learning and
2.0 technologies to extend the classroom. Some
development: Perspectives from educational
of the advantages associated with implementa-
psychology (pp. 67-88). Mahwah, NJ: Lawrence
tion of Web 2.0 were listed and discussed. The
Erlbaum.
potential for a revolution of learning that involves
constructivism will inherently involve these tools. Brown, J. S., & Adler, R. P. (2008). Minds on
“Constructing an understanding requires that the fire: Open education, the long tail, and learning
students have opportunities to articulate their ideas, 2.0. EDUCAUSE Review, (January/February):
to test those ideas through experimentation and 17–32.
conversation, and to consider connections between
Computerworld. (2008). IT will “take head out
the phenomena that they are examining and other
of sand” on Web 2.0 in 2008, 42(6), 5.

197
The Power and Promise of Web 2.0 Tools

Connor, E. (2007). Using Wiki technology to Ge, S., & Tok, M. (2003). Enhancing online edu-
build a faculty publications database. Journal of cation using collaboration solutions. Journal of
Electronic Resources in Medical Libraries, 4(4), Educational Technology Systems, 31(4), 361–380.
11–25. doi:10.1300/J383v04n04_02 doi:10.2190/BGQM-PFF7-84AD-XCA8
Conrad, R. M., & Donaldson, J. A. (2004). Glogoff, S. (2005). Instructional blogging: Pro-
Engaging the online learner: Activities and re- moting e-learning, student-centered learning, and
sources for creative instruction. San Francisco: peer input. Innovate, 1 (5). http://www.innova-
Jossey-Bass. teonline.info/index.php?view=article&id=126
Davies, A., Ramsay, J., & Linfield, H. (2005). Hung, D. (2001). Design principles for web-based
Building learning communities: Foundations for learning: Implications for Vygotskian thought.
good practice. British Journal of Educational Educational Technology, 41(3), 33–40.
Technology, 36(4), 615–628. doi:10.1111/j.1467-
It’s a Web 2.0 world: Are you ready for it? (2003,
8535.2005.00539.x
April). Campus Technology (pp. 43-50).
Dearstyne, B. W. (2007). Blogs, mashups, &
Jonassen, D. (1999). Designing constructivist
wikis. The Information Management Journal,
learning environments. In Reigeluth, C. M. (1999).
41(4), 24–33.
Instructional Design Theories and Models: A
Dede, C. (1996). Distance learning: Distributed New Paradigm of Instructional Theory, 2, 215-
learning: Making the transformation. Learning 241. Mahwah, New Jersey: Lawrence Erlbaum
and Leading with Technology, 23, 25–30. Associates, Publishers.
Doll, W. (1993). A post-modern perspective on Jonassen, D., Davidson, M., Collins, M., Camp-
curriculum. New York: Teachers College Press. bell, J., & Haag, B. B. (1995). Constructivism and
computer-mediated communication in distance
Doolittle, P. (1999). Constructivism and online
education. American Journal of Distance Educa-
education. Virginia Polytechnic Institute and
tion, 9(2), 7–26.
State University. Retrieved July 2008 from:
edpsychserver.ed.vt.edu/workshops/tohe1999/ Julyan, C., & Duckworth, E. (1996). A constructiv-
text/doo2.pdf ist perspective on teaching and learning science. In
C. Twomey-Fosnot (Ed.), Constructivism: Theory,
Eijkman, H. (2008). Web 2.0 as a non-founda-
perspectives, and practice. (pp. 55-72). Columbia
tional network-centric learning space. Campus-
University. Teachers College Press.
Wide Information Systems, 25(20), 93–104.
doi:10.1108/10650740810866567 Karpinski, J. L. (2008). Disconnected in a con-
nected world: Knowledge and understanding of
Franklin, G. (2000). Augmenting print-based dis-
Web 2.0 tools at the University of Pennsylvania
tance learning with other technologies can enhance
Medical Center. Medical Reference Services
its effectiveness for adult learners. Journal of
Quarterly, 27(1), 53–72.
Educational Technology Systems, 28(4), 327–334.
doi:10.2190/AJ1U-VD37-9J59-PH24 Lave, J., & Wenger, E. (1991). Situated learning:
Legitimate peripheral participation. Cambridge,
Gagne, R. M., Wager, W. W., Golas, K. C., & Keller,
MA: Cambridge University Press.
J. M. (2005). Principles of instructional design.
(5th ed) Belmont, CA: Wadsworth Publishing.

198
The Power and Promise of Web 2.0 Tools

Lee, M. J., Mcloughlin, C., & Chan, A. (2008). Somerville, M. M., & Nino, M. (2007). Collab-
Talk the talk: Learner-generated podcasts as orative co-design: A user-centric approach for
catalysts for knowledge creation. British Jour- advancement of organizational learning. Perfor-
nal of Educational Technology, 39(3), 501–521. mance Measurement & Metrics, 8(3), 180–188.
doi:10.1111/j.1467-8535.2007.00746.x doi:10.1108/14678040710841054
Legon, R. (2007). Online courses lead the way in The Chronicle of Higher Education. (2008, Sep-
quality assurance. Sloan-C View: Perspectives in tember 11). Professor uses web ‘Widgets’ to share
Quality Online Education 6(1). Retrieved July 20, course content. Retrieved October 10, 2008, from
2007 from http://www.sloan-c.org/publications/ http://chronicle.com/wiredcampus/article/3307/
view/v6n1/viewv6n1.htm professor-uses-web-widgets-to-share-course-
content.
McFedries, P. (2006). The web, take two.
IEEE Spectrum, 43(6), 68–68. doi:10.1109/ Thompson, (2007). Web 2.0 takes on colleges and
MSPEC.2006.1638049 universities: The dawn of education 2.0. Retrieved
April 29, 2008 from http://www.masternewmedia.
Palloff, R., & Pratt, K. (2005). Collaborating
org/news/2007/04/20/Web_20_takes_on_col-
online: Learning together in community. San
leges.htm
Francisco: Jossey-Bass.
Tufekci, Z. (2008). Grooming, gossip, Fa-
Prensky, M. (2006). Engage me or enrage me: What
cebook and Myspace. Information Com-
today’s learner’s demand. Educause October, 60-
munication and Society, 11(4), 544–564.
64. Retrieved September 17, 2008 from http://net.
doi:10.1080/13691180801999050
educause.edu/ir/library/pdf/erm0553.pdf
Twomey-Fosnot, C. (1996). Constructivism: A
Sandars, J., & Schroter, S. (2007). Web 2.0 tech-
psychological theory of learning. (pp. 8-33). In C.
nologies for undergraduate and postgraduate
Twomey-Fosnot (Ed.), Constructivism: Theory,
medical education: an online survey. Postgraduate
perspectives, and practice. New York: Teachers
Medical Journal, 83(986), 759–762. doi:10.1136/
College Press.
pgmj.2007.063123
Von Glasersfeld, E. (1996). Introduction: As-
Schrum, L., & Hong, S. (2002). Dimensions and
pects of Constructivism. In C. Twomey-Fosnot
strategies for online success: voices from expe-
(Ed.). Constructivism: Theory, perspectives and
rienced educators. JALN, 6(1). Retrieved August
practice (pp. 3-7). Teachers College, Columbia
11, 2002 from http://www.aln.org/alnWeb/journal/
University.
jaln-vol6issue1.htm#schrum
Woo, Y., & Reeves, T. C. (2007). Meaning-
Sherry, L. (2000). The nature and purpose of
ful interaction in web-based learning: A social
online conversations: A brief synthesis of current
constructivist interpretation. The Internet and
research. International Journal of Educational
Higher Education, 10(1), 15–25. doi:10.1016/j.
Telecommunications, 6(1), 19–52.
iheduc.2006.10.005
Zualkernan, I. A. (2006). A framework and a
methodology for developing environments. Edu-
cational Technology & Society, 9(2), 198–212.

199
The Power and Promise of Web 2.0 Tools

Appendix

Tool Example Link


Open Source Open Office http://Openoffice.org
Tools Linux http://www.linux.org
The Open Source Initiative http://www.opensource.org/
Source Forge http://sourceforge.net/
Open CourseWare Massachusetts Institute of Tech (MIT) http://ocw.mit.edu
Carnegie-Melon University http://www.cmu.edu/oli/
Utah State University http://ocw.usu.edu/
Portals Merlot (Multimedia Educational http://merlot.org
Resources for Learning and Online http://www.uportal.org/
Teaching)
UPortal
Collaborative design Jot http://jot.com
& Project Gnosh http://gnosh.com
Management Project Zoho http://projects.zoho.com/jsp/home.jsp
Survey Instruments Zoomerang http://zoomerang.com
Survey Monkey http://surveymonkey.com
Video Blogs Video Blogs http://www.vblog.com/
Directory of Web 2.0 http://www.go2Web20.net/ http://www.allthingsWeb2.com/
Tools Centre for Learning & Performance http://c4lpt.co.uk/handbook/elearning20.html
Technologies
Social Networks Ning.com http://education.ning.com
http://www.classroom20.com/
Blogs Blogger.com http://asdtechnology.blogspot.com
Wikis PBwiki.com http://pbwiki.com
wikispaces.com http://wikispaces.com
Video Blogs Vobbo http://www.vobbo.com/
Webpage design Weebly http://Weebly.com
FreeWebs http://webs.com
Google websites http://sites.google.com/
Conferencing HorizonLive Elluminate Live! Breeze, http://Elluminate.com
Software WizIQ; www.wimba.com
http://www.adobe.com/resources/breeze/
http://www.wiziq.com
Social bookmarking De.li.ci.ous http://delicious.com/
Widgets Widget Box http://www.widgetbox.com/tag/education
http://pageflakes.com
Online Answers One Big U http://www.onebigu.com/user/homepage.php
Games Online Games http://onlinegames.net

This work was previously published in Information Technology and Constructivism in Higher Education: Progressive Learn-
ing Frameworks, edited by C. R. Payne, pp. 146-158, copyright 2009 by Information Science Reference (an imprint of IGI
Global).

200
201

Chapter 1.13
Integrating Web 2.0
Technologies within
the Enterprise
R. Todd Stephens
AT&T Corporation Collaboration & Online Services, USA

Abstract What is missing is a framework or roadmap in


which organizations can plan out their execution
This chapter examines the elements of the new Web of We 2.0 deployment.
2.0 technology base and reviews the lessons learned
when implementing these technologies. Collab-
orative applications have made enormous inroads Introduction
into the enterprise and bring unprecedented speed
and transparency to communications. Researchers Enterprises are being transformed from an old
and practitioners alike are focusing on how col- business model built around the command and
laborative applications can replace the one-way control aspects information management to a new
communications inherent to Intranet sites. This one where collaboration is the essential component.
chapter is intended for individuals who are look- We are witnessing this transformation outside the
ing toward the possibility of integrating these new enterprise with the success of Web 2.0 technologies
technologies into the core communication medium. like Wikipedia, YouTube, and Flickr. Yet, within
Unfortunately, there are still large barriers such as the walls of the organization, progress is slow.
politics, turf battles, integration, and poor usability This chapter will examine why the adoption rates
with the current product set. A company’s ability to for Collaborative and Social Software are low and
manage information effectively over its life cycle, what can be done to improve them. This research
including sensing, collecting, organizing, process- emerges from three Fortune 500 companies which
ing, and maintaining information, is crucial to the the resulting methodology has worked to increase
long term success in a global economy. The success the adoption rates ten fold.
or failure of this integration may very well create Traditional implementation methodologies fo-
or lose a competitive advantage for the enterprise. cused on the hardware, software, and the associated
functionality. Our research indicates that while these
DOI: 10.4018/978-1-60566-106-3.ch037 are important they do not lead to mass adoption of

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Integrating Web 2.0 Technologies within the Enterprise

the application by the enterprise. Many functions information as well as built a sense of community
that information workers perform are dictated by across organizational boundaries (McNay, 2000).
the business and current transactional systems like Typical organizations will have office-based
CRM, ERP, or accounting systems. Collabora- employees in various locations, telecommuting,
tion and Social Software need to be integrated and off-shoring staff. The traditional day by day
as situational applications and although they are communication landscape has changed from
optional, they are required to reach higher levels personal to electronic. The migration to electronic
of performance. Members of complex teams are communications emerged as standards, technol-
less likely to utilize collaborative tools in the ogy and infrastructure matured. This allowed more
absence of other influences such as executive information sharing and community building to
encouragement, modeling collaborative behavior, occur without a requirement of physical loca-
creating a “gift” culture, training, supporting a tion. Over the past several years Intranets have
strong sense of culture, assigning team leaders, emerged as the key delivery mechanism for ap-
building, and understanding roles (Erickson & plication and business information. Intranets may
Gratton, 2007). Our research indicates barriers be thought of as providing the infrastructure for
to adoption including social issues, cultural is- intra-organizational electronic commerce (Chel-
sues, awareness issues, educational issues, and lappa & Gupta, 2002). This allows organizations
political issues. This research will focus on the to utilize the technology to achieve its organiza-
awareness and educational issues since most tional goals and objectives. Web 1.0 allowed the
organizations implementing Web 2.0 will face organization to govern the information flow and
these first issues first. focus on achieving the business goals.
Unfortunately, most technologies fail to deliver
competitive advantages over an extended period
Background of time. Investments in information technology,
while profoundly important, are less and less likely
Web 1.0 Intranets to deliver a competitive edge to an individual
company (Carr, 2003). This is especially true in the
The term Web 1.0 emerged from the research world of the Web 1.0 since much of the knowledge
around Web 2.0. Basically, Web 1.0 focused on a and information is disseminated all over the world
read only Web interface while Web 2.0 focuses on as quickly as it gets published. Organizations are
a read-write interface where value emerges from beginning to see that the command and control
the contribution of a large volume of users. The model is no longer effective at developing a high
Internet as well as the Intranet initially focused performance work force which opens the door for
on the command and control of the information the next evolution in technologies as described by
itself. Information was controlled by a relative the Web 2.0 framework.
small number of resources but distributed to a
large number which spawned the massive growth Web 2.0 Defined
of the Web itself. Like television, the Web allowed
for the broadcasting of information to a large While Web 2.0 has been debated by researchers
number of users. as to who and when the concepts emerged, little
Inside the organization, the Intranet has argument exists that the technology and demand
changed the way organizations structure and has arrived. Unlike Web 1.0, this new technol-
operate their business. Specifically, the Intranet ogy encourages user participation and derives its
has centralized communications and corporate greatest value when large communities contribute

202
Integrating Web 2.0 Technologies within the Enterprise

content. User generated metadata, information, straints by providing a standard user interface
and designs enable a much richer environment that does not require customization. Weblogs
where the value is generated by the volume of originally emerged as a repository for linking but
employees. Sometimes referred to as sharing, soon evolved to the ability to publish content and
collaboration, aggregate knowledge, or com- allow readers to become content providers. The
munity driven content, social software creates essence of a blog can be defined by the format
the foundation of collective intelligence (Weiss, which includes small chunks of content referred
2005). Much of the Web 2.0 technology is dif- to as posts, date stamped, reverse chronological
ficult to nail down with an exact definition; the order, and content expanded to include links, text,
basic truth is that Web 2.0 emphasizes employee and images (Baoill, 2004). The biggest advance-
interaction, community, and openness (Millard ment made with Weblogs is the permanence of the
& Ross, 2006). Along with these characteristics, content which has a unique Universal Resource
Smith and Valdes (2005) added simple and light- Locator (URL). This allows the content to be
weight technologies and decentralized processing posted and along with the comments to define a
to the mix. O’Reilly (2005) defined Web 2.0 as permanent record of information. This is critical
a platform, spanning all connected devices; Web in that having a collaborative record that can be
2.0 applications are those that make the most of indexed by search engines will increase the utility
the intrinsic advantages of that platform: deliv- and spread the information to a larger audience.
ering software as a continually-updated service With the advent of software like Wordpress and
that gets better the more people use it, consuming Typepad, along with blog service companies like
and remixing data from multiple sources, includ- blogger.com, the weblog is fast becoming the
ing individual users, while providing their own communication medium of the new Web.
data and services in a form that allows remixing
by others, creating network effects through an Sample Weblog URLs
“architecture of participation,” and going beyond
the page metaphor of Web 1.0 to deliver rich user • Andrew McAfee Web 2.0 Blog (http://
experiences. While Web 2.0 has many and often blog.hbs.edu/faculty/amcafee/)
confusing definitions most include the concepts • Randy Basler’s Boeing Blog (http://boe-
of Weblogs, Wikis, Really Simple Syndication ingblogs.com/randy/)
(RSS) Functionality, social tagging, mashups, • Jonathan Schwartz’s Sun Blog (http://
and user defined content. blogs.sun.com/jonathan/)
• Rough Type by Nicholas Carr (http://www.
Weblogs or Blogs roughtype.com)

Weblogs or blogs have become so ubiquitous that Wikis


many people use the term synonymous for a “per-
sonal Web site” (Blood, 2004). Unlike traditional A Wiki is a Web site that promotes the collaborative
Hypertext Markup Language (HTML) Web pages, creation of content. Wiki pages can be edited by
blogs offer the ability for the non-programmer anyone at anytime. Informational content can be
to communicate on a regular basis. Traditional created and easily organized within the wiki envi-
HTML style pages required knowledge of style, ronment and then reorganized as required (O’Neill,
coding, and design in order to publish content 2005). Wikis currently are in high demand in a large
that was basically read only from the consumer’s variety of fields, due to their simplicity and flex-
point of view. Weblogs remove much of the con- ibility nature. Documentation, reporting, project

203
Integrating Web 2.0 Technologies within the Enterprise

management, online glossaries, and dictionaries, date, and copyrights to anyone that subscribes to
discussion groups, or general information applica- the feed. The end user is required to have an ap-
tions are just a few a examples of where the end plication called an aggregator in order to receive
user can provide value (Reinhold, 2006). The the information. By having the RSS aggregator
major difference between a wiki and blog is that application, end users are not required to visit each
the wiki user can alter the original content while site in order to obtain information. From an end
the blog user can only add information in the form user perspective, the RSS technology changes the
of comments. While stating that anyone can alter communication method from a search and discover
content, some large scale wiki environments have to a notification model. Users can locate content
extensive role definitions which define who can that is pertinent to their job and subscribe to the
perform functions of update, restore, delete, and communication.
creation. Wikipedia, like many wiki type projects,
have readers, editors, administrators, patrollers, Sample RSS URLs
policy makers, subject matter experts, content
maintainers, software developers, and system • Newsgator (http://www.newsgator.com/)
operators (Riehle, 2006), all of which create an • FeedBurner (http://www.feedburner.com/)
environment open to sharing information and • Pluck (http://www.pluck.com/)
knowledge to a large group of users. • Blog Lines (http://www.bloglines.com/)

Sample Wiki URLs Social Tagging

• Disney’s Parent Wiki (http://family. Social tagging describes the collaborative activity
go.com/parentpedia) of marking shared online content with keywords
• Wikipedia (http://www.wikipedia.org/) or tags as a way to organize content for future
• Reuters Financial Glossary (http://glos- navigation, filtering, or search (Gibson, Teasley,
sary.reuters.com/) & Yew, 2006). Traditional information architec-
• Internet 2 (https://wiki.internet2.edu/con- ture utilized a central taxonomy or classification
fluence/dashboard.action) scheme in order to place information into specific
pre-defined bucket or category. The assumption
RSS Technologies was that trained librarians understood more about
information content and context than the average
Originally developed by Netscape, RSS was user. While this might have been true for the local
intended to publish news type information based library with the utilization of the Dewey Decimal
upon a subscription framework (Lerner, 2004). system, the enormous amount of content on the
Many Internet users have experienced the frus- Internet makes this type of system un-manageable.
tration of searching Internet sites for hours at Tagging offers a number of benefits to the end user
a time to find relevant information. RSS is an community. Perhaps the most important feature
XML based content-syndication protocol that to the individual is able to bookmark the informa-
allows Web sites to share information as well as tion in a way that is easier for them to recall at a
aggregate information based upon the users needs later date. The benefit of this ability on a personal
(Cold, 2006). In the simplest form, RSS shares basis is obvious but what about the impact to the
the metadata about the content without actually community at large. The idea of social tagging is
delivering the entire information source. An au- allowing multiple users to tag content in a way that
thor might publish the title, description, publish makes sense to them; by combining these tags,

204
Integrating Web 2.0 Technologies within the Enterprise

users create an environment where the opinions services will enable greater extensions of mash-
of the majority define the appropriateness of the ups and combine many different businesses and
tags themselves. The act of creating a collection business models. Organizations, like Amazon and
of popular tags is referred to as a folksonomy Microsoft are embracing the mash-up movement
which is defined as a folk taxonomy of important by offering developers easier access to their data
and emerging content within the user community and services. Moreover, they’re programming their
(Ahn, Davis, Fake, Fox, Furnas, Golder, Marlow, services so that more computing tasks, such as
Naaman, & Schachter, 2006). The vocabulary displaying maps onscreen, get done on the users’
problem is defined by the fact that different users Personal Computers rather than on their far-flung
define content in different ways. The disagreement servers (Hof, 2005)
can lead to missed information or inefficient user
interactions (Boyd, Davis, Marlow, & Naaman, Sample Mashup URLs
2006). One of the best examples of social tagging
is Flickr which allows user to upload images and • Housing Maps: (http://www.housingmaps.
“tag” them with appropriate metadata keywords. com/)
Other users, who view your images, can also tag • Chicago Crime (http://www.chicagocrime.
them with their concept of appropriate keywords. org)
After a critical mass has been reached, the result- • Healthcare Product (http://www.vimo.
ing tag collection will identify images correctly com/)
and without bias. • Global Disease Map (http://healthmap.
org/)
Sample Social Tagging URLs
User Contributed Content
• Flickr (http://www.flickr.com/)
• YouTube (http://www.youtube.com/) One of the basic themes of Web 2.0 is user con-
• Del.icio.us (http://del.icio.us/) tributed information. The value derived from the
• Technorati (http://technorati.com/) contributed content comes not from a subject
matter expert, but rather from individuals whose
Mashups: Integrating Information small contributions add up. One example of user
contributed content is the product review systems
The final Web 2.0 technology describes the ef- like Amazon.com and reputation systems used
forts around information integration or sometimes with ebay.com. A common practice of online
referred to as “mashups.” These applications can merchants is to enable their customers to review
be combined to deliver additional value that the or to express opinions on the products they have
individual parts could not deliver on their own. purchased (Hu & Liu, 2004). Online reviews are
One example is HousingMaps.com that combines a major source of information for consumers and
the Google mapping application with a real estate demonstrated enormous implications for a wide
listing service on Craiglists.com (Jhingran, 2006). range of management activities, such as brand
Other examples include Chicagocrime.org who building, customer acquisition and retention,
overlays local crime statistics onto Google Maps product development, and quality assurance (Hu,
so end users can see what crimes were commit- Pavlou, & Zhang, 2006). A person’s reputation is
ted recently in the neighborhood. Another site a valuable piece of information that can be used
synchronizes Yahoo! Inc.’s real-time traffic data when deciding whether or not to interact or do busi-
with Google Maps. Much of the work with Web ness with. A reputation system is a bi-directional

205
Integrating Web 2.0 Technologies within the Enterprise

medium where buyers post feedback on sellers Web 1.0 Compared to Web 2.0
and vice versa. For example, eBay buyers vol-
untarily comment on the quality of service, their While the differences between Web 1.0 and 2.0 are
satisfaction with the item traded, and promptness grey at best, we can attempt to draw some segmen-
of shipping. Sellers comment about the prompt tation by reviewing the high level characteristics.
payment from buyers, or respond to comments Table 1 provides a side by side comparison of
left by the buyer (Christodorescu, Ganapathy, these technologies.
Giffin, Kruger, Rubin, & Wang, 2005). Reputa- In the Web 1.0 environment, information was
tion systems may be categorized in three basic largely static and controlled by a few resources.
types: ranking, rating, and collaborative. Rank- Specifically, the individual or organization that
ing systems use quantifiable measures of users’ produced this information pushed information to
behavior to generate and rating. Rating systems the end user by either controlling the access or
use explicit evaluations given by users in order limiting the feedback options. Web 2.0 turns that
to define a measure of interest or trust. Finally, model around and create a far greater dynamic
collaborative filtering systems determine the level environment where each consumer has the ability
of relationship between the two individuals before to contribute to the overall value of the information
placing a weight on the information. For example, itself. Instead of searching and browsing topics,
if a user has reviewed similar items in the past Web 2.0 users are allowed to publish and subscribe
then the relevancy of a new rating will be higher to the content which results is a more bottom up
(Davis, Farnham, & Jensen, 2002). implementation. The following section will review
how these new technologies can be integrated into
the current knowledge environments that have
Sample User Contributed traditionally followed the command and control
Content URLs model of information.

• Amazon.com (http://www.amazon.com) Enterprise 2.0


• Ebay (http://www.ebay.com)
• Trip Advisor (http://www.tripadvisor.com/) Enterprise 2.0 is a term used to describe the
• Review Centre (http://www.reviewcentre. integration of the Web 2.0 technology portfolio
com/) inside of the organization. Both the producers and

Table 1. Characteristics of Web 1.0 and Web 2.0

Web 1.0 Characteristics Web 2.0 Characteristics


Static Content Dynamic Content
Producer Based Information Participatory Based Information
Messages Pushed to Consumer Messages Pulled by Consumer
Institutional Control Individual Enabled
Top Down Implementation Bottom Up Implementation
Users Search and Browse Users Publish and Subscribe
Transactional Based Interactions Relationship Based Interactions
Goal of Mass Adoption Goal of Niche Adoption
Taxonomy Folksonomy

206
Integrating Web 2.0 Technologies within the Enterprise

consumers of the information will reside inside Social Software. More importantly, end users have
the organization. If either of the customer clas- not heard of the internal product offering, if one
sifications involve outside entities then the Web exists. Communication is one of the most critical
2.0 tag should be used. McAfee (2006) indicates aspects of letting people know that a collabora-
a new wave of business communication tools tive or social application is available. Traditional
which allow for more spontaneous, knowledge- information technology solutions were focused on
based collaboration. These new tools, the author a single business process and the aspects of mar-
contends, may well supplant other communication keting and branding were unnecessary. However,
and knowledge management systems with their for enterprise services this awareness can be one
superior ability to capture tacit knowledge, best of the most critical functions performed early in
practices and relevant experiences from through- the product’s life cycle. Like e-mail and desktop
out a company and make them readily available Office applications, you want a high degree of
to more users. For all its appeal to the young and awareness across the entire enterprise. A high de-
the wired, Web 2.0 may end up making its greatest gree of awareness would be some where between
impact in business. And that could usher in more 90-100 percent of the information workers within
changes in corporations, already in the throes of the organization.
such tech-driven transformations as globalization
and outsourcing. Indeed, what some are calling Educational Issues
Enterprise 2.0 could flatten a raft of organizational
boundaries; between managers and employees End users may have heard of Web 2.0 through
and between the company and its partners and the media but they still not understand how the
customers (Hof, 2005). technology can be used in a business setting. Once
an end user becomes aware of an application, the
Barriers to Adoption next phase is to ensure that they understand how
the application should be used. The educational
This chapter does not address issues around in- area is critical since most employees above the
frastructure or software selection. The research age of 35 have not used these new types of tech-
wanted to look at the barriers to adoption assum- nologies.
ing all other variables are constant and normally
taken into account on most implementations. The Cultural and Social Issues
following barriers of adoption are not related to
the specific technology. Rather, they focus on the When organizations have overcome the aware-
end user and the major issues impacted them. This ness and educational gaps, then we can make the
makes sense in Web 2.0 the end user contributes statement that the majority of the organization
as much to the success of the implementation as knows the technology is available and what can
any other component. be done with it. They may still choose to use
their older technology that has been used in the
Awareness Issues past. Not with standing political pressure, we are
looking at cultural or social issues. These issues
The awareness issue describes an environment can emerge when end users fear change, afraid
where of the majority of users have never heard of new methods, or prefer to work in a command
of Web 2.0, Enterprise 2.0, Collaboration, and and control model.

207
Integrating Web 2.0 Technologies within the Enterprise

Political Issues which is a collaborative tool that has most of the


Web 2.0 elements described in the prior section.
The final area focuses on the political pressure (See Figure 1.)
organizations place on users. Political pressure
may focus around strategic direction, vendor as- Common Situation
sociations, or organizational structures. In smaller
organizations, these issues may not be as big an While all three implementations varied by size of
impact as in a large distributed environment. the firm, number of employees, and basic infra-
structure, the implementations had one common
characteristic. Flat line growth occurred within
Integration of Web six months in each of the deployments. Flat line
2.0 Technologies growth occurs when new orders show no growth
over a three to six month period. Figure 2 provides
The studied organization is a Fortune 500 tele- the different site metrics collected prior to the
communications company that has gone through implementation of the proposed framework. The
several acquisitions over the past few years. With lines have been cut off to indicate the point in time
the integration of three companies, the presented the framework was applied to the organization.
framework went through several different field In all three cases, the program had a solid be-
trials over the course of three years. This allowed ginning but reached a level of saturation between
the research to apply the framework into three 90 and 120 collaborative sites. For clarity, the
different companies in order to test the validity in number of collaborative sites continued to grow
a real business environment. The framework was but could not outpace the same number of dele-
developed after five years of trials in traditional tions. A deletion occurs when a program, project,
knowledge management systems. The application or resource no longer needs the collaborative or
of the framework to collaborative tools started in social software environment. On average, the
2004 and progressed through 2007. The initial leveling off of site demand occurred between five
deployment focused on Microsoft’s Sharepoint and six months.

Figure 1. Adoption rates prior to implementing the framework

208
Integrating Web 2.0 Technologies within the Enterprise

Figure 2. Implementation business model

Business Model Framework by these two groups.


Starting at the top of Figure 2, traditional
At the highest level, a business model is how an implementations of information technology would
organization creates value. Timmers (1998) de- include the operational and architecture functions.
fine a business in respect to the architecture for Operations would include activities such as hard-
the product, service, and information flows, the ware monitoring, software installation, backup,
benefits for the various actors, and the sources recovery, security, and maintenance. Computer
of revenue. In reality, an organization can have Operations is a critical function to ensure delivery
a variety of business models, each is simply an of a reliable, scalable, and functional infrastruc-
artificial representation of reality which detracts ture. This area must be governed with a high
focus from certain aspects while concentrating degree of control in order to maintain the stability
on others (Kittl, Petrovic, & Teksten, 2001). of the environment. Architecture focuses on the
This research uses the concepts of a business design, planning, and software selection within
model, not in terms of commerce, but focused on the enterprise. Generally speaking, architecture
the various interacting parts required to deliver includes the activities of defining and modeling
business value. Figure 2 provides an overview of the environment which may include the following
the business model used to describe functional architectures: business, application, data, informa-
designation of work. Some researchers refer to tion, technology, and product architecture (Pereira
this as the resource model. & Sousa, 2004). Traditionally, these components
The model describes seven functional areas and focused on ensuring that the environment did not
four portfolio or product areas. Leadership and fail from an infrastructure point of view. The vast
management are two functional areas that will not majority of Web 2.0 implementations will focus
be covered in this chapter to any depth. The basic on these core elements to ensure that the program
idea is that all information technology activities operates effectively. The idea of an implementation
would include coordination, communication, and failing would indicate the hardware or software
cooperation activities which are usually performed failed to deliver the business value consistently

209
Integrating Web 2.0 Technologies within the Enterprise

over time. The problem with this approach is that Business Development
having a perfect infrastructure does not guaran-
tee mass adoption which is the truest measure Located at the bottom right side of Figure 2 is the
of success. In the case of internally developed representation of the customer which could either
Web 2.0 applications or the utilization of open be the producer or a consumer of the information.
source, the architecture area could be expanded The overall framework centers around the cus-
to encompass the entire System Development tomer behavior. The box on the left represents the
Lifecycle (SDLC). functions of business development which we can
We can define success (mass adoption) from define as any activity that impacts awareness and
two perspectives. In any knowledge type of ap- education before someone becomes a customer.
plication, you will have two key customer classes: Like the business functions of marketing, brand-
the producer of the information and consumer of ing, and selling, business development strives to
the information. The producer is the person, com- encourage participation. The ultimate goal is to
munity, or application that creates a reusable asset ensure that every person in the enterprise knows
in the form of information utilizing the Web 2.0 these tools exist, understand how they can be used,
tools. This might include a wiki page or ownership and knows where to go to get engaged.
to a specific weblog. The consumer is responsible
for locating and accessing the information, assess- Client-Support
ing the ability to reuse the information, adapting
to the information and integrating the information The vast majority of information workers are not
into the business. The consumer might not actually familiar with virtual solutions and need guidance
contribute to the environment in the form of com- on how to best utilize and integrate this technol-
ments or informational update. Production rates ogy into their day to day operations. End users
for Web 2.0 applications are still relatively low will ask five basic questions of the collaborative
as compared to the number of users that consume environment:
the information. In one survey, only 11 percent of
respondents would even consider contributing to • What collaborative products and services
Wikipedia while the actual number of contributors are available to me?
is less than 1 percent overall. When you have mil- • How can I utilize these products and ser-
lions of consumers, a 1 percent contributor rate is vices within my environment?
pretty good. However, in an enterprise of 20,000 • Who can help me in case I need some pro-
people that would indicate you will only have 20 fessional guidance?
contributors. This demonstrates the criticality of • Are the collaborative applications ready
building up a producer community towards the for enterprise usage?
long term goal of mass adoption. Not only do we • How am I doing in comparison to others or
need to focus on the information contained within against best practices?
the environment but also with the utilization of that
information. The content must be used and to a In order to address these questions, organiza-
greater degree, the utilization of the content drives tions should look toward developing a support
the return on investment. This is not a trivial point; group that can enable the end user rather than
organizations must focus on the components of hindering their understanding of a collaborative
success and understand that having great hardware, environment. Meeting the needs of the customer
software and functionality is simply the price of may vary depending on the level of knowledge
entry into the Web 2.0 environment. the user brings to the environment. Customers

210
Integrating Web 2.0 Technologies within the Enterprise

who are new to technology expect a high level Sharepoint, Confluence, Social Text, or IBM’s
of reliability and support in order to gain the Connections. These could be considered products
greatest value possible (Johnston & Supra, 1997). from an architecture or operations point of view.
Customer service should not be homogeneous However, from the customer perspective these
and both the online and physical support environ- are tools. A product solves a problem or generates
ments need to take into account the experience value based on the consumption or utilization. In
level of the end user (Dutta & Roy, 2006). The other words, the product is what you can do with
customer wants to know what products, services, the tool. Products in the Web 2.0 space include
and documentation are available to them within weblogs, wikis, collaborative intranets, virtual
the collaborative environment. The content of an workspaces, RSS feed readers, book marking, and
online environment is not limited to the product professional profiles. Each of these can be catego-
or services provided. Rather, content includes the rized as a product regardless of the tool selected to
solutions and strategies employed to make it easy perform the function. Services would include both
for the user to accomplish important tasks, such tangible and intangible value-add activities that go
as information retrieval, search, and obtaining along with the products. Services might include
feedback (Calongne, 2001). Support information templates, user guides, editing, PDF conversion,
or content should include the product and service education, and training. Services must be delivered
quantity, quality, and relevance to the customer to the customer and cannot be inventoried for
(Palmer, 2002). Technologists often make the later use. Taken together, products and services
mistake of assuming a certain level of expertise provide the customer experience that encourages
with the user community. Unlike e-mail or Of- participation and end user involvement.
fice products (Word Processing or Spreadsheet),
collaborative tools are fairly unknown to the end Portfolios
user. A Client-Support environment would include
many of the following components: Not counting the infrastructure, the model iden-
tifies three additional portfolios including the
• Training and Education product portfolio, the service portfolio, and the
• Subject Matter Expert Information business processes. The product portfolio would
• Technical Supports and Operational include the various products developed in the prior
Information section. Since most products are meta-physical in
• Best Practices and “How To” nature, they must be demonstrated in the online
Documentation support environment. The service portfolio will
• Community of Practice describe the services available to the end user.
• Ordering Processes The idea is that products must be demonstrated
• Product, Service and Solution Overviews while services must be described. Finally, the
• End User Metrics of Content and Usage business process portfolio defines the business
processes required to engage in the environment.
Product and Service Development Taken together, a customer may engage with
several products, services and business solutions
Normally when you discuss the concepts around which constitute a solution offering. Assuming
a product, you think of products like Microsoft’s the ultimate goal of any Web 2.0 application is

211
Integrating Web 2.0 Technologies within the Enterprise

the mass adoption of the customer base then you Field Trials
want to move up the value chain. The value-chain
has been well documented by Joseph Pine II and The first implementation of the framework oc-
James H. Gilmore. curred in 2004. As seen in Figure 1, the deployment
Pine and Gilmore (1997) discuss “the ex- had already reached the point of saturation when
perience economy” by tracing the value added executive leadership contacted the author’s team to
to the coffee bean in its various iterations from see if the framework could be applied to the Web
pure “commodity” to pure “experience.” In their 2.0 area. In the past, the framework had been suc-
evolutionary construct there are four stages, in cessfully implemented in knowledge management
ascending order of sophistication the stages are type implementations. These systems included
commodity, good, service, experience. They point prior research in metadata repositories, registries,
out that coffee is traded on the futures market at and other knowledge applications. Focusing on
roughly $1 a pound (thus, about 2 cents a cup at the left most line (Figure 1), demand had leveled
the “commodity” level). After manufacturers roast, off at around 100 collaborative environments
grind, package and distribute the bean for retail, with an average monthly variance of +/-5 percent.
the price jumps to between 5 and 25 cents a cup The initial review and prioritization of activities
(the “goods” level). At a “run-of-the-mill” diner focused on providing the online support environ-
a cup might run from 50 cents to $1 a cup (the ment, automating the procurement process, and
traditional “service” level). The authors contend developing marketing plans. These were seen as
that one can “Serve that same coffee in a five-star obvious gaps in the prior implementation model
restaurant or espresso bar, where the ordering, based on the author’s observations. The author
creation, and consumption of the cup embodies was also able contact various user communities in
a heightened ambience or sense of theatre, and order to ascertain if these gap assumptions were
consumers gladly pay anywhere from $2 to $5 true. No official survey was used to collect the
for each cup.” Thus, by creating value at the information other than informal conversations.
“experience” level, the seller is able to charge an Figure 3 provides the results of implementing the
extremely high premium over that charged by framework over a 36 month period of time.
the “service” provider. In defining their terms By the fall of 2007, the studied organization
they argue that, “When a person buys a service, had over 13,000 collaborative sites with an aver-
he purchases a set of intangible activities carried age monthly growth rate of 423.62 percent. As the
out on his behalf. But when he buys an experi- implementation matured, less focus was placed
ence, he pays to spend time enjoying a series of on the business opportunity area and more on the
memorable events that a company stages, as in a client-support area. This would make sense, since
theatrical play, to engage him in a personal way.” the number of employees that utilizing the collab-
The idea is that organizations that support the Web orative environment was around 95 percent of the
2.0 implementation must move up the value chain employee population. That is to say that 95 percent
in order to obtain the mass adoption of the employees had heard of and used some form
of the application which was determined by the
unique user id logged into the system.
In 2006, the studied organization was pur-
chased by a larger telecommunications company.

212
Integrating Web 2.0 Technologies within the Enterprise

Figure 3. Collaborative sites after the framework application

This purchase brought together three different 1. Have you ever heard of Collaborative or
companies, all of which had an implementation Social Software? (Determine Awareness)
of the collaborative suite. The adoption rates were 2. If you have heard of these, does your orga-
similar to those found in Figure 1, represented by nization utilize them? (Cultural, Political,
the two lines located on the right side of the chart. or Social)
Executive leadership reviewed the implementa- 3. What is the primary use of the Collaborative
tions and determined that the framework needed and Social Tools within your organization?
to be applied in the other two companies as well. (Education)
The initial step would be to survey the new orga- 4. Please describe your overall impression
nizations to see where the issues lie in adoption (use, purpose) of the products in a few
and determine which parts of the framework sentences?
should be emphasized. Unlike the first field trial, 5. Please describe your experience with these
the author had no insight into the new companies tools in a few sentences (accessed infre-
prior to the integration. quently, heavy user, loved it, etc.).

Survey Tool Notice that no specific questions were asked


to differentiate between the cultural, political or
An employee survey was used to determine which social barriers. The reason for this was that the
of the adoption barriers were prevalent. Based percentage was expected to be low as compared
upon some initial interviews, the survey was to the other issues of awareness and education.
distributed via e-mail, and incentives were used In order to ensure the survey group didn’t know
to encourage participation. One hundred employ- Collaborative and Social Software by other names,
ees were randomly selected from the corporate a product matrix was included that showed the
directory. Two follow up e-mails were used to specific products like Sharepoint, Confluence,
encourage users to participate and the average or Open Source. The results of the surveys are
return rate was 72 percent. The questions on the presented in Figure 4.
survey included the following: The surveys were issued in January of 2007

213
Integrating Web 2.0 Technologies within the Enterprise

Figure 4. Survey results (percentage)

and September of 2007. The first response to the community to obtain information and support in
survey showed an enormous awareness issue a self service fashion. The results of these field
in which the vast majority of end users had not trials showed a dramatic increase in demand in
heard of the social offering or did not understand just a few months. On average, the new organiza-
how the tool could be used. Over 77 percent of tions increased by 1,608 percent and 78 percent
the respondents indicated that the lack of aware- respectively. Key actions by the implementation
ness or education was the primary reason for not team included the following:
implementing collaborative solutions in their
environment. Only 2 percent acknowledged the • Expanded the Client-Support environment
actual use of the tools during the few years and to include all three organizations
the remaining 21 percent focused on the cultural, • Expanded the product offering (new releas-
social and political issues. This result would es, Web components, and applications)
indicate the need to address the awareness and • Expanded the service offering for addition-
educational issues first. al training, education, and consulting
• Developed new reward and recognition
Field Trials II, III and IV programs
• Published articles in various organizational
Based upon the results of the survey, awareness newsletters
and education were identified as the two main • Posted to the corporate weblog and wiki
issues. The user community simply did not know daily
the tools existed or how they could be used in a • Developed audio and video training
business environment. The current deployment programs
could easily be expanded to the new organizations • Consolidated business process into a single
will very little effort. For example, the community customer experience
of practice was expanded to include the new or- • Expanded the Community of Practice
ganizations as was the ordering process replicated (CoP)
across all three companies. The client-support was • Developed additional audio and video pro-
centralized as a one stop location for the end user grams for awareness and education

214
Integrating Web 2.0 Technologies within the Enterprise

• Integrated with other Collaborative tools focus their resources on the components that define
like Podcasting, Audio Conferencing, a successful implementation. The results of the
Instant Messaging, and Social Software field trials indicate that within any environment
• Celebrated Key Milestones like the 25,000th or culture, mass adoption can be obtained. With
Collaborative Environment mass adoption, the return on investment of these
technologies will be high.
At the time of this writing, the total number
of collaborative sites had grown to over 28,000
with 2-3 million page views per month. The References
number of document objects exceeded 2 million.
An additional survey was conducted in late 2007 Ahn, L., Davis, M., Fake, C., Fox, K., Furnas, G.,
which was similar to the one described earlier. The Golder, S., et al. (2006). Why do tagging systems
awareness and education problem of 77 percent work? Proceedings of the SIGCHI conference
had dropped to 33 percent within the nine months on Human Factors in computing systems. Mon-
of implementing the framework. treal, Canada: The Association of Computing
The final field trial occurred in mid 2007 Machinery.
with the deployment of Social Software which Baoill, A. (2004). Conceptualizing the weblog:
included weblog, wiki, and book marking applica- Understanding what it is in order to imagine what
tions. The major difference in this effort was that it can be. Interfacings: Journal of Contemporary
the framework was applied from the beginning. Media Studies, 5(2), 1–8.
Similar activities were performed including adding
an online client-support environment, training, Blood, R. (2004). How blogging software reshapes
education, marketing, and extending the offering the online community. Communications of theACM,
with additional products and services. Within 4 47(12), 53–55. doi:10.1145/1035134.1035165
months, the total number of information points
Boyd, D., Davis, M., Marlow, C., & Naaman, M.
exceeded 5,500.
(2006). Social networks, networking & virtual
communities: HT06, tagging paper, taxonomy,
Flickr, academic article, to read. Proceedings
Conclusion of the seventeenth conference on Hypertext and
hypermedia. Odense, Denmark: The Association
In this chapter, we have laid a framework to
of Computing Machinery.
support the implementation of collaborative and
social software. As discussed, users come to this Calongne, C. (2001, March). Designing for
technology with a wide variety of experience website usability. Journal of Computing in Small
levels which cannot be assumed by the technology Colleges, 16(3), 39–45.
community. Client-support must be implemented
Carr, N. (2003). Does IT Matter? Information
if the business wants universal adoption and a high
technology and the corrosion of competitive
degree of business value. The different compo-
advantage. Boston: Harvard Business School
nents of the framework address the concerns of
Press.
the end user which can put them at ease and create
a more open environment for integrating this new Chellappa, R., & Gupta, A. (2002). Managing
technology. The framework focused on the issues computing resources in active intranets. Interna-
with awareness, education, and the offering. By tional Journal of Network Management, 12(2),
addressing these components, organizations will 117–128. doi:10.1002/nem.427

215
Integrating Web 2.0 Technologies within the Enterprise

Christodorescu, M., Ganapathy, V., Giffin, J., Hof, R. (2005). Mix, Match, And Mutate.
Kruger, L., Rubin, S., & Wang, H. (2005). An Business Week Online. Retrieved Octo-
auctioning reputation system based on anomaly ber 1, 2006 from http://www.businessweek.
detection. Proceedings of the 12th ACM confer- com/@@76IH*ocQ34AvyQMA/magazine/
con-
ence on Computer and communications security. tent/05_30/b3944108_mz063.htm
Alexandria, VA: The Association of Computing
Hu, M., & Liu, B. (2004). Mining and Summariz-
Machinery.
ing Customer Reviews. Proceedings of the 10th
Cold, S. (2006). Using Really Simple Conference on Knowledge Discovery and Data
Syndication (RSS) to enhance student re- Mining. Seattle, WA: The Association of Comput-
search. ACM SIGITE Newsletter, 3(1), 6–9. ing Machinery.
doi:10.1145/1113378.1113379
Hu, N., Pavlou, P., & Zhang, J. (2006). Can online
Davis, J., Farnham, S., & Jensen, C. (2002). reviews reveal a product’s true quality? Empirical
Finding others online: Reputation systems for findings and analytical modeling of online word-
social online spaces. Proceedings of the SIGCHI of-mouth communication. Proceedings of the
conference on Human factors in computing sys- 7th ACM conference on Electronic commerce.
tems: Changing our world, changing ourselves. Ann Arbor, MI: The Association of Computing
Minneapolis, MN: The Association of Computing Machinery.
Machinery.
Jhingran, A. (2006). Enterprise information mash-
Dutta, A., & Roy, R. (2006). Managing customer ups: Integrating information simply. Proceedings
service levels and sustainable growth: A model of the 32nd international conference on Very
for decision support. Proceedings of the 39th An- large data bases. Seoul, Korea: The Association
nual Hawaii International Conference on System of Computing Machinery.
Sciences. Kona, HI: Institute of Electrical and
Johnston, B., & Supra, J. (1997). Toward an inte-
Electronics Engineers, Inc.
grated approach to information services support.
Erickson, T., & Gratton, L. (2007). Eight Ways Proceedings of the 25th annual ACM SIGUCCS
to Build Collaborative Teams. Harvard Business conference on User services. Monterey, CA: The
Review, 10(7). Association of Computing Machinery.
Gibson, F., Teasley, S., & Yew, J. (2006). Learn- Kittl, C., Petrovic, O., & Teksten, R. (2001).
ing by tagging: group knowledge formation in a Developing Business Models for eBusiness. In-
self-organizing learning community. Proceedings ternational Conference on Electronic Commerce
of the 7th international conference on learning 2001. Vienna, Austria: International Center for
sciences. Bloomington, IA: The Association of Electronic Commerce.
Computing Machinery.
Lerner, R. (2006). At the forge: Creating mashups.
Gilmore, J., & Pine, J. (2001). The experience Linux Journal, 147, 10.
economy: Work is theater and every business a
McAfee, A. (2006). Enterprise 2.0: The dawn
stage. Boston: Harvard Business School Press.
of emergent collaboration. Sloan Management
Review, 47(3), 21–28.

216
Integrating Web 2.0 Technologies within the Enterprise

McNay, H. E. (2000). Corporate Intranets: Build- Smith, D., & Valdes, R. (2005). Web 2.0: Get
ing Communities with Data. IEEE Technology & ready for the next old thing. Gartner Research
Teamwork, 197-201. Paper. Stamford, CT.
Millard, D., & Ross, M. (2006). Blogs, wikis Timmers, P. (1998). Business models for elec-
& rss: Web 2.0: hypertext by any other name? tronic markets. Electronic Markets, 8(2), 3–8.
Proceedings of the seventeenth conference on doi:10.1080/10196789800000016
Hypertext and hypermedia. Odense, Denmark:
Weiss, A. (2005). The power of collective intel-
The Association of Computing Machinery.
ligence. netWorker, 9(3), 16-23.
O’Neill, M. (2005). Automated use of a wiki for
collaborative lecture notes. Proceedings of the
36th SIGCSE technical symposium on Computer
Key Terms
science education SIGCSE ‘05. St. Louis, MO:
The Association of Computing Machinery. Client-Support: Client-Support is a term used
O’Reilly, T. (2005). What Is Web 2.0: Design to describe the various efforts to ensure the success
patterns and business models for the next genera- of an environment. These efforts would include
tion of software. Retrieved July 17, 2006 from education, training, communities of practice,
http://www.oreillynet.com/pub/a/oreilly/tim/ online documentation and automated business
news/2005/09/30/what-is-web-20.html processes for procurement.
Collaboration: Collaboration is defined as
Palmer, J. (2002). Designing for Web site us- people working together on non-routine cognitive
ability. Computer, 35(7), 102–103. doi:10.1109/ work. This activity is about behavior, work habits,
MC.2002.1016906 culture, management, and business goals and value
Pereira, C., & Sousa, P. (2004). A method to de- generated we people from diverse backgrounds
fine an enterprise architecture using the Zachman come together.
framework. Proceedings of the 2004 ACM sym- Information Worker: The information worker
posium on Applied computing. Nicosia, Cyprus: is a label placed on individuals that primarily work
The Association of Computing Machinery. with information and data. Information workers
perform non-routine, cognitive, or creative work
Reinhold, S. (2006). Wikitrails: Augmenting wiki that often requires both structured and unstructured
structure for collaborative, interdisciplinary learn- information inputs from multiple sources.
ing. Proceedings of the 2006 international sym- RSS: In the simplest form, RSS shares the
posium on Wikis WikiSym ‘06. Odense, Denmark: metadata about the content without actually de-
The Association of Computing Machinery. livering the entire information source. An author
Riehle, D. (2006). How and why wikipedia works: might publish the title, description, publish date,
An interview with Angela Beesley, Elisabeth and copyrights to anyone that subscribes to the
Bauer, and Kizu Naoko. Proceedings of the 2006 feed. A feed reader application is required just as
international symposium on Wikis WikiSym ‚06. an e-mail client is required to read e-mail.
Odense, Denmark: The Association of Comput- Social Tagging: Social tagging describes the
ing Machinery. collaborative activity of marking shared online
content with keywords or tags as a way to or-
ganize content for future navigation, filtering,
or search.

217
Integrating Web 2.0 Technologies within the Enterprise

Weblog: A blog (short for weblog) is a per- well as from the consumer. Additionally, Web 2.0
sonal online journal that is frequently updated and embraces more of a thin client architecture which
intended for general public consumption. Blogs allows for the assembly of various components.
are a series of entries posted to a single page in Together, end user conent and thin client applica-
reverse-chronological order. These original entries tions make the Web 2.0 environment.
cannot be edit by others but can be commented Wiki: A wiki is software that allows users to
on by anyone. easily create, edit, and link pages together. Un-
Web 2.0: Web 2.0 is a term used to decribe like a blog, the end user can actually update the
the next generation of Web applications where original authors information.
information flows both from the producer as

This work was previously published in Handbook of Research on Electronic Collaboration and Organizational Synergy, edited
by J. Salmons; L. Wilson, pp. 560-577, copyright 2009 by Information Science Reference (an imprint of IGI Global).

218
219

Chapter 1.14
Security in a Web 2.0 World
Richard T. Barnes
SunGard Higher Education, USA

Abstract Introduction

Web 2.0 has brought enumerable benefits as well as With the multitude of benefits derived from the
daunting problems of securing transactions, com- various Web 2.0 technologies, it is unfortunate that
puters, and identities. Powerful hacker techniques, this book needs a chapter on security. Although the
including cross-site scripting (XSS) and cross-site collaborations, synergies, and transformations of
request forgery (CSRF), are used to exploit applica- the collective Web technologies (known as Web
tions to reveal and steal, at the worst, confidential 2.0) have immeasurably changed society in a good
information and money, or, at the least, cause way, there is a bad element that we must recognize,
trouble and waste time and money for reasons that understand, and defend against.
may be best described as fun or simply possible to The relatively open and participative nature
do. The people interested in transgressing Web 2.0 of Web 2.0 is, at once, a strength and weakness.
applications do so for money, prestige, or for the Opening sites to user content and comment creates
challenge. An infamous hacker from the early days synergies that would not exist had the sites been
of the Internet now heads his own Internet security restricted to a select few. However, it is difficult to
company. A more recent hacker of some infamy restrict user input to only positive discourse; vari-
has created a stir of concern and consternation as ous motivations compel some to poison this well
to how pervasive and potentially destructive hacker we know as Web 2.0.
attacks can be. Securing Web 2.0 applications re- Collectively, the responsibility and burden falls
quires a multifaceted approach involving improved on organizations and individuals to share in con-
code development standards, organizational policy straining the enablers to minimize the damage to our
changes, protected servers and workstations, and 2.0 Web sites. Although it is helpful to understand
aggressive law enforcement. the motivations behind the various (and growing)
attacks, it is more important to follow best practices
DOI: 10.4018/978-1-60566-122-3.ch005 in code development and security design (Evers,

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Security in a Web 2.0 World

2007). The adage “the best defense is a good of- Background


fense” does not apply well to Web 2.0 security. We
cannot proactively prosecute and punish someone It is perhaps ironic that the following definition
before they commit a Web attack; we may be on for application security comes from one of the
the road to a changing world, but constitutional best known wikis, Wikipedia. Application security
rights cannot be trampled upon. encompasses measures taken to prevent excep-
It is likely that some are dissuaded by the tions in the security policy of an application or
possibility of punishment if caught; but if only the underlying system through flaws in the design,
a few carry out Web attacks, our best approach, development, or deployment of the application.
still, is to mount our best defense. It is of course This definition is an excellent start in addressing
equally important to prosecute security offenses. a very large problem. However, it does not really
The threat of punishment has to be more than tell us why; that is, why is it necessary to prevent
theoretical: Offenders must know that if they are exceptions to security policy?
caught, there will be consequences. A broader definition may help. There are sev-
This chapter will explore the motivations, eral definitions of the word security: The freedom
methods, and defenses against the malicious be- from danger or the freedom from fear and anxi-
haviors that cost time and money, and lessen the ety are two variants that tell us why application
positives that can come from these technologies. security is so important to Web 2.0 applications.
There have been notable attacks to prominent Identity theft, corporate espionage or sabotage,
Web sites; a few of these will be examined for and/or simple maliciousness are certainly enough
their causes and associated effects. The evolution to give most of us some pause or anxiety. Applica-
of the World Wide Web into version 2.0 has had tion security, as it relates to Web 2.0, is now an
social impacts, too. What are these impacts, and area of great attention because of our collective
are there trends evident that may help us predict need to be free of these dangers.
where security attacks and defense strategies will A confluence of factors has complicated our
go in the future? Some possibilities are explored lives as Web 2.0 becomes a more significant
here and in subsequent chapters. presence. The graphics-rich functionality, col-
There is an old adage that says those who for- laboration, and opportunities have not only yielded
get the past are condemned to repeat it. This idea “serendipitous innovation” (Tapscott & Williams,
cannot be forgotten in Web 2.0 security. We must 2006), but less desirable consequences, too.
remember how attacks happened before so we can Consequences such as cross-site scripting
avoid similar attacks in the future. By examining (XSS) and cross-site request forgeries (CSRFs)
the trends, analyzing our mistakes, and understand- were not anticipated when foundational Web
ing our needs, we can improve on Web 2.0 and 2.0 technologies were created. Asynchronous
make it better. That is how we got to version 2.0 JavaScript and XML (extensible markup lan-
from 1.0. Perhaps, as the Web evolves into what guage), or AJAX, is a set of Web development
some in the community are calling Web 3.0, the techniques that enable Web sites to be interactive
lessons learned here will not be forgotten. and rich with features that make the static Web
pages of a few years ago seem, well, static. How-
ever, it is through AJAX and other technologies

220
Security in a Web 2.0 World

that Web attackers have created innovative ways essential enabler of the Web 2.0 era, AJAX serves
to, at the least, cause mischief, and at the worst, as the technical nexus for the development of the
cause severe harm. rich, interactive Web sites that we now take for
The subject of Web 2.0 security has a number granted. Capable of making direct communication
of important acronyms; AJAX is one. Others with the Web server, AJAX Web sites can request
include XML, RSS (really simple syndication), server data without reloading the Web page. These
and SOAP (simple object access protocol; Shah, behind-the-scenes data exchanges occur without
2006). This chapter is not necessarily intended the user’s knowledge and provide a transparency
to define and explain these terms, but more to to applications that make them distinctly more
put them in context with the larger problem (and advanced than their Web 1.0 predecessors.
challenge) of Web 2.0 security. Much of the recent This direct communication between the Web
literature talks of where the burden lies: with the server and Web application also provides an
developers, with the companies, and with the opening for application security transgressions
users. The reality, however, is that at least some (Enright, 2007). One method of attack, known
of the burden falls on each of us. as XSS, involves the injection of malicious code
Developers can build security into the applica- into a Web page viewed by others. The silent (i.e.,
tions they write instead of adding security later as hidden) execution of the AJAX code is perhaps
an afterthought. Companies can elevate security one of the biggest areas of concern for IT security
to an enterprise-level initiative and build it into experts. Because transactions are occurring in the
the products they sell. Although companies have background without the user’s knowledge and
been victims of security incursions, especially input, the potential for the execution of malicious
of late, the user has often been victimized, and code running unimpeded is large. This code may
it is often at the user level where good security run on the computers of unsuspecting Web site
hygiene is easiest and best applied. visitors, exposing the visitors to the possible theft
A comprehensive approach to securing Web of sensitive information (e.g., banking informa-
2.0 applications stands the greatest chance for tion, authentication credentials, etc.). There are
success. It is not enough to say that developers several types of XSS attack methods; however,
need to do a better job or companies cannot rush the common denominator in the end result is that
products to market when everyone can contribute the user’s browser always executes code that is,
to better security. How ironic that one of the great at least, not authorized, and at worst, destructive
strengths of Web 2.0, collaboration, can, in effect, or compromising (McMillan, 2007).
help solve one of its greatest weaknesses! Attacks using the CSRF method have been less
common, but a large unrealized potential exists for
more attacks (and more harm) in the future. The
Security Problems cross-site in CSRF is derived from the XSS method
and Weaknesses explained above and denotes some of the simi-
larities between the two methods. Where the two
AJAX in this context is not a cleanser made famous methods are most different can be denoted in the
by a television jingle, nor is it a reference to the forgery part of the name. According to Merriam-
Greek who out of madness quite literally fell on Webster Online, a forgery is defined as an imitation
his sword. AJAX in this context is a programming passed off as genuine. This is quite literally what
technique that employs a few other successful the CSRF attack method does: Unauthorized (or
technologies including, as the acronym implies, forged) commands from a user are transmitted
XML and JavaScript (O’Reilly, 2005). Seen as an to or from a trusted Web site. The reliance and

221
Security in a Web 2.0 World

trust on an established (authenticated) identity is If a node is compromised, a SOAP message en


how the Web has worked, and the CSRF method route can be intercepted and modified as part of
exploits this common characteristic by imitating an attack. Although there are several variations on
the authenticated user (Waters, 2007). the Web services theme, these types of attacks are
XML poisoning is another method of attack. less prevalent than the JavaScript-based attacks.
Web 2.0 applications transmit XML data between
the client and server as part of their normal op-
erations. Although poisoning may sound a bit Who are the Attackers, and
overstated, the method will corrupt (or poison) What are their Motivations?
the XML data in such a way as to disrupt the
processing of the information. The effect of this Just as Web applications have evolved from the
poisoning can range from denial of service (DoS), static, even boring presentation of webmaster-
where the targeted server is bombarded with spuri- defined content, so too have the people who stage
ous requests that will, in effect, bring the server attacks on the Web. Because evolution may imply
down or render it nonfunctional, to compromised betterment, perhaps it is incorrect to say that at-
confidential information. tackers have evolved; what has really happened
RSS is a method and an XML 1.0-compliant is that their methods have become more sophisti-
format that aggregates disparate sources of Web cated as the Web infrastructure has become more
information into a feed for the user. The user sub- complex. But, who are the attackers and why do
scribes to the feed and sets the criteria for what they do what they do?
information should be included (or excluded) in There are no clear definitions or labels for
the feed. While this may sound like an incred- these Web attackers. While the word hacker has
ible time saver for the user, what it also can do been generally used in a pejorative sense by the
is expose the user to malicious code embedded media, there are other words that may be more
in the source information. This code may also be precise. The word cracker, also known as a black
aggregated along with the legitimate information, hat, refers to a person who illegally compromises
and the user will not necessarily know anything is the security of a computer system or network to
amiss. This RSS injection method, as it is known, reach a malicious end. A white hat will (attempt
can install and launch software on the user’s to) compromise security, but will also have valid
computer, potentially compromising confidential permission from the system or network owners;
information. the objective, of course, is to locate security holes
Web services can be a fertile ground for Web 2.0 and plug them before they are exploited. Not to
security intrusions. SOAP is the common protocol imply a standard curriculum has been established,
used for Web services remote procedure calls, but to acquire the necessary skills to be a cracker,
and is therefore an often-used method for attack. one should first be a hacker (Walker, 2005). The
Parameter manipulation and node exploitation usage of the word hacker has changed over the
are common techniques used in Web-services- years. Its etymology has mostly had a creative
related attacks. In parameter manipulation, the connotation. However, since the movie Wargames
variables that are passed in services calls can be in 1983, the general (public) meaning has been
manipulated to suit the attacker’s needs. If a site to creatively manipulate computer code or pro-
has insufficient validation of received parameters, cedures to achieve a desired outcome, generally
the site is open to attack and compromise. Nodes a mischievous or malicious one. For purposes
on the Internet cannot be assumed to be secure: of brevity, the word hacker will be used in this

222
Security in a Web 2.0 World

chapter not to be imprecise, but more specifically Like Mitnick who preceded him, but perhaps
to be consistent with the general connotation of to a lesser degree, Samy may have gained more
the word. than he has lost from his exploit. Although Samy
The rebellious, antiauthority teenager may still Kamkar was sentenced to probation, community
exist, but the harm that erstwhile person can do service, and a certain amount of restitution, his
is now less than before, mostly thanks to much reputation has been indelibly etched into Web
higher numbers of installed security software. history. One online news source even referred to
While it is certainly not ubiquitous, the awareness his exploit as “cunning” (Orlowski, 2005).
of computer viruses and worms is higher than The motivations behind Web 2.0 attacks have
what it was, therefore leading to a greater number changed. While it may have been a prank or an
of protected computers. There are exceptions, of adventure a few years ago, the development of
course, but the days are mostly passed when a new Web exploits is now the stuff of nationalism,
lone and unassisted teenager can cause significant social castes, and real money. A denial-of-service
harm (without a significant response) from his attack that hit Estonia in early 2007 is suspected
basement computer. However, the problem will to be rooted in Russian nationalism, although
never go away. As new, more complex applica- some security experts dispute the nationalism
tions are deployed, new exploitable vulnerabilities claim (Brenner, 2007), saying it was motivated
are found. by anger over a government decision to move a
Perhaps one of the earliest and best known revered monument. Despite the lingering issue
hackers who had the mind-set, skills, and intel- of why the Estonian attack happened, Russia is
ligence for finding vulnerabilities is now out of thought to be a proverbial hotbed of malware
jail. Having satisfied all penalties levied against (malicious software) engineering (Brenner). Other
him, he now has his own security consulting com- than the booming petro economy, the Russian
pany and has published more than a few books. “other” economy is bleak, and programmers not
According to a recent interview, Kevin Mitnick’s only see malware as a money opportunity, they
original motivation was “fun and entertainment.” also do not believe developing malware is wrong.
There was a thrill to gaining access to confidential Still, the nationalist bent cannot be discounted; a
or proprietary information (Brandon, 2007). small Russian newspaper reportedly praised the
A hacker of more recent fame, the creator of local perpetrators of a hacking attack for their
the infamous Samy worm that forced MySpace accomplishment at the capitalists’ expense (Cla-
off line after an XSS attack, created it as a prank burn, 2007). With a report that a Russian Web
according to an interview shortly after the incident. server was found to be hosting approximately 400
Despite the greater prevalence of security software, malware applications (Brenner), and a report that
the Samy worm raised awareness to the increased China’s production of malware exceeds Russia,
risks associated with applications under the Web how long will it be before nationalist Web attacks
2.0 umbrella. When it became known that Samy’s are directed to the Western hemisphere?
very first AJAX project created the Samy worm, In the West, a social order has developed among
business and security leaders took note; if this is the people who develop and/or deploy malware
his first, what could be next? It may seem a bit on the Web (Claburn, 2007). Even though Kevin
perverse that Samy, the perpetrator of the attack, Mitnick has a legitimate career now after paying
did not fault himself or MySpace for the worm’s his dues to society, he is held in reverence and awe
success, but instead blamed the client browsers by young and aspiring hackers. Samy Kamkar is
for making it easy to run unauthorized JavaScript substantially younger, but his “Samy is my hero”
code on them (Lenssen, 2005). prank on MySpace is still admired by people in

223
Security in a Web 2.0 World

their teens, 20s, and even 30s. Even though they that look legitimate, but are in fact facades devel-
are legitimate now because of their admiring fans, oped to acquire credit card numbers, passwords,
both Mitnick and Kamkar are certainly near the social security numbers, and other personal data,
high end of the hacker social order. phishing is proving to be a lucrative criminal
The lure of an income, perhaps a large income, method of cyber attack. Security firms such as
is a major motivator in both the West and in Asia. Symantec have included phishing protection in
No longer is malware just written by and for the their software, and Microsoft has phishing filter
exclusive use of its writer. Malware kits, that is, functionality included in Internet Explorer 7 to
kits for developing malware, are readily available help minimize the dangers phishing can pose to
for sale on the Internet. Originators of malware the unwitting user.
programs or kits are at or near the top of the hacker
social order. These are the people who stand to
earn the greatest amount of money from their ac- What Can We Do?
tions, and the money can be considerable. Lower
in the hacker social order are the people who buy What can be done to mitigate the security risks
these kits to develop their own malware, but that and minimize the problems that have evolved
does not mean they are of lesser significance; from Web 2.0? We cannot eliminate the problem
the availability and the use of malware kits are a of cyber attacks on Web 2.0 applications or other-
significant factor in the proliferation of malware wise. The best we can hope for is to minimize the
on the Internet. Though the road can be tough probability of an attack, or if attacked, minimize
(Kevin Mitnick was arrested several times), the the damage done.
hope of a lucrative hacking career and the high The field is a rapidly changing one, where
status it may bring can be compelling. hackers find a weakness, companies or individu-
Malware kits are one source of income. als respond, and society slowly incorporates and
Another money-earning method involves rent- adapts to the changes over the long term. During
ing botnets. Botnets are networks (or groups) the 1980s, few could have imagined what the In-
of security-compromised computers known as ternet in general and Web 2.0 in particular could
zombies. Botnets are also used in spamming (mass have become in 2007, but what was unimagined
junk e-mail distributions), focused attacks to steal then is reality now: Society will continue to adapt.
data, and denial-of-service attacks as client-naïve One constant, though, is that hackers will exploit
instruments. They have even been used to extort a weakness, and the rest of us will respond. So-
money under threat of a DoS attack. ciety’s innovation will always be challenged by
Depending on the size of the botnet, that is, the hacker’s innovation.
how many zombie computers are part of the botnet, Many blame the rush-to-market mentality of
the weekly rental income can be several hundred companies in getting their Web 2.0 applications
dollars. More income can be earned with the sale up and running. In their haste to meet deadlines,
of an unpublished security vulnerability, perhaps developers forego what few best practices there are
up to US$1,000 or more (Evers, 2007). What is to complete their tasks. There are two things that
less known, but feared more, are the markets for can be done to address this part of the problem.
identity-related data such as credit card numbers, In 1996, Alan Greenspan, the Federal Reserve
social security numbers, and company-related con- chairman, used the term “irrational exuberance”
fidential information to be used in espionage. to characterize the overinflated values of the stock
Phishing scams are relatively new, but their market; perhaps companies venturing into Web 2.0
impact cannot be understated. With Web sites territory need to tamp down their exuberance to a

224
Security in a Web 2.0 World

level where their developers can properly include data should be guarded as closely as company
security into their Web 2.0 applications. Secondly, data. Home computers can be current on the lat-
Web security is a rapidly changing field; to keep up est security software just as work computers are
with the changes, developers should be continually (usually). Individuals should recognize the public
trained on threats, techniques, and best practices. nature of the Internet and the Web, and realize that
As mentioned before, there is not one answer to the public does not mean safe. Web 2.0 has enabled
security problem, but as companies work toward the Web to become more of a conversation than a
a whole solution, they should concentrate on two billboard; that conversation will have questionable
of the biggest threat sources: cross-site scripting and nefarious participants, and individuals have
and cross-site request forgeries. to remember that.
There is more that companies can do to address
Web security issues. Widely acknowledged as a
problem, company confidential or proprietary data What is on the Horizon
can appear in Web 2.0 weblogs (blogs) that may for Web Security?
eventually become known to the company’s com-
petitors. This is not caused by hackers, but more As suggested in the previous section, companies
likely company employees, disgruntled or simply and individuals are reacting to the Web security
naïve. Companies need to develop (better) data issues that have been exacerbated with Web 2.0
policies: Their data need to be identified, managed, technologies. Some believe there really is one
and controlled (Vijayan, 2007). Knowing what magic solution, while others embrace the idea
data the company has and where they are located that a comprehensive strategy and solution is the
is an essential first step. Also essential to know is best approach.
who has access to data; that is, companies should One notable and positive trend is the telltale
know who should have access and put controls in movement to thin-client applications and hard-
place to enforce company policy on data. ware. Thin clients are essentially input and output
A more problematic approach is determining devices only; the applications, heavy processing,
if and how to limit employee access to blogs. and data reside on a server. With the advent of fast
Whether deliberate or accidental, when an em- corporate and broadband networks, thin-client
ployee puts company data on a blog, they become computing is more possible than ever. In 1995,
public and potentially compromise the position Oracle founder Larry Ellison presaged a future
of the company. An outright ban on company where network computers replaced the personal
computers is certainly possible. Web sites can be computer (Bock, 2006). Though his reasoning
blocked, firewall rules can be stringently applied, was based on the growing complexity of personal
and so forth (Fanning, 2007), but how can a ban computers during the 1990s and not the security
be enforced on noncompany computers? The short concerns raised in the 2000s, his prediction was
answer is it cannot, so the likely answer for com- nevertheless prophetic. Microsoft now has in beta a
panies is to take a middle position; develop, apply, server-based version of Office that will further the
and enforce a comprehensive data policy that can trend of thin-client applications. With computing
extend beyond the company’s parking lot. power and data moving to more protected Web
The onus should not be completely on compa- servers, opportunities for intrusion are lessened
nies to address Web security problems. Organiza- and malware attacks will be reduced.
tions are generally more strident about security, The best practices for Web security are devel-
and individuals should learn from their employers oping and becoming better known. When Web 2.0
and become more security conscious. Personal applications began to appear, the best practices

225
Security in a Web 2.0 World

had not caught up to the technology, but that is and profit. However, the market has responded
changing now. Control Objectives for Informa- and will continue to respond.
tion and Related Technology (COBIT) is both a In the context of Web security, terms such
format and forum for governance of all things IT. as AJAX and SOAP may never become well-
As practices become established and accepted known outside IT circles, however the impact
by the IT community, COBIT-codified practices they have (and will) have on all of us cannot be
are updated and disseminated through traditional understated.
channels such as conferences and published This is certainly not to suggest or imply support
standards. Web 2.0, podcasts, blogs, and wikis, of hackers, their social order, or the nationalistic
though perhaps nontraditional channels now, are motivations behind Web security incursions and
becoming evermore important in communicating attacks, but what is the underlying net effect of
best practices to the people and organizations that hackers, nationalists, IT, and security companies?
need them. Is the net effect positive, negative, or neutral? Be-
There may be a developing trend with negative sides the obvious inconveniences and annoyances,
consequences for Web security, at least in the short the negative side includes companies sabotaged,
term. An increasing number of acquisitions of Web extorted, and coerced out of uncounted sums of
security specialty firms have some security experts money. IT departments have had to divert resources
concerned. With IBM’s acquisition of Watchfire to the security problem in order to plug the leaks
and HP’s recent acquisition of SPI Dynamics, will known and identify the holes not yet known, and
the products of those companies be less available individuals have had identities stolen and bank
to the general market? These smaller companies, accounts drained.
before their acquisitions, were autonomous leaders Is there a plus side? Perhaps yes—more
in Web 2.0 security. However, with their indepen- companies have been created that specialize in
dence gone, it is unclear whether their products Web security, IT department budgets have been
will be directly available to the market at large, or increased to augment their security staffs and tools,
limited as offerings from the acquiring companies. and while of questionable value, some (formerly
This should be a short-term problem because their unemployed) people in Asia and elsewhere are
technology, and technology from other similarly now working to help meet the burgeoning demand
acquired companies, will eventually make its way for malware.
to the market (Germain, 2007). From a dispassionate viewpoint, one might
argue that the plusses outweigh the minuses; that
is, the response to Web security problems has
Conclusion been more beneficial to the many than harmful
to the few. A victim of identify theft or the CEO
Where are we now with Web security? This chap- of a company that was impacted by an attack or
ter has examined the background and evolution extortion will surely disagree.
of where we were. It is evident now that with the It should be clear that there will always be
advent of Web 2.0, the security technology for concerns over Web security. A certain trend is that
Web applications lagged behind the technology knowledge of the security threats arising from
for the social software embodied in Web 2.0. Web 2.0 is spreading, and users, developers, and
This gap in technologies created an opportunity companies are responding with education, new
for makers of mischief, maliciousness, espionage, products, and new strategies to mitigate the risks.

226
Security in a Web 2.0 World

In this Web 2.0 world of podcasts, blogs, and wikis, Evers, J. (2007b). The security risk in Web 2.0.
there is an unfortunate but real repeating process Retrieved July 17, 2007, from http://news.com.
that helps keep IT security managers employed. com/2102-1002_3-6099228.html
As security awareness increases, knowledge in-
Fanning, E. (2007). Editor’s note: Security for Web
creases; as knowledge increases, the potential for
2.0. Retrieved July 17, 2007, from http://www.
new security threats increases. Although security
computerworld.com/action/article=283283
is fleeting, the field of IT security is forever.
Germain, J. M. (2007). IT security and the no good,
very bad Web app nightmare. Retrieved December
References 24, 2007, from http://www.technewsworld.com/
story/60208.html
Bock, W. (2006). Larry Ellison and the network
computer that wasn’t. Retrieved December Lenssen, P. (2005). Samy, their hero: Inter-
30, 2007, from http://www.mondaymemo. view. Retrieved December 29, 2007, from
net/031103feature.htm http:blogoscoped.com/archive/2005-10-14-n81.
html
Brandon, J. (2007). Q&A with the forefa-
ther of hacking. Retrieved December 24, McMillan, R. (2007). Researchers: Web 2.0 se-
2007, from http://www.pcmag.com/print_ curity seriously flawed. Retrieved December 24,
article2/0,1217,a=213916,00.asp 2007, from http://www.pcworld.com/printable/
article/id,131215
Brenner, B. (2007). Black hat 2007: Estonian
attacks were a cyber riot, not warfare. Retrieved O’Reilly, T. (2005). What is Web 2.0: Design
December 30, 2007, from http://searchsecurity. patterns and business models for the next gen-
techtarget.com/originalContent/0,289142,sid14_ eration of software. Retrieved July 1, 2007, from
gci1266728,00.html http:www.oreilly.net.com/lpt/a/6228

Brenner, B. (2007). How Russia became a mal- Orlowski, A. (2005). Web 2.0 worm downs MyS-
ware hornet’s next. Retrieved December 30, pace. Retrieved July 17, 2007, from http://www.
2007, from http://searchsecurity.techtarget.com/ theregister.co.uk/2005/10/17/web20_worm_
originalContent/0,289142,sid14_gci1275987,00. knocks_out_myspaces/print.html
html Shah, S. (2006). Top 10 Web 2.0 attack vectors.
Claburn, T. (2007). Hacker profile becomes more Retrieved July 17, 2007, from http://www.net-
social, adds women. Retrieved December 24, 2007, security.org/article.php?id=949&p=1
from http://www.informationweek.com/shared/ Tapscott, D., & Williams, A. (2006). Wikinomics:
printableArticle.jhtml?articleID=205101618 How mass collaboration changes everything. New
Enright, G. (2007). Web 2.0 vulnerabilities to watch York: Portfolio.
for. Retrieved July 22, 2007, from http://www. Vijayan, J. (2007). Six ways to stop data leaks.
computerworld.com/action/article=9027342 Retrieved July 17, 2007, from http://computer-
Evers, J. (2007a). Hacking for dollars. Retrieved world.com/action/article=285138
December 24, 2007, from http://www.news.
com/2102-7349_3-5772238.html

227
Security in a Web 2.0 World

Walker, A. (2005). How and why hackers want Waters, J. K. (2007). Web 2.0 entails “sleep-
to get inside your machine. Retrieved December ing giant” security risk. Retrieved December
24, 2007, from http://www.informit.com/articles/ 24, 2007, from http://www.adtmag.com/print.
printerfriendly.aspx?p=425380 aspx?id=21499

This work was previously published in Social Software and Web 2.0 Technology Trends, edited by P. Deans, pp. 58-67, copyright
2009 by Information Science Reference (an imprint of IGI Global).

228
229

Chapter 1.15
Web Site Localization Practices:
Some Insights into the Localization
Industry

Nitish Singh
Saint Louis University, USA

John E. Spillan
University of North Carolina, USA

Joseph P. Little
Saint Louis University, USA

Abstract future. [Article copies are available for purchase


from InfoSci-on-Demand.com]
The e-commerce industry has experienced spec-
tacular growth, change and development. This
situation has initiated an enormous business Introduction
revolution that has affected the process of global-
ization tremendously. The goal of this study was From the mid-1990s to the present day, the e-
to analyze the Web sites of localization companies commerce industry has experienced spectacular
that provide localization and translation services growth, change and development. The global
to other companies and see if they themselves are online population is estimated to reach 1.8 billion
practicing what they are preaching. The results by 2012 (Jupiter Research, 2008). In 2008 North
suggest that localization companies are indeed America accounted for only 17.5% of the online
not practicing what they are preaching. Analysis population and that percentage is in decline as
shows that localization company Web sites are countries such as China, Brazil, India, and Russia
less localized than the Web sites of their clients, show the highest level of online population growth
the multinational companies. The findings provide (Internet World Stats, 2008). This situation has
some implications to domestic and international initiated an enormous business revolution that has
marketers who currently operate in or are plan- affected the process of globalization tremendously
ning to enter into the global markets in the near (Cyr & Lew, 2003). During the past several years

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Web Site Localization Practices

an entire industry (the localization industry) has • Modifying graphics, data fields, tables,
grown up around helping companies design mul- forms, layout, colors and tables etc.
tilingual Web sites and software applications for • Modifying the cultural content of the site
different countries. An industry report estimates or software to be congruent to the local
the size of the worldwide translation and localiza- culture.
tion services market at US$ 8.8 billion (DePalma • International e-commerce readiness for
& Beninatto, 2006). According to this report the multi-country transactions.
commercial market for localization services is • Web site navigational modifications to meet
estimated at US$6 billion and the government local preferences.
market at US$ 2.8 billion worldwide.
Localization is the process of adapting products This study analyzes the Web sites of localiza-
and services (Web sites, manuals and software) in tion service providers in order to understand to
accordance to linguistic, cultural, technical and what extent these companies are translating and
other locale-specific requirements of the target localizing their own sites, and modifying graph-
market (Localization Program at California State ics, layout, colors, text, policies, navigation, and
University, Chico, College of Business, 2008). cultural content. The study then compares the
Localization is now being seen by multination- localization efforts of the localization vendor
als as a necessary process to develop multilin- company with their clients, who are generally
gual and multicultural content to effectively tap multinational companies. This analysis will reveal
global markets. Forester research estimates put if the companies that are preaching localization
the 2006 global e-commerce revenues at around are also effectively implementing it on their own
$12.6 trillion. Furthermore, research has shown Web sites. After all, the localization service pro-
that consumers prefer Web sites in their native viders should set the benchmarks for their clients
language and Web sites that reflect their local to follow. Moreover, the companies that will be
preferences (Singh, Furrer, & Ostinelli, 2004). the winners in this fast growing and consolidat-
Thus, companies around the world are creating ing localization industry will be the ones that are
multilingual Web sites to tap this vibrant online able to highlight and exemplify the need for Web
market. Companies like IBM, Oracle, Intel and site localization. What is a localization service
other have almost 90 international sites to take provider telling their current clients along with
advantage of the global online markets and com- potential future clients if their own Web sites are
municate with their global customers. This surge not sufficiently localized? The goal of this article is
in creating multilingual online content and soft- to gain understanding of the localization processes
ware has also led to the growth of the localization used in the localization industry and to focus on
industry which is helping these companies by the current trends in the localization industry.
effectively translating their Web sites, user inter- This article is composed of seven sections. The
faces, software, and manuals. Beyond translation first section presents and introduces a review of the
the localization companies are also involved in globalization and localization literature. Section
the following (Esselink, 2000): two explains the research methodology. Section
three provides a presentation of the analysis.
• Making visual or graphics, technical and Section four provides a discussion of the results,
textual modifications to the site content. section five describes some managerial implica-
• Rewriting the text, translating the text, and tions, section six explains some limitations along
ensuring translation, idiomatic, and concep- with future research ideas, and finally section
tual equivalence of the translated text. seven provides a conclusion.

230
Web Site Localization Practices

Literature Review their use of rhetorical style and use of metaphors.


Therefore, a simple machine translation, without
Globalization and the an understanding of a culture and its language,
Localization Industry may result in a cultural faux pas (Singh et al.,
2004). Beyond linguistics companies also need
Increased ownership of computers and Internet to consider such culturally sensitive areas such as
usage is growing every day. Throughout the persuasion techniques, colors, icons, signs, Web
world the Internet is rapidly becoming the main page layout, and cultural values when localizing
source for information, shopping and services. Web sites (Singh & Pereira, 2005).
Furthermore, computer and Internet users are in- Previous research has shown that culturally
creasingly from non-English speaking countries. sensitive Web content enhances the site’s usability
One estimate indicates that 32% of Internet users (Fock, 2000; Luna, Peracchio, & de Juan, 2002;
are non-native English speakers. This number is Singh & Pereira, 2005; Simon, 2001). So, in or-
increasing. The result of this huge Internet expan- der to effectively communicate to foreign online
sion motivated businesses to recognize the value consumers it is beneficial for a firm to adapt their
of Website localization (Kwintessential, 2009). Web sites to the targeted market. Furthermore,
Moreover, the unending process of globaliza- research has shown that not only does Web site
tion is fundamentally altering the manner in localization enhance usability but also attitude
which enterprises do business. When businesses towards the Web site, perception of the ease of
globalize their e-business, there is a great need site navigation, and ultimately purchase intention
recognize that language, cultural expectations (Singh et al., 2004).
and trust play a huge role when building online The cultural impact is substantial and Hall
Web capabilities (Culnan & Armstrong, 1999; (1976) believes that it is very difficult to act or
Jarvenpaa, Tractinsky, & Vitale, 2000; Singh & interact in any meaningful way if they do not un-
Pereira, 2005; Violino, 2001). For an organization derstand language and culture. The consequence
to be successful in this demanding setting, they of not including language and culture, when
must adjust their offerings so that their products considering global Web presence, is discarding
and services present the appearance and feel of profitable global online consumer. The Internet,
being produced locally. The process of localization similar to any other advertising document, is a
begins with an understanding of a wide range of replica of the culture of the country or locale (Cyr
linguistic, cultural, content, and technical issues. & Trevor-Smith, 2004; Hermeking, 2005; Singh
A product or service presentation has to be tailored & Matsuo, 2004). According to Mooij (1998)
to the local customs and practices of a country or advertising mimics a society’s values. It can only
region. For example, producing a Web site in only be effective when it is inseparably connected to
the English language is not sufficient because the the primary culture of the group for which it is
majority of the world does not understand English. targeted. Studies have demonstrated that advertis-
Furthermore, even if consumers do understand ing that is harmonious with local cultural values
English research has shown that they prefer Web is significantly more compelling than standard-
sites in their native language (Singh et al., 2004). ized advertising. Several researchers, therefore,
While many companies use machine translations have emphasized the use of country-specific
to adapt the language used on their Web sites cultural values appeal when developing interna-
(Singh & Boughton, 2002), this type of transla- tional advertising campaigns and communication
tion is not sufficient either. Languages not only material (Albers-Miller & Gelb, 1996; Han &
differ in their use of characters or syntax, but also Shavitt, 1994). Research indicates that Web site

231
Web Site Localization Practices

localization and cultural customization promotes The differences in cultures require interna-
a better opinion regarding the site ultimately tional businesses to find ways to make their Web
influencing people’s purchase intentions (Singh sites communicate with different cultures in dif-
& Pereira, 2005). Luna et al. (2002) discovered ferent parts of the world. As such, the key acronym
that culturally harmonious Web content creates that has emerged in this new arena of business
a more user friendly environment where the user operation is GILT or Globalization, Internation-
has clear instructions and comes away from the alization, Localization and Translation (Lommel,
Website with a better attitude about the content 2003). Globalization addresses the enterprise
that is presented. Consequently, the localization issues associated with making a company truly
of Web sites also necessitates culturally tailoring global. So, for products and services this means
the Web sites to be congruent with the cultural integrating the internal and external business
requirements of the local environment. Miscom- functions with marketing, sales, and customer
munications, in the international context, gener- support in the world market (The Localization
ally take place when the message is seemingly Industry Standards Association, 2008). More
mismatched with the local culture and does not specifically, Web site globalization includes two
produce the response that was expected towards complementary processes: Internationalization
the businesses products or services. The foreign and Localization.
language, signs and symbols, and Web content Internationalization is the process of gener-
that is culturally different, creates confusion, alizing a product so that it can handle multiple
frustration, offensiveness and in the long run a languages and cultural conventions without the
loss of business (Luna et al., 2002). need for redesign. In more technical terms, it is the
While company Web sites provide a major op- process through which back-end technologies are
portunity to impart and promote a corporate image used to create modular, extendible, and accessible
and to sell products and services, the effectiveness global Web site templates that support front-end
of the Website depends almost exclusively on the customization (Singh & Boughton, 2005). This
value of its content (Pollach, 2005). An effective process enables company Web sites to be locally
Web site is the one where the consumers invest responsive to the end-user through front-end
a considerable amount of time reviewing the customization. Internationalization takes place at
content of interest, requesting more information, the level of program design and Web document
and buying the goods or services offered (Liu, development (Singh & Little, 2009).
Marchewka, & Ku, 2004). As such, the quality Localization and translation is the process
and value of a Web site will be influenced by how of adjusting a product or service and making
the Web site mirrors the culture of the nation for it linguistically and culturally appropriate to
which it has been designed (Fletcher, 2006; Singh the target locale. More specific to the current
& Pereira, 2005). study, Web site localization is the process of the
Culture influences just about everything we front-end customization, whereby Web sites are
do, say, read, hear and think. Web sites are not adapted to meet the needs of an international
immune to the affect of culture (Kwintessential, target market (Singh & Boughton, 2005; Singh
2009). Those companies that are able to develop, & Little, 2009).
manage and customize their business Web sites The localization industry can trace back its
to the culture of the country they are doing busi- roots to early 1980’s when the software industry
ness, will generate more interest in their company was emerging as an upcoming sector of the US
and ultimately increase the sales of their products economy, and felt a need to translate software
and services. products in multiple languages (Globalization

232
Web Site Localization Practices

Industry Primer (LISA), 2007). As the applica- and technology integration from corporate con-
tion of software grew across a cross-section of solidations is not a recipe for long term growth
industries and with the growth of the Internet, the and sustainability of this industry. Lee (2005)
localization industry also saw sustained growth. recommends that for the localization industry
Now the localization industry is seeing a growth to be viable it needs to expand its role from just
phase with the need for translation and localiza- a translation or technology-solution provider to
tion of software, manuals, packaging, and most truly embracing the wider concept of localization
importantly multilingual Web sites. The industry by providing international marketing expertise.
is also undergoing considerable consolidation. This international marketing orientation will then
During the 1990’s the trend toward industry help the localization companies to expand their
consolidation started with small vendors joining offerings by delivering localized information and
hands to offer “one-stop shopping” for large soft- comprehensive resident knowledge in designing
ware developers like Microsoft, Oracle, and IBM marketing campaigns for geopolitical and ethno-
who needed translation and localization services graphic regions in areas of print advertisement,
in multiple languages (Cyr & Lew, 2003). The online brand valuation, and Website usability
industry consolidation leads to the emergence of analysis (Lee, 2005).
multi-language vendors (MLVs) which special- The next sections of this article will detail the
ized in completing multi-language, multi-service methodology, sample, and the analyses used in this
localization/translation projects. These MLVs also study. Additionally, insights into the current level
used an outsourcing model where they outsourced of localization practiced by localization vendors
the core translation services to single-language and their multinational clients are presented.
vendors (SLVs); Single Language Vendors nor-
mally work into one target language only, from
one or more source languages (Esselink, 2000). Methodology
The acquisition of Bowne Global Solutions by
Lionbridge Technologies in 2005 lead to the To analyze the quality and extent of localization
emergence of Lionbridge as one of the largest depicted on the localization vendor Web sites,
Globalization and off-shoring companies in the the study conducted a content analysis of the
industry. Similarly, SDL International, another vendor Web sites and Web sites of multinational
major player in the localization industry, enhanced companies. More specifically, content analysis
its portfolio by acquiring Trados Inc., which was methodology was used and a coding system was
a major translation technology solution provider. developed to measure various facets of the local-
As the localization industry grows, and serves ization efforts. The coding system used in this
new and bigger clients across a cross-section study was adapted from Singh, Toy and Wright
of industries, it will need localization vendor (2009). The coding sheet included items like:
companies to invest in process and product in-
novations and R&D to be competitive. In order • Ease of finding global gateway on the Web
for these large investments and comprehensive site
solutions to be provided, vendors will need to • Use of country code domain names of
be backed by substantial capital investments. ccTLD
However, Leon Z. Lee (2005), an industry ex- • Translation depth
pert, warns that the current focus of large and • Local customer support
small localization companies toward primarily • E-commerce information and policies
cost leadership, automated enterprise workflows, • Navigational outlay

233
Web Site Localization Practices

• Web site page structure/layout ternational Web sites and Web sites specifically
• Use of Locale-specific graphics, colors and for both Germany and Spain. Thus, a total of
values. 100 multinationals with 300 country sites (U.S.
English, Germany and Spain) were analyzed for
To perform the content analysis two coders this study.
were trained in the coding scheme and jointly
coded several Web sites. The inter-coder reliability
on the sample of vendor and multinational client Results and Analysis
Web sites ranged from .82 to .86. It is suggested
that inter-coder reliability needs to be above .80 Number Unique Languages
in order to be acceptable (Grant & Davis, 1997). Supported
Thus, the coder reliability exceeded the suggested
threshold. The purpose of finding the number of unique lan-
guages was to understand how many languages is
the company providing its services in. The results
Sample show that on average a vendor company site had
about 7 unique languages depicted. On the other
Analyzing all country sites (which can range from hand the Vendor clients, such as Multinational
10-90 and may include more than 2000 pages) company Web sites had on average 19 unique
to measure localization efforts was beyond the languages supported. This shows that Localiza-
scope of this study. Thus, the study measured the tion vendors are far behind their clients in terms
localization efforts on the German and Spanish of languages supported on the site (see Table 1).
Web sites of each vendor company. German and In fact the mean number of languages depicted
Spanish have been forecasted as some of the top by multinational Web sites (19.38) exceeded the
languages in which multinationals are localizing maximum depicted (16) by the vendor sites. An
their sites. independent sample t-test indicates the means are
To find a sample of localization vendor significantly different (F = 37.708, p = .000).
company Web sites the study used the vendor
company data base provided at the Globalization Ease of Finding Global Gateway
and Localization Association Web site. In total
the study was able to include only 53 localiza- The aim here is measure how visible the link for
tion vendor company Web sites in the sample, as international sites is from the company’s U.S.
these were the only companies we found having English home page. Based on the comparison
international Web sites for Germany and Spain. data between the vendor and client Web site, it
Thus, 53 companies and their German and Spanish seems 30 percent of client (Multinational) sites
sites served as the final sample, which included have a dedicated global gateway page compared to
almost 106 Web sites and more than a thousand which only about 9 percent of vendor sites have a
Web pages. The study also analyzed the company dedicated gateway page (see Table 2). The results
home site (mostly in English) to see the structure of a chi-square test for two independent samples
of the global gateway, Web page structure, and indicates there is a significant difference in the
Web content depth and navigation. The sample presentation of a global gateway page between
of multinational company Web sites was selected vendor and multinational sites (χ2 = 22.191, p =
from Forbes top 500 international company list. .000). However, it seems both vendor and client
100 multinationals were identified that had in- sites are lacking quality gateway pages, which

234
Web Site Localization Practices

Table 1. Number of languages used


Number of Languages Used
M i n i - S t d .
N Maximum Mean
mum Dev.
Vendor No. of Languages 53 3 16 7.11 3.06
Multinational No. Of Lan-
102 5 38 19.38 7.44
guages

Table 2. Web site global gateway page


Vendor Multinational Mu lt i n at ion a l
Vendor %
Frequency Frequency %
No Link 0 7 0 6.80
Not Easy to Locate 6 20 11.32 19.42
Located at Middle Third 18 12 33.96 11.65
Located at Upper Right Corner 24 33 45.28 32.04
A Dedicated Global Gateway
5 31 9.43 30.10
Page
Total 53 103 100.00 100.00

are crucial to drive international online traffic


to country-specific sites. Localization Assessment of Country-
Specific Web Sites (Germany and
Use of Country-Specific Domain Spain)

The goal here is to see if the company has invested • Translation Depth: Translation depth was
in buying the country code top level domains also measured to see to what extent are the com-
called cctld for the country. The use of ccTLD helps panies translating their Web pages relative
in international search engine optimization and to U.S. English Web pages. To measure
also shows commitment of the company to that translation depth the study counted the
country market (see Table 3). The analysis shows number of English page and local language
that no vendor site was using ccTLD exclusively to primary links or main links on the home
create international sites. Most vendors were using page of English and local language site.
some extension of .com/Spain of .com/Germany. The results, in table 4, show that on vendor
The client multinational Web sites did relatively Web sites about 85 percent of English pages
better in terms of use of ccTLD. About 26 percent links were translated. On the other hand
of multinational sites were fully using ccTLD in terms of Multinational sites, on average
for their international Web sites. The results of about 67 percent of English page links were
a chi-square test for two independent samples translated.
indicates there is a significant difference in the • Content Localization: This Category mea-
use of ccTLD between vendor and multinational sures to what extent the company has local-
sites (χ2 = 20.346, p = .000). ized its Web site content in terms of local

235
Web Site Localization Practices

Table 3. Web site use of country-specific domain


Vendor Multinational Mu lt i nat ional
Vendor %
Frequency Frequency %
No ccTLD 22 20 42.31 19.80
Not Fully Using ccTLD 30 54 57.69 53.47
ccTLD Used 0 27 0.00 26.73
Total 52 101 100.00 100.00

Table 4. Web site English vs local language links


M i n i -
N Maximum Mean Std. Dev.
mum
Vendor - English 53 5 92 28.51 20.04
Multinational - English 101 4 299 56.08 47.10
Vendor - Local Language 102 2 96 24.23 19.01
Multinational - Local Language 202 3 177 39.73 30.55

support, e-commerce related information, which were equivalent to their US Web


and navigational ease. site, compared to 24 percent by multina-
• Local Customer Support and Contact: tionals. The results of a chi-square test for
By analyzing the level of local customer two independent samples indicates there
support it can be measured to what extent is a significant difference in the use of
is the company localizing its customer local customer support between vendor
service efforts for a specific-locale (see and multinational sites (χ2 = 21.200, p =
Table 5). The results show that while both .000).
the vendors and the client multination- • Availability of all policies and e-com-
als are not fully localizing their sites, merce information such as shipping
the vendor sites depict far less degree of policy, return, privacy, terms of use,
localization efforts. Only about 4 percent copyright etc.
of vendor sites had local support pages

Table 5. Web site level of customer support


Vendor Multinational
Vendor % Multinational %
Frequency Frequency
No Online Support 6 14 5.94 6.93
Basic Support 15 20 14.85 9.90
Basic Support - Customer Con-
38 68 37.62 33.66
tact
Several Pages of Support 38 51 37.62 25.25
Equivalent to English Site 4 49 3.96 24.26
Total 101 202 100.00 100.00

236
Web Site Localization Practices

Table 6. Web site e-commerce and information use policies

Vendor Multinational
Vendor % Multinational %
Frequency Frequency
Not Available 70 26 69.31 12.87
Only One or Two of the Policies 2 30 1.98 14.85
2-3 Policies Available 3 18 2.97 8.91
Most Policies Available 4 32 3.96 15.84
All Policies Available 22 96 21.78 47.52
Total 101 202 100.00 100.00

Table 7. Web site navigational ease


Vendor Multinational
Vendor % Multinational %
Frequency Frequency
Very Poor 6 7 5.94 3.47

Poor 24 27 23.76 13.37


Few Navigational Elements 35 54 34.65 26.73
Navigational Elements Seen 28 61 27.72 30.20
Navigation Elements Equivalent to English
8 53 7.92 26.24
Site
Total 101 202 100.00 100.00

Under this category the results for vendor and the other hand the 26 percent of the mul-
client sides were very different. Almost 70 percent tinational client site had navigational ele-
of vendor sites did not have policies related to ments equivalent to the US English pages.
e-commerce and information use (see Table 6). In general, both Vendor and Client Web
This shows that most vendors are not very global sites were not highly localized in terms of
in terms of conducting e-commerce. On the other navigation. However, the results of a chi-
hand, almost 47 percent of multinational sites had square test for two independent samples
all the policies available. These results suggest indicates there is a significant difference
that at least half the multinationals are localizing in the navigational ease between vendor
their site in terms of e-commerce readiness. The and multinational sites (χ2 = 18.085, p <=
results of a chi-square test for two independent .001).
samples indicates there is a significant difference • Layout and cultural adaptation
in the availability of policies between vendor and • Web Page Structure (see Table 8): The
multinational sites (χ2 = 101.136, p = .000). Overall Look of the Site and Design:
Under this category the objective is to
• Navigational ease in terms of sitemap, measure to what extent the look and the
local search, navigation buttons etc (see layout of the Web site has been localized
Table 7): Analysis of navigation revealed for a specific-locale. Surprisingly, almost
that only about 8 percent of local vendor 92 percent of vendor international sites
sites had navigational elements that were for Spain and Germany were basically
equivalent to their US home pages. On standardized templates of their US site.

237
Web Site Localization Practices

Table 8. Web site page structure


Vendor Frequency Multinational Frequency Vendor % Multinational %
Standardized 96 92 92.31 45.10
Mostly St andard-
5 27 4.81 13.24
ized
Some Differences 3 37 2.88 18.14
Localized 0 28 0.00 13.73
Highly Localized 0 20 0.00 9.80
Total 104 204 100.00 100.00

Table 9. Web site local culture


Vendor Frequency Multinational Frequency Vendor % Multinational %
Standardized 97 72 93.27 35.29
Mostly St andard-
4 52 3.85 25.49
ized
Some Differences 1 38 0.96 18.63
Localized 0 34 0.00 16.67
Highly Localized 2 8 1.92 3.92
Total 104 204 100.00 100.00

Multinational sites also did not seem to though a growing body of evidence is
achieve much localization under this cat- suggesting that cultural customization of
egory with almost 45 percent international sites leads to better attitude and intentions
sites being standardized. The results of to buy online (Singh & Pereira, 2005).
a chi-square test for two independent The results of a chi-square test for two
samples indicates there is a significant independent samples indicates there is a
difference in the structure of the Web significant difference in the use of local
pages between vendor and multinational culture between vendor and multinational
sites (χ2 = 66.671, p = .000) (see Table sites (χ2 = 95.101, p = .000).
9).
• Use of local models, graphics, colors
and other cultural markers.Under this Discussion
category the study measures if the
site uses local models, different colors The results clearly show that companies (localiza-
more appropriate for the country, and tion vendors and multinationals) are currently not
cultural symbols. The results show that fully localizing their sites in terms of using country
only about 2 percent of vendor sites are code top level domains, global gateway pages,
localized or highly localized, compared customer support, e-commerce and information
to 20 percent of multinational sites. It use policies, navigation, Web site structure, layout,
seems neither vendors or multinational colors, and graphics. It is even more concerning
clients are truly focusing their efforts that companies selling localization services are
to culturally customize their sites, even actually localizing their own sites to a much lesser

238
Web Site Localization Practices

extent than multinational companies, the firms that dors are able to serve without much competition.
tend to be their clients. The localization industry However, as the localization industry consolidates
is not practicing what they preach. and matures, it will be difficult for small localiza-
In every single category examined in this tion vendors to remain competitive and profitable
study the multinational Web sites were shown by just providing generic translation/localization
to be more localized than the vendor Web sites. services to niche markets or local country mar-
A telling comparison is the number of distinct kets. Even large sized localization vendors risk
languages used. Vendor sites average using seven losing their competitive position due to industry
different languages compared to multi national consolidation, over-reliance on generic translation
sites that average nineteen different languages. services, and cut throat price-based competition.
This result exemplifies the lack of localization If the localization industry wants to keep its com-
practices being used by the vendors themselves. petitive position and provide a healthy industry
Furthermore, over 93 percent of vendor sites are environment for both small and large localization
culturally standardized. vendors to grow, it must go beyond generic product
None of the vendor sites were found to be us- offerings in the form of translation services, and
ing a ccTLD which is surprising as international expand the definition of localization to include
domains are crucial for international search engine not just translation but to also offer:
optimization. This may be due to the fragmented
nature of the localization industry wherein small • Localization of the Website lay out and
localization vendors from a specific country tend navigation based on locale-specific require-
to serve their own local market and are content ments.
with their local customer base. However, large • Cross-cultural Web site and user interface
localization vendor Web sites also seemed to usability research.
show lack of ccTLD use and an overall low level • Country-market analysis
of localization on various parameters we used • Cultural customization of Web sites to spe-
in this study. So, do these vendors really believe cific locales
that Web site localization practices are important? • International online business strategy and
From this study’s results, the picture we get is marketing expertise
that vendor sites are lacking commitment toward • Online branding and advertising localiza-
localization. However, before reaching any con- tion
clusion we should consider a bigger picture and • International search engine optimization
understand what are the reasons for localization and search engine marketing
vendors to not sufficiently localize their sites? In
the limitations and future research section we These are just some ways the industry can
discuss some of these issues. expand the definition of localization and offer
complementary services that can help companies
differentiate their offerings from their competitors
Managerial Implications and stay profitable.
Localization services firms, in order to be
With few large localization vendors like Lion- profitable in the future, must practice what they
bridge Technologies, SDL International, and preach. They must practice not only to exemplify
Translations.com holding the top positions there their services being sold, but also to appeal to an
seems to be a large segment of niche markets that international market. Research has shown that
small localization vendors or single language ven- individuals prefer Web sites that are localized

239
Web Site Localization Practices

to their own language and culture (Singh et al., international expansion?


2004). To further shed light into why localization
vendors are not actively localizing their sites we
need further research to investigate their overall
Limitations and Future globalization strategy by asking questions such
Research as:

The current study is an exploratory study focusing • Are the localization vendors just targeting
on the amount of Web site localization used by some large multinational companies from
vendors offering Web site localization services predominantly English speaking countries
compared to their clients level of Web site localiza- (U.S., U.K., Australia etc.) and a few other
tion (multinational company Web sites). So, the non-English speaking countries?
data analysis here is a simple frequency examina- • What resource and marketing constraints
tion. The goal of this study was to examine the do these vendor companies face?
use of localization practices frequency and that • Another interesting question to investigate
goal was met. However, further, more in-depth is to understand the top management will-
data analysis could be used in the future. Another ingness to globalization and their vision for
limitation to this study was the sample used. The globalization. It seems several small and
vendor Web sites may not be intended for an in- medium sized localization companies are
ternational audience. As stated earlier, many of run by top management which has primarily
the smaller localization firms are serving single, a translation background. So is it the lack
niche markets. On the other hand, multinational of business education background that is
company Web sites, by their nature, are meant restricting the global expansion of localiza-
for international consumption. Therefore, it is tion vendors?
expected that vendor sites may be less localized
than their client’s sites. However, the results show Thus, to get a full picture of Web globaliza-
that even the large localization services firms lack tion efforts of localization vendors, we should not
localized content on their Web sites to meet the lonely study localization vendors Web sites but
needs of an international market. Furthermore, also understand their overall globalization strategy
even the smaller firms should aspire to exemplify and how it has evolved over time.
the practices that they preach.
Future research directions should include
a closer examination of localization practices Conclusion
within the localization industry. A comparison
between the large localization service firms Web The findings of this research suggest that both
sites and their smaller, niche market, counterparts localization vendor and multinationals are barely
is needed. Also, a longitudinal study examining localizing their Web site offerings. This may not
the increased amount of localization used on be all bad news, as more multinationals seek to
the internet is warranted. Is the trend to local- tap online markets and compete for them, the
ize Web sites to a greater extent, or are more winner will be the multinational sites that are
Websites trying to serve a smaller, local niche truly localized and speak to their international
market instead of an international market? If customers in their language and culture. Local-
the trend is to serve international markets then ization service sales may be increased by simply
are the localization practices keeping up with practicing what they preach.

240
Web Site Localization Practices

Not only has research shown that consumers for Information Science and Technology, 55(13),
prefer localized Web site content, but research has 1199-1208.
also shown that by localizing you can increase
DePalma, D., & Beninatto, R. (2006). Language
traffic to Web sites (Ferranti, 1999), and increase
services 2006: Supply-side outlook. Research
willingness to purchase (Singh et al., 2004). Lo-
Report by Common Sense Advisory. Retrieved
calization services vendors may be missing out
July 28 2008, from www.commonsenseadvisory.
on increased sales by simply preaching and not
com
practicing. Actually, applying what they preach
to their own Web sites, according to research, Esselink, B. (2000). A practical guide to localiza-
should increase the amount of traffic to their tion (Language International world directory).
sites while also increasing the willingness to Philadelphia: John Benjamins Publishing Co.
purchase their localization services. This means
Ferranti, M. (1999). From global to local. Info-
that the localization industry can look forward
world, 21(41), 36-37.
to significant growth, but only if it can educate
its multinational clients about the importance Fletcher, R. (2006). The impact of culture on Web
of localization and the best way to do that is to site content, designs, and structure: An interna-
practice what they preach. tional and a multicultural perspective. Journal of
Communication Management, 10(3), 259-273.
Fock, H. (2000, September). Cultural influences
ACKNOWLEDGMENT
on marketing communication on the World Wide
Web. Paper presented at the Multicultural Market-
The data collection for this study was supported
ing Conference, Hong Kong.
by students and Gary Muddyman, CEO of Con-
versis Global. Globalization Industry Primer (LISA) (2007). Glo-
balization industry primer. LISA. Retrieved April
3, 2009, from http://www.lisa.org/Globalization-
References indust.468.0.html#c261.
Albers-Miller, N., & Gelb, B. (1996). Business
Grant, J., & Davis, L. (1997). Selection and use
advertising appeals as mirror of cultural dimen-
of content experts for instrument development.
sions: A study of eleven countries. Journal of
Research in Nursing and Health, 20, 269-274.
Advertising, 25(Winter), 57-70.
Hall, E. T. (1976). Beyond culture. Garden City,
Culnan, M. J., & Armstrong, P. K. (1999). Informa-
NY: Doubleday & Company.
tion privacy concerns, procedural fairness and
Han, S. P., & Shavitt, S. (1994). Persuasion and
impersonal trust: An empirical investigation.
culture: Advertising appeals in individualistic and
Organization Science, 10, 104-115.
collectivistic societies. Journal of Experimental
Cyr, D., & Lew, R. (2003). Emerging challenges Psychology, 30, 8-18.
in the software localization industry. Thunderbird
Hermeking, M. (2005). Culture and Internet
International Business Review, 45(3), 337-358.
consumption: Contributions from cross-cultural
Cyr, D., & Trevor-Smith, H. (2004). Localiza- marketing and advertising research. Journal of
tion of Web design: An empirical comparison Computer-Mediated Communication, 11(1). Re-
of German, Japanese and United Stats Web site trieved April 3, 2009, from http://jcmc.indiana.
characteristics. Journal of the American Society edu/vol11/issue1/hermeking.html.

241
Web Site Localization Practices

Internet World Stats (2008). World Internet us- Mooij, M. D. (1998). Global marketing and adver-
ers and population stats. Internet World Stats. tising. Understanding cultural paradox. Thousand
Retrieved July 28, 2008, from http://www.inter- Oaks, CA: Sage Publications.
networldstats.com/stats.htm.
Pollach, I. (2005). Corporate self-presentation on
Jarvenpaa, S. L., Tractinsky, N., Vitale, M. (2000). the WWW: Strategies for enhancing usability,
Consumer trust in an Internet store. Information credibility and utility. Corporate Communica-
Technology and Management, 1(1-2), 45-71. tions, 10(4), 285-301.
Jupiter Research (2008). Jupiter research. Jupiter Simon, S. J. (2001). The impact of culture and
Research. Retrieved July 28, 2008, from http:// gender on Web sites: An empirical study. Da-
www.jupiterresearch.com/bin/item.pl/home/. tabase for Advances in Information Systems,
32(1), 18-37.
Kwintessential (2009). Culture and Web site
localization. Kwintessential. Retrieved April 3, Singh, N., & Boughton, P. (2002). Measuring Web
2009, from http://www.kwintessential.co.uk/ site globalization: A cross-sectional country and
translation/articles/culture-Website-localization. industry level analysis. Proceedings from Ameri-
html. can Marketing Association Educators’ Confer-
ence (Winter), Austin, TX. Chicago: American
Lee, L. Z. (2005). Evolving localization and its
Marketing Association.
brand extension. Galaxy Newsletter, Q4. Retrieved
April 3, 2009, from http://www.gala-global.org/ Singh, N., & Boughton, P. (2005). Measuring
newsletters/newsletter_3516.html. Web site globalization: A cross-sectional country
and industry level analysis. Journal of Web Site
Liu, C., Marchewka, J., & Ku, C. (2004). American
Promotion, 1(3), 3-20.
and Taiwanese perceptions concerning privacy,
trust, and behavioral intentions in electronic com- Singh, N., Furrer, O., & Ostinelli, M. (2004). To
merce. Journal of Global Information Manage- localize or to standardize on the Web: Empirical
ment, 12(1), 18-40. evidence from Italy, India, Netherlands, Spain,
and Switzerland. Multinational Business Review,
Localization Program at California State Univer-
12(1), 69-88.
sity, Chico, College of Business (2008). What is lo-
calization?. The Localization Program. Retrieved Singh, N., & Little, J. (2009). Culturally custom-
November 26, 2008, from http://www.csuchico. izing international Web sites. In M. A. Shareef,
edu/localize/whatislocalization.html. Y. K. Dwivedi, M. D. Williams, & N. Singh
(Eds.), Proliferation of the Internet economy: E-
Lommel, A. (2003). LISA, The localiza-
commerce for global adoption, resistance, and
tion industry primer, 2nd edition. LISA. Re-
cultural evolution. Hershey, PA: IGI Global.
trieved July 28, 2008, from http://www.lisa.
org/Globalization-Indust.468.0.html?&no_ Singh, N., & Matsuo, H. (2004). Measuring cul-
cache=1&sword _list[]=indust r y&sword _ tural adaptation on the Web: A study of U.S. and
list[]=localizatio&sword_list[]=primer. Japanese Web sites. Journal of Business Research,
57(8), 864-872.
Luna, D., Peracchio. L. A., & de Juan, M. D.
(2002). Cross-cultural and cognitive aspects of Singh, N., & Pereira, A. (2005). The culturally
Web site navigation. Journal of the Academy of customized Web site: Customizing Web sites for the
Marketing Science, 30(4), 397-410. global marketplace. Burlington, MA: Elsevier.

242
Web Site Localization Practices

Singh, N., Toy, D. R., & Wright, L. K. (2009). A The Localization Industry Standards Association
diagnostic framework for measuring Web site (2008). What is globalization? The Localiza-
localization. Thunderbird International Business tion Industry Standards Association. Retrieved
Review, 51(3), 281-295. July 28, 2008, from http://lisa.org/What-Is-
Globalization.48.0.html.
Violino, B. (2001). E-business lurches abroad.
Internet Week, March 19th. Retrieved July 28,
2008, from http://www.internetweek.com.

This work was previously published in International Journal of E-Adoption, edited by S. Sharma, pp. 36-54, copyright 2009
by IGI Publishing (an imprint of IGI Global).

243
Section II
Development and Design
Methodologies

This section provides in-depth coverage of conceptual architectures, frameworks and methodologies
related to the design and implementation of Web technologies. Throughout these contributions, research
fundamentals in the discipline are presented and discussed. From broad examinations to specific dis-
cussions on particular frameworks and infrastructures, the research found within this section spans the
discipline while also offering detailed, specific discussions. Basic designs, as well as abstract develop-
ments, are explained within these chapters, and frameworks for designing successful Web sites, Web-
based applications, and Web portals are provided.
245

Chapter 2.1
Perceptions of Mobile
Device Website Design:
Culture, Gender and Age Comparisons
Dianne Cyr
Simon Fraser University, Canada

Milena Head
McMaster University, Canada

Alex Ivanov
Simon Fraser University, Canada

ABSTRACT were tested in a controlled laboratory experiment on


an Internet enabled phone. The results of a quantita-
Anytime anywhere services offered through mobile tive analysis were statistically inconclusive in terms
commerce hold great potential to serve customers of cultural and gender differences, but significant
in wireless environments. However, there is limited differences were found between older and younger
understanding of how mobile Web site design is per- users. However, an in-depth qualitative analysis
ceived by diverse users. This chapter explores how of interview transcripts revealed some interesting
users who differ by culture, age, and gender perceive differences among cultural, gender and age groups.
the design of a mobile device and their subsequent Consistent with findings in the stationary Internet
level of satisfaction with the device. Sixty subjects domain, design elements were found to impact
satisfaction with mobile services.
DOI: 10.4018/978-1-60566-366-1.ch010

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Perceptions of Mobile Device Website Design

INTRODUCTION examines cohorts by age (younger versus older


users) is practically nonexistent when design is
Organisations will be well served to not be com- considered, although some studies have focused
placent with their wireless site design efforts… on special needs and preferences of older users
[But] before wireless site designers can address the (Goodman et al., 2005), or features of a mobile
usability challenge, and before organisations can device perceived by user groups as adding value
leverage the commercial benefits of m-commerce, (Anckar & D’Incau 2002).
a deeper understanding of what aspects of usability To explore the role of user differences con-
is important to users and how they may differ in cerning the perception of the design of a mobile
a wireless context is required. (Venkatesh et al., device, users who differ on cultural, gender and
2003, p.56) age dimensions were tested on an Internet enabled
cellular phone. Related to culture, Canadian and
Chinese cultures were chosen due to acknowl-
Mobile commerce1 (or m-commerce) has huge edged diversity (Hofstede, 1980). Between-group
potential to serve customers in wireless environ- comparisons were conducted with respect to screen
ments. The adoption of m-commerce is dependent design (including information design, navigation
on consumer acceptance of new and well-designed design, and visual design), and satisfaction with
technologies (Ancker & D’Incau, 2002; Coursaris the mobile device. In an exploration of these
et al., 2003: Kim et al., 2002; Kumar & Zahn topics, the paper provides a review of relevant
2003; Nysveen et al., 2005; Perry et al., 2001; literature leading to the hypotheses for testing, the
Schrott & Gluckler, 2004; Yang, 2005). It is ex- methodology used, an elaboration of results, and
pected design characteristics may influence user discussion of the findings. Given the increased
perceptions towards a mobile device. Congruent diversity of mobile users, developing an expanded
with work done by Cyr (2008), Information De- understanding of user perceptions and preferences
sign (ID), Navigation Design (ND), and Visual not only has theoretical importance, but also serves
Design (VD) may all contribute to user adoption to enhance the reengineering of devices to best
of a technology, as well as to satisfaction with a meet consumer requirements.
mobile technology.
Concerning user attitudes of handheld devices,
it is also expected that diverse categories of us- MOBILITY AND DESIGN
ers based on culture, gender, or age may react
differently to using the device. There is growing In the realm of the stationary Internet, effective
literature on cross-cultural website design, mostly website design engages and attracts online con-
evaluated within the context of the stationary sumers (Agarwal & Venkatesh, 2002; Fogg &
Internet (Becker, 2002; Chau et al., 2000; Cyr, Tzeng, 1999; Hui & Triandis, 1985; Morgan &
2008; Cyr et al., 2006; Cyr & Trevor-Smith, 2004; Hunt, 1994; Schultz, 2003). According to Gom-
Marcus & Gould, 2000; Sun, 2001). More recently, mans et al. (2001), ‘A website has to be designed
research has examined culture and mobile data for a targeted customer segment…’ Chau et al.
services (Choi et al., 2006). Investigations have (2000) argue the modes of information presented
likewise examined gender and design in the context on the Internet, and the quality of graphics has a
of mobile devices, but research results are mixed significant impact on user experience. Research
(Anckar & D’Incau, 2002: Kwon & Chidamba- in design suggests various guidelines for effective
ram, 2000; Teo et al., 1999). Finally, research that Web navigation (Childers et al., 2001; Farkas &

246
Perceptions of Mobile Device Website Design

Farkas, 2000), criteria for optimal Web design ciples of interface design can be transferred to
(Bernard, 2002; Egger 2001), and how aesthetics mobile devices. He examines various issues such
and usability might be linked (Tractinsky, 1997). as content, user interaction with the device, issues
The sensory experience of the website can help of reading text on small screens, rapid serial visual
to determine if a user stays and shops at a site presentation, and browser types.
(Rosen & Purinton, 2004; Yoon, 2002). Relevant to the current research, design catego-
The quality of handheld displays that favor ries for information design, navigation design, and
enhanced information design and visual design visual design as suggested by Garrett (2003) were
is steadily increasing and affects user perceived selected for systematic examination. A definition
effectiveness of the presentation (Rau et al., 2006). of each category follows.
While mobile screens are much smaller than those
available on the stationary Internet (Schmidt & a. Information design: Elements of the site
Frick, 2000), various studies demonstrate com- that convey accurate or inaccurate informa-
prehension rates on smaller screens are generally tion to a user. For instance, the location of
equivalent to their larger counterparts (Dillon & an icon on the screen would be the domain
McKnight, 1990; Duchnicky & Kolers, 1983; of information architecture, but whether
Resiel & Shneiderman, 1987). Other researchers or not that icon conveys the right informa-
consider the smaller screens of mobile devices tion to a user is the domain of information
a ‘serious obstacle to usability of the mobile design. Clear and logical presentation of
Internet’ (Chae & Kim, 2004, p. 165). Sarker & information about services or products is
Wells (2003) examined interface characteristics also a component of information design.
and network capabilities that affect the implemen- b. Navigation design: The navigational
tation and acceptance of wireless phones. They scheme used to help or hinder users as they
discovered while users were ‘quite forgiving of access different sections of the site, such as
physical limitations of the device due to techno- the location and format of navigation aids.
logical constraints, they were bothered by flaws c. Visual Design: Elements that deal with the
in the interface of the devices’ (p. 37). balance, emotional appeal, aesthetics, and
In recent years there has been increased atten- uniformity of the website overall graphical
tion to mobile usage and in 2004 the International look. This includes colors, photographs,
Journal of Human-Computer Studies devoted a shapes, or font type.
special issue to this topic. The issue addressed
mobile use in a variety of contexts including These categories are represented in other work
human characteristics and interface systems, in design (Agarwal & Venkatesh, 2002; Yoon,
although design aspects of the mobile website 2002; Flavian et al., 2005; Palmer, 2002; Simon,
were not specifically considered. Further, in 2005 2001), and while not exhaustive are representative
Behaviour & Information Technology devoted a of key elements of website usability. The same cat-
special issue to mobility from a human computer egories were used by Cyr (2008) to study website
interaction (HCI) perspective. Topics included design across cultures, and by Cyr and Bonanni
navigation support, user acceptance and trust, and (2005) regarding website design and gender.
user evaluation of usability of mobile devices. In Further, in a study of mobile services Choi et al.
research in which interface design and usability (2006) consider three categories of user experi-
are examined for wireless devices in m-commerce, ence across cultures. These are content (similar
Tarasewich (2003) suggests many current prin- to information design), information architecture

247
Perceptions of Mobile Device Website Design

(which includes navigation design), and graphical significant cultural differences regarding value
user interface (which is similar to visual design). structures of the mobile Internet and their effect
In this research, information design, navigation on users’ satisfaction.
design, and visual design are considered in the More specifically, and with respect to infor-
specific context of a mobile device. mation design, research comparing user prefer-
ences in Canada, the U.S., Germany and Japan
Culture and Design for perceived access and presentation of product
information on a stationary computer uncovered
User preferences for website design features are few significant differences between the U.S.,
known to vary across cultures (Barber and Badre, Canada, and Germany, but significant differences
2001; Cyr, 2008; Evers and Day, 1997; Nielsen (p<.01) between these countries and a highly
and DelGaldo, 1996; Sun, 2001). Cyr and Trevor- collectivist2 culture like Japan (Cyr et al., 2005).
Smith (2004) examined design elements for 30 Based on qualitative comments from the study,
municipal websites in each of Germany, Japan, there appeared a desire on the part of Canadians,
and the United States. Significant differences Americans, and Germans for utility - at least as far
were found across countries for use of symbols as obtaining site information is concerned. Choi
and graphics, color preferences, site features, et al. (2006) examined cultural characteristics and
language, and content. In a study in which user user experience attributes in mobile data services
impressions were evaluated toward eight website in Korea, Japan, and Finland. Based on qualita-
design features, numerous differences were de- tive findings, the authors found user experience
tected between collectivist Japanese and Chinese attributes correlated to the user’s culture and
users with individualist British users (Hu et al., ‘Finnish participants showed a cultural profile
2004). In the current investigation participants opposite to that of the Koreans’ (p. 192). For ex-
are tested who are either Canadian or Chinese. ample, in the area of information design Koreans
With respect to these cultures, Singh et al. (2003) and Japanese (both collectivists), preferred large
compared domestic and Chinese versions of amounts of information on a single screen while
websites for 40 American-based companies and Finns (individualists) preferred direct, explicit
found differences in all the cultural categories communication and reacted negatively to large
examined. amounts of content. Related to the preceding stud-
Studies of m-commerce in different countries ies, like the Finns, Canadians are individualists.
and considering different cultures are rare, al- Alternately, Chinese are collectivists in alignment
though ‘an understanding of the cultural dimen- with Koreans and Japanese.
sions of a market can aid marketers immensely in
developing appropriate m-commerce services…’ H1a: There will be differences between Chinese
(Harris et al., 2005). In response, research in the and Canadian users in the perception of informa-
area of culture and mobility is beginning to emerge. tion design of a mobile device. Canadian users
Cross-country differences were found for adoption will prefer utility in information design, while
of mobile applications in Hong Kong, Japan, and Chinese prefer more detail and depth of informa-
Korea (Kim et al., 2004), between the UK and tion presented in a mobile medium.
Hong Kong (Harris et al., 2005), and between
France and the USA (Carlson et al., 1999). Lee Regardless of culture, users prefer easy to navi-
et al. (2002) compared Japan and South Korea gate websites. In an experiment using a stationary
in a study of m-commerce usage, and found website, Simon (2001) found that North Americans

248
Perceptions of Mobile Device Website Design

prefer navigation that enhances movement and concerned than Canadians with aesthetic beauty
makes the interface simpler to use. In the study in visual design.
by Choi et al. (2006) in mobile data services as
already mentioned, Koreans and Japanese liked
clear and logical ordering of menu items, while Gender and Design
Finns mentioned they most liked search facilities.
Despite the paucity of prior research on navigation Gender is frequently used as a basis for segmenta-
design across culture, and particularly in the con- tion, and researchers have attempted to understand
text of mobile devices the following exploratory the fundamental similarities and differences be-
hypothesis is offered. tween the men and women for decades (Deaux &
Kite, 1987; Putrevu, 2001). Past empirical studies
H1b. There will be no differences between Chinese have shown significant gender differences across
and Canadian users in the perception of navigation a variety of tasks and domains. For example,
design of a mobile device. Both Canadians and men often perform better than women on spatial
Chinese will prefer simple and logical navigation orientation tasks, whereas females tend to score
formats. better on verbal or linguistic tasks (Simon, 2001;
Deaux & Kite, 1987). Similarly, men and women
User preferences vary by culture with respect differ in their reactions to visual images, affecting
to visual design of the interface. Color varies by recall and recognition (Jones et al., 1998).
culture. Red means happiness in China, but danger There are also gender differences in computer
in the US (Barber and Badre 2001). When applied usage. Men and women diverge in Web acceptance,
to Web design, color may impact user expecta- with perceived usefulness found to positively
tions of the interface as well as overall satisfac- influence intention to use the Web more in men
tion (Barber & Badre, 2001). In a cross-cultural than women (Sanchez-Franco, 2006). Women
study on website design, a Japanese respondent use computers for collaboration and networking,
indicated a preference for more pictures and an while men view computers as a tool for obtain-
“emotional approach” (Cyr et al., 2005). In other ing and evaluating content (Gefen & Straub,
work specifically focused on images used in 1997). Pearson et al. (2003) examined gender
website design, in qualitative analyses Canadians as a moderating variable to end-user computer
perceived images to have aesthetic, affective and efficacy, and found no differences between men
functional qualities while Japanese respondents and women although women were somewhat less
focused only on affective qualities (Cyr et al., confident to learn new computer applications. In
2006). Sun (2001) found that users from cultures other research, narrowing of differences between
such as China or Japan have a strong preference men and women has occurred concerning software
for visuals and aesthetic beauty of the interface. In use, anxiety, and enthusiasm (Rainer et al., 2003).
a mobile context, Koreans and Japanese preferred Specific to mobile commerce, studies indicate
colorful screen design and Finns preferred simple that among Internet users men are predisposed to
screen design with less emphasis on color (Choi mobile adoption more than women (Yang, 2005;
et al., 2006). Brennan, 2000; Joines et al., 2003; Park & Jun,
2003; Rohm & Swaminathan, 2004). Contrary to
H1c. There will be differences between Chinese these studies, gender differences were not detected
and Canadian users in the perception of visual in Spanish users related to shopping patterns and
design of a mobile device. Chinese will be more m-commerce adoption (Bigne et al., 2005).

249
Perceptions of Mobile Device Website Design

It is anticipated the design of a website will Age and Design


impact user preferences, which in turn may pro-
duce different reactions between men and women Age is another common dimension used to
(Chen & Dhillon, 2003). In one investigation in segment consumer and user groups. There has
which gender and design are considered, Simon been a growing level of interest and research in
(2001) tests users’ perceptions of a site, which issues relating human computer interaction and
refer to information richness, communication ef- age groups (Goodman & Lundell, 2005). Some
fectiveness, and communication interface. Women researchers have focused on the unique design
were found to have a less satisfied perception of characteristics posed by older adults. Hawthorn
the websites than men. Other work demonstrates (2000) lists various physical, sensory and cognitive
differences between men and women for content limitations that may alter with increased age and
and navigation (Maltby et al., 2003) and preference their implications for interface design. Physical
for color or graphics (Rodgers & Harris, 2003). In limitations, such as reduced dexterity and preci-
a study of website design, Cyr and Bonanni (2005) sion, can make the use of small and delicate input
found specific information design elements (such devices (as found on various mobile devices) more
as site organization and presentation of product difficult. Sensory limitations can create limitations
information) were perceived more favorably by for the design of computer output and cognitive
men than women. Further, men found the sites limitations can affect the design of the interface
easier to navigate, and liked certain visual design itself. For example, cognitive spatial ability has
aspects such as degree of interaction and anima- predicted computer performance (Kelley and
tions more than women. Charness 1995), and was demonstrated to decline
The above findings on a stationary website with increasing age (Salthouse, 1992). In other
suggest there will be differences between the research, age affected the retention of computer
preferences of men and women for interface training (Brown, 2001), and confidence in learn-
design in a mobile context. This assumption will ing new applications (Crosby et al., 2003). Older
be explored in the following hypotheses. employees generally exhibited a less positive
attitude towards computers (Brown, 2001), and
H2a. There will be differences between men and were less satisfied users (Simmers & Anandarajan,
women in the perception of information design 2001) than younger employees. It is expected
of a mobile device. that well designed visual cues such as text links
and icons are able to support the needs of older
users. The format for organizing Web contents
H2b. There will be differences between men and and the amount of information appearing on a
women in the perception of navigation design of screen enable higher performance for older users
a mobile device. as their visual search skills and selective attention
diminish (Ellis & Kurniawan, 2000).
As with gender, studies that investigate mobile
H2c. There will be differences between men and use or m-commerce with consideration of age or
women in the perception of visual design of a design are few. Stroetmann et al. (2002) found
mobile device. that 43% of elderly people surveyed had at least

250
Perceptions of Mobile Device Website Design

some difficulty with mobile devices and 21% had Design and Satisfaction
considerable difficulty because of some physical
or cognitive impairment. In a study of Swed- For many years customer satisfaction has been
ish teenagers, Weilenmann and Larsson (2000) studied in physical environments (Balasubrama-
reported that young people use a mobile phone nian et al., 2003; Parasuraman et al., 1988; Oliver,
in radically different ways from more mature 1980 and 19991999). More recently, research into
adults. Younger users use a mobile device more consumer satisfaction has turned to the Web do-
for expression than for information, and for social main and examines ‘stickiness’ and ‘the sum of all
purposes rather than for coordination or efficiency. the website qualities that induce visitors to remain
In a study with Spanish users, younger people at the website rather than move to another site’
are more predisposed to m-commerce adoption (Holland & Baker, 2001). According to Anderson
than older Internet users (Bigne et al., 2005). In and Srinivasan (2002), e-satisfaction is defined
a report on mobile use in India (MACRO 2004), as the contentment of the customer with respect
limited adoption of mobile devices among older to his or her prior purchasing experience with
users resulted from small buttons on the hand- a given electronic firm. In the present research
set and tiny screens that impede user visibility. we adapt the definition for online satisfaction
Goodman et al. (2005) found that text, speech, presented by Anderson and Srinivasan to suggest
and photographs were all effective ways to pres- mobile satisfaction refers to contentment of the
ent landmark information to older users using a customer with the experience of using the mobile
mobile navigation aid. interface. However, there is no requirement for
In this investigation we are interested to ex- completion of a purchase in the current context.
amine design characteristics (such information This definition is in alignment with Chae et al.
design, navigation design, and visual design) as (2002) who examined information quality related
outlined by Garrett (2003) for a mobile device, to user satisfaction for mobile Internet services.
in this case an Internet enabled phone. To our Online satisfaction motivates online shoppers
knowledge this is the first attempt to systemati- to stay at the site and return to the site in the future
cally examine these design features and how they (Flavian et al., 2005; Bhattacherjee, 2001; Doll &
may differ for older or younger mobile users. To Torkzadesh, 1988; McKinney et al., 2002) thus
develop some understanding in this area, the fol- yielding a loyal customer outcome. In research
lowing exploratory hypotheses are offered. into stationary websites, customer satisfaction is
affected by content and context of the website
H3a. There will be differences between older and (Flavian et al., 2005; Teo et al., 2003). More
younger users in the perception of information specific to information design, Szymanski and
design of a mobile device. Hise (2000) discovered that product information
and site design are critical to creating a satisfying
customer experience. The experience of online
H2b. There will be differences between older and shopping can be affected by the richness of product
younger users in the perception of navigation information presented (Palmer, 2002; McKinney
design of a mobile device. et al., 2002), and is a dominant concern of the user
(Kateranttanakul & Siau, 1999; Pitt et al., 1995;
Zhang et al., 2000). A positive navigation experi-
H3c. There will be differences between older and ence and perception of a well-designed site may
younger users in the perception of visual design likewise result in online consumer satisfaction
of a mobile device. (Agarwal & Venketesh, 2002; Fogg & Tzeng,

251
Perceptions of Mobile Device Website Design

1999; Palmer, 2002; Fogg et al., 2002; Hoffman & and Benbasat (2003) found that mobile interface
Novak, 1996; Koufaris, 2002; Nielsen, 2001), an features can positively influence users’ attitudes
enjoyable online shopping experience (Childers et regarding product presentations. Graphical user
al, 2001), and sales (Lohse & Spiller, 1999). Yoon interfaces, information architecture, and content
(2002) found navigation functionality resulted in all contributed to differences in perceived sat-
satisfaction, and induced Web visitors to remain isfaction with mobile data services (Choi et al.,
at the site. Lohse and Spiller (1998) demonstrated 2006). The preceding considerations lead to the
that designing online stores with friendly user final hypotheses:
interfaces positively influences traffic and sales.
Cyr (2008) found information design, navigation H4a. Perception of information design of a mobile
design, and visual design to all positively impact device will impact user perceived satisfaction with
satisfaction for users from multiple countries. the mobile device.
While design research is limited within the
mobile context, Tarasewich (2003) concluded:
‘[A]esthetics, along with usability, may also H4b. Perception of navigation design of a mobile
be part of designing an overall enjoyable user device will impact user perceived satisfaction with
experience with mobile devices’ (p. 12). Jiang the mobile device.

Figure 1. Research model for mobile services

Table 1. Country comparisons (Source: based on Hofstede, 1980)

Country Dimension Canada China


Power Distance Low (39) High (80)
Uncertainty Avoidance Low (48) Medium (60)
Masculine Medium (52) Medium (50)
Individualism High (80) Very Low (20)
Long-Term Orientation Very Low (23) Very High (118)

252
Perceptions of Mobile Device Website Design

H4c. Perception of visual design of a mobile de- experienced users, and had owned a cell phone
vice will impact user perceived satisfaction with for approximately 4.5 years. Demographics of the
the mobile device. sample appear in Table 2.

The above hypotheses as examined in this Experimental Site and Device


research are visually depicted in Figure 1.
Given the emphasis on aesthetics in usability, an
attractive site was required, without interference
RESEARCH METHODOLOGY from a slow or faulty live Internet connection.
Most WAP sites in North America and Europe are
Participants primarily text based, however the Lonely Planet
offered some visual treatment (www.lonelyplanet.
Sixty participants were recruited who were either com with access from a mobile phone at wap.
Chinese or Canadian in origin (30 in each cat- lonelyplanet.com). Therefore the City Guide ver-
egory). These countries were chosen to represent sion of this site was chosen for the experiment and
diverse cultural characteristics as per Hofstede further enhanced. The site for this study featured
(1980). Refer to Table 1. a colored background and text cells, as well as
Participants were also categorized by gender photographs, maps and icons, among other fea-
and age. Consistent with Aarnio et al. (2002), tures. Each version of the site was created in both
younger participants were 34 and younger and English and Chinese. For the Chinese version,
older participants were 35 or older. A relatively site contents were translated to Chinese, ‘back-
small sample size of 30 was chosen for each group translated’ from Chinese to English, and then this
due to the individualized and in-depth require- version was compared with the original English
ments of the data collection process. To participate, version to verify content equivalence. Screenshot
individuals must have used a cell phone for at least pages appear in Appendix A. Testing occurred on
one year. In fact, all participants in the study were a Nokia 6600 Internet enabled cell phone. This

Table 2. Participant demographics

Canadian (n=30) Chinese (n=30)


Age < 35: 67% < 35: 60%
≥ 35: 33% ≥ 35: 40%
Gender Male: 53% Male: 43%
Female: 47% Female: 57%
Highest level of education High school: 40% High school: 27%
Technical degree: 3% Technical degree: 3%
Undergraduate/College: 43% Undergraduate/College: 53%
Masters/Doctorate: 13% Masters/Doctorate: 17%
Time spent online/week 22 hours 17.4 hours
Time owned a cell phone 4.8 years 4.4 years
Mobile Internet browsing experience 20% 17%
Use of the mobile Internet Buy movie/concert tickets: 33% Buy movie/concert tickets: 30%
(if available) Download games/ring tones: 40% Download games/ring tones: 60%
Check news: 50% Check news: 67%
Browse places to eat, shop, etc.: 47% Browse places to eat, shop, etc.: 67%
Other: 33% Other: 7%

253
Perceptions of Mobile Device Website Design

phone was considered very suitable to the aims comfortable with the device they were read the
of the investigation as it has one of the largest following narrative:
screens on the market and has a superior color
display (65,536 TFT) capable of laptop quality Imagine that you have just arrived in San Fran-
images. The interface included a joystick option cisco to meet up with an old friend. Your friend
for easy navigation. has suggested that you select a restaurant on your
To prevent problems with download and brows- cell phone, and call her back with the address. Use
ing delays typical of a real website using WAP the bookmarked CityGuide site to accomplish this
technology, website pages for the Lonely Planet task. Spend as much time as you need browsing
were downloaded and saved locally on the cell through the featured listings for San Francisco.
phone. Based on feedback obtained from a pilot There is no need to actually write down any infor-
focus group (of 6), participants were not aware of mation or make any calls. Just let me know what
this manipulation and perceived the connection to your selection is when you’re finished.
be ‘real’. Important to this study, saving the site
locally allowed the content to be modified and The device was then handed to the participant
the display speeds to be controlled. with the browser opened at the introductory page of
the site. The site listing for San Francisco featured
Experimental Tasks four restaurants. Participants first completed the
task, and then responded to a survey and were
The pilot study with 6 participants was used to pre- interviewed regarding their experience. At the end
test potential tasks and the experimental protocol of the experiment, participants were debriefed and
including survey items and interview questions. received a $20 honorarium for their time.
Participants in the pilot study were initially asked
to perform three information retrieval tasks: find- Survey Instrument
ing movie listings at a local theatre, choosing a
restaurant in a different city, and booking a hotel in Following the completion of the experimental task,
a different city. It was decided the restaurant task subjects were asked to complete a paper-based
was most suitable because it was preferred by the survey. In this section, we provide an overview
pilot subjects and afforded excellent visual design of the survey measurement items, focusing on its
opportunities. For the restaurant task, photos of content and construct validity.
the venue’s interior as well as the map showing Content validity considers how representative
the location were included. and comprehensive the items are in creating the
In the full study, each participant went through experimental constructs. Validity is assessed by
the experiment individually, under the supervi- examining the process by which the construct
sion of a research investigator. For Chinese items were generated (Straub, 1989). Constructs
participants, all documents were translated and should draw representative questions (items) from
back translated, and a translator was available as a universal pool (Cronbach, 1971; Kerlinger,
required. The session began with a brief introduc- 1964). In this research, survey items were adapted
tion and completion of a background data sheet, from previously validated work on Information
followed by familiarization with the Nokia 6600 design (Cyr et al., 2004 and 20052005), Navi-
device including a written summary of key func- gation design (Cyr et al., 2004 and 20052005),
tions. Once it was determined participants were Visual design (Cyr et al., 2004 and 2005;2005 van

254
Perceptions of Mobile Device Website Design

Table 3. Principle components analysis and reliability

ID ND VD SAT
Cronbach alpha .61 .84 .89 .84
AVE .57 .68 .60 .56

Items
ID1 .775 .221 .094 .101
ID2 .730 .020 .311 .245
ID3 .598 .271 .559 .176
ID4 .546 .227 .537 .300
ND1 .250 .700 .302 .175
ND2 .004 .891 .072 .249
ND3 .222 .862 .131 .039
VD1 .110 .093 .859 .146
VD2 .336 .157 .833 .182
VD3 .101 .122 .601 .510
VD4 .214 .200 .789 .291
SAT1 -.159 .240 .247 .661
SAT2 .326 .060 .223 .857
SAT3 .313 .102 .137 .810
SAT4 .408 .279 .290 .646

Notes: ID=Information Design; ND=Navigation Design; VD=Visual Design; SAT=Satisfaction

der Heijden, 2003), and Satisfaction (Cyr et al., However information design had two items (ID3
2004 and 20052005). All items were constructed and ID4) that had high cross-loadings with items
as agree-disagree statements on a seven-point in the visual design construct. As such, ID3 and
Likert scale. The complete survey appears in ID4 were removed from our analysis in order to
Appendix B. maintain discriminant validity.
Construct validity assesses the extent to which Discriminant validity can also be assessed by
a construct measures the variable of interest. In the average variance extracted (AVE) for each
other words, there should be high correlations construct. As shown in Table 2, the AVEs were
between items of the same construct (convergent all above the recommended 0.50 level (Fornell &
validity), and low correlations between items Larcker, 1981), which meant that more than one-
of different constructs (discriminant validity) half of the variances observed in the items were
(Straub, 1989). Results of the principal compo- accounted for by their hypothesized factors.
nents analysis with varimax rotation appear in Construct reliability (internal consistency) of
Table 3. The loadings for navigation design, visual the four factors was examined using Cronbach’s
design, and satisfaction construct items exceed α-value. As shown in Table 2, α-values ranged
recommended thresholds (Hair et al., 1998). from 0.61 (for information design) to 0.89 (for

255
Perceptions of Mobile Device Website Design

visual design). Rivard and Huff (1988) suggest would change, and whether or not they found the
that this measure for reliability should be higher device useful.
than 0.5 and ideally higher than 0.7. Navigation Responses were content analyzed and coded
design, visual design and satisfaction α-values using Atlas.ti. This software provides an effective
are well past this recommended threshold, and means to analyze qualitative data such as interview
the α-value for information design (with its two transcripts. The qualitative analysis process con-
items dropped) is also in an acceptable range. sisted of the following steps: (1) data preparation
Therefore, our survey instrument encompassed (i.e. interview transcription and formatting); (2)
satisfactory content validity (as evidenced from in vivo coding (use of participants’ words as code
drawing construct items from existing validated labels) and open coding (use of arbitrary labels for
literature); satisfactory convergent validity (as code labels); (3) category and concept building in
evidenced from high item loadings and construct which semantic relationships between codes are
reliability); satisfactory discriminant validity (as identified to build higher conceptual abstractions;
evidenced from low cross-loadings of factor items and (4) theory building based on interpretation
and the AVE for each factor); and satisfactory of the results.
construct reliability (as evidenced from Cron-
bach’s α-values).
RESULTS
Interviews
The descriptive statistics and correlations for the
Following the completion of the survey, subjects perceptual constructs are shown in Table 4. Each
were asked open-ended questions in a tape- of the design variables (ID, ND and VD) was cor-
recorded interview. The interview questions were related to each other, as well as to overall satisfac-
meant to solicit additional information about the tion. From the demographic variables, only age is
participants’ experiences with the experimental positively correlated with visual design.
task and interface. The questions probed how Overall, respondents have a rather favorable
participants liked the design of the site, what they impression of the mobile interface, with mean

Table 4. Descriptive statistics and correlations

Mean Std. Dev. ID ND VD SAT GEN CUL AGE


ID 5.43 1.04
ND 5.73 0.90 .37***
VD 5.28 1.09 .53**** .43***
SAT 4.98 1.11 .49**** .43*** .59****
GEN ---- ---- .09 .01 .09 .08
CUL ---- ---- -.06 -.09 .02 .17 .10
AGE ---- ---- .04 .07 .26** .18 -.10 .07

Notes:
1 ID=Information Design; ND=Navigation Design; VD=Visual Design; SAT=Satisfaction
2. Descriptive statistics not provided for GEN, CUL and AGE, as these are dichotomous variables
3. * p<.10; ** p<.05; *** p<.01; **** p<.001

256
Perceptions of Mobile Device Website Design

scores of over 5 out of 7 for information design exact words as the basis for a code), and (ii) open
(ID), navigation design (ND), and visual design coding (using arbitrary labels to code the data).
(VD). The mean score for satisfaction is 4.98, Interview responses were systematically catego-
also indicating overall satisfaction with the site. rized in each design area (information, navigation
To probe these results further, a word count was and visual design). Altas.ti was used to create a
run on transcribed interviews using the atlas.ti concept map for the design areas, highlighting the
software. The word ‘easy’ appeared 69 times, actual number of responses in each code based
mostly in response to the question ‘Try to de- on gender and culture. Refer to Appendix C. Al-
scribe the navigation experience on the site’. As though the numbers are relatively low, they signify
such, the majority of users find the navigation response trends between the groups. In addition,
of the device easy. One respondent elaborates, representative quotations are included. Further,
‘[U]sing the joystick was pretty straightforward, more theoretical themes were created from the
once I remembered to go left and right to go up emerging concepts, across the design areas. The
the links. It was more intuitive to use the scroll main themes from our interview data were:
button up or down…’
Although information design receives a rela- • Information breadth: The number of al-
tively high mean score (5.43), more than half the ternative (restaurant) choices. This is en-
respondents thought there could have been more capsulated by comment such as ‘choices
information about the restaurants on the site. This are limited’ and unmet expectations for ‘a
split into two category codes: one related to not lot of restaurants represented’.
enough information for each restaurant, which • Information depth: The amount of detail
typically was lack of a menu (code menu) and two, for each alternative (restaurant) choice.
not enough choice of restaurants (under the code This is encapsulated by participants seek-
choice). In terms of visual design of the site, the ing ‘detailed menus’, ‘prices’, ‘famous
following quote captures some of the favorable dishes’, ‘parking’, ‘exterior shots’, ‘re-
sentiments as expressed by a number of users: views’ and ‘hours of operation’.
‘[E]asy to use, attractively displayed, something • Visual ease: Utilitarian view on the in-
that awes people, this display is very graphically terface design’s capability to facilitate the
appealing. I really want to play with this.’ Another task. This is captured through comments
user comments: ‘The colors were good. The colors such as ‘adequate resolution’, ‘easy to
were actually fairly robust. Resolution seemed to read’ and ‘easy to look at’.
be pretty good.’ • Visual beauty: Hedonic view on the beau-
ty of the interface design. This is captured
Tests of Culture, Gender, through comments such as ‘cute’ design,
and Age as Moderators ‘should be more charming’, ‘put some mu-
sic’, and ‘animation would be good’.
T-tests of differences between culture, gender, and • Navigation layout: The layout of the in-
age group means for design and satisfaction are formation within the site. Captured by
shown in Tables 5, 6, and 7 respectively. comments such as ‘laid out in a logical
Additionally, interview data was coded using way’, and too much ‘scrolling’.
two methods: (i) in vivo (using the participant’s

257
Perceptions of Mobile Device Website Design

Table 5. T-test of differences between culture group means for design and satisfaction

Group Means t-value p-value


ID Can: 5.48 0.433 .667
Ch: 5.37
ND Can: 5.81 0.688 .507
Ch: 5.66
VD Can: 5.27 -0.176 .861
Ch: 5.32
SAT Can: 4.80 -1.279 .206
Ch: 5.17
Notes: Can= Canadian; Ch=Chinese

Table 6. T-test of differences between gender group means for design and satisfaction

Group Means t-value p-value


ID M: 5.33 -0.698 .488
F: 5.52
ND M: 5.72 -0.076 .939
F: 5.74
VD M: 5.19 -0.685 .497
F: 5.39
SAT M: 4.87 -0.576 .567
F: 5.06
Notes: M=Males; F=Females

Table 7. T-test of differences between age group means for design and satisfaction

Group Means t-value p-value


ID <35: 5.39 -0.277 .784
≥35: 5.48
ND <35: 5.68 -0.626 .534
≥35: 5.82
VD <35: 5.08 -2.274 .027**
≥35: 5.66
SAT <35: 4.83 -1.501 .139
≥35: 5.25
Notes: * p < .1; ** p < .05; *** p < .01

258
Perceptions of Mobile Device Website Design

Table 8. Summary of interview analysis across culture, gender and age groups

Design Area Culture Gender Age


Information Design Canadian: focused on Information Male: no noticeable patterns Young: focused on Information
Breadth Female: focused on Information Breadth
Chinese: focused on Information Depth (in particular, Chinese Old: commented on both Infor-
Depth females) mation Breadth and Information
Depth, with more focus on Depth
Navigation Design Canadian: commented on both Male: some comments on Naviga- Young: very few comments on
Navigation Layout and Challenges tion Layout. navigation
(in particular, Canadian women Female: focused on Navigation Old: some comments on Naviga-
focused on Navigation Chal- Challenges (in particular, Cana- tion Challenges
lenges) dian women).
Chinese: very few comments on
navigation
Visual Design Canadian: focused on Visual Ease Male: No men commented on Young: some comments on
(in particular, Canadian men) Visual Beauty; Canadian men Visual Ease
Chinese: focused on Visual Beauty commented on Visual Ease Old: commented on both Visual
(mostly providing suggestions to Female: focused on Visual Beauty Ease and Visual Beauty (females
augment beauty) (in particular, Chinese females) only for Visual Beauty).

• Navigation challenges: Navigation/inter- prefer simple and logical navigation formats. In


action challenges mostly stemming from the case of Canadians they thought navigation
inexperience with the new technology. could in fact be simpler, and the Chinese made
Encapsulated by comments such as ‘coun- few specific comments. Finally, qualitative results
terintuitive joystick’ and not convenient to also support H1c. It was predicted Chinese would
‘press the left side to select the options’. be more concerned with aesthetic beauty of the
mobile interface than Canadians. In fact, Chinese
Table 8 summarizes the analysis of the in- (women only) even commented on how to aug-
terview transcripts across the above emerging ment the beauty of the interface, and Canadians
themes and individual differences (culture, gen- focused more on visual ease.
der, age). It is interesting to note that no male from either
Based on the survey data, it is surprising no culture made comments on the visual beauty or
statistically significant differences are evident for hedonic elements of interface design. However
culture or gender. However based on the qualitative there were some cases where women not only
analysis of the interview data some differences in commented on their hedonic preferences, but also
these categories are indicated. made distinctions of what they prefer versus what
These results support H1a that Canadians prefer they think men prefer. Overall, while our quan-
utility of information design, while Chinese prefer titative analysis rejects hypothesis 2 concerning
more detail and depth of information. Canadians differences in gender, our in-depth qualitative
were much more disappointed with the number analysis suggests there may be some interesting
of restaurant choices (Information Breadth) than differences in these categories.
Chinese, who focused on the lack of restaurant Some support is found for hypothesis 3. Using
details (Information Depth). This Chinese focus survey data, statistically significant differences ex-
on Information Depth was particularly evident ist between older users (35 or older) and younger
among Chinese women. H1b was also generally users (under 35) for visual design, with older
supported in that both Canadians and Chinese respondents indicating the design of the mobile

259
Perceptions of Mobile Device Website Design

Table 9. Predicting satisfaction

R2 ∆R2 β Sig.

GENDER .063 .013 .174 .551


AGE .415 .171
CULTURE .322 .270

ID .422 .391 .220 .097*


ND .230 .118
VD .420 .002***

ID x GENDER ..031 -.021 -.120 .583


ND x GENDER .192 .432
VD x GENDER -.025 .924

ID x AGE .175 .130 .061 .532


ND x AGE -.111 .343
VD x AGE .180 .122

ID x CULTURE .071 .022 .306 .258


ND x CULTURE -.105 .663
VD x CULTURE -.104 .738

ID x GENDER x AGE .018 -.035 -.005 .602


ND x GENER x AGE .077 .654
VD x GENDER x AGE .011 .948

ID x GENDER x CULTURE .012 -.041 .086 .848


ND x GENDER x CULTURE .181 .546
VD x GENDER x CULTURE -.261 .573

ID x AGE x CULTURE .131 .084 .361 .051*


ND x AGE x CULTURE -.091 .636
VD x AGE x CULTURE -.190 .448

ID x AGE x GENDER x CULTURE .013 -.040 .166 .580


ND x AGE x GENDER x CULTURE .116 .661
VD x AGE x GENDER x CULTURE -.263 .452

Notes: 1. ID=Information Design; ND=Navigation Design; VD=Visual Design


2. * p < .1; ** p < .05; *** p < .01

260
Perceptions of Mobile Device Website Design

interface more appealing. Generally speaking, The effect size of independent variables on a
older respondents seem more impressed with dependent variable can be determined by com-
the novelty of the device and its design than the paring the R2 of the dependent variable with and
younger group. This is evidenced in the following without the presence of each independent variable
comment from an older subject: ‘The newness of (Chin, 1998). The calculation for effect size (f2)
it, it’s captivating. Being efficient is fun sometime. is as follows:
Not having to wade through a whole bunch of stuff.
Look at this, look at what it can do, you can do R2 included - R2 excluded
f2 =
it really easily, and you can get the information 1-R 2
included
before everybody else does.’
The effect size of perceived information
Predicting Satisfaction
design, navigation design and visual design on
satisfaction were f2=0.09, f2=0.08 and f2=0.23,
Regression analysis was performed to assess
respectively. Cohen (1988) provides the follow-
the determinants of satisfaction based on main
ing criteria for interpreting effect size: (i) for
effects and interaction effects of the variables in
small effect size, 0.02<f2≤0.15; (ii) for medium
this research. Gender and culture categorization
effect size, 0.15<f2≤0.35; and (iii) for large effect
was straightforward, where dummy variables
size, f2>0.35. Therefore, both information and
represented males/females and Canadian/Chinese
navigation design were shown to have a small
participants. Age was also coded as a dichoto-
effect size on satisfaction, while visual design
mous variable for clarity in presentation (Morris
can be classified as having a medium effect size
et al., 2005). Participants who are 34 years of
on satisfaction.
age or younger were categorized as ‘younger’,
The only interaction effect shown to have a
while those 35 or older were placed in the ‘older’
statistical impact on satisfaction is information
category (as per Aarnio et al., 2002). Morris et
design by age by culture. Not all groups were
al. (2005) suggested that gender differences in
equally satisfied with information provided at the
technology perceptions are more pronounced
site. This is supported in our qualitative analysis.
among older workers, and the interplay between
Canadians were primarily concerned about the
key demographic variables should be examined
limited number of restaurant options (Information
in addition to investigating isolated demographic
Breadth), whereas Chinese sought more detailed
characteristics. This notion is supported by others
information for each of the restaurant choices
in the information systems field (Butler, 2000;
(Information Depth). As one Chinese participant
Venkatesh et al., 2000), as well as the field of
noted, ‘You need to know price of the restaurant,
psychology (Nosek et al., 2002; Kubeck et al.,
the surroundings, and what the location and street
1996). Therefore, in Table 9 the regression analy-
looks like, but maybe I would have to call them
sis examines various combinations of three-way
to get information and details’. In contrast, a Ca-
and four-way interactions between demographic
nadian participant commented: ‘I was expecting
variables (e.g., visual design X gender X age).
there would be a lot of restaurants represented.
In general, results of the regression analysis in
For San Francisco I would be disappointed, I
Table 9 support hypothesis 4 that design elements
wouldn’t trust the source’. This was particularly
do impact satisfaction with mobile services.
evident among the older participants.

261
Perceptions of Mobile Device Website Design

DISCUSSION AND CONCLUSION design, while Chinese focused on ‘Information


Depth’ (more detail about the existing choices).
Egan (1998) makes the case for exploring indi- This finding is in alignment with the qualitative
vidual differences in interface design, and claims results of Choi et al. (2006) who discovered that
‘differences among people usually account for collectivist Koreans and Japanese preferred large
much more variability in performance than differ- amounts of detail on a single screen. With respect
ences in system designs or differences in training to gender and information design, no specific pat-
procedures’ (p. 543). Individual differences may terns emerged for men however women tended
affect what users seek in a system’s interface, and to focus more on ‘Information Depth’, especially
how they interpret such interfaces. Elements of a Chinese women. This finding suggests women
user interface appropriate for one group may not more than men desire detailed information content,
be appropriate for another. perhaps related to different modes of information
For mobile devices, evidence of the impact of processing (Maltby et al., 2003). With respect to
individual differences on design and satisfaction age, younger participants focused on ‘Information
has been preliminary, scattered, and incomplete. Breadth’ while older users desired both breadth
In fact, the majority of previous research tends and depth of information, with more emphasis
to examine the adoption of mobile devices by on depth. Taken collectively, younger Canadian
culture (Harris et al., 2005; Kim et al., 2004; users were most concerned with ‘Information
Carlson et al., 1999), gender (Yang, 2005; Bren- Breadth’ while older Chinese females were most
nen, 2000; Joines et al., 2003; Park & Jun, 2003; focused on ‘Information Depth’. These results
Rohm & Swaminathan, 2004) or age (Wielenmann have implications for mobile interface designers
& Larsson, 2000) rather than based on design who aim to best connect with users.
considerations. On the stationary Internet, there is evidence
In contrast, the current research explores user that preferences for visual design vary by culture
perceptions of mobile design by culture, gender (Cyr, 2008; Cyr & Trevor-Smith, 2004). Users
and age. This has included a focus of information from collectivist cultures such as China desire
design, navigation design and visual design and visuals and aesthetic beauty of the interface, with
how each impacts user satisfaction with an Internet emphasis on “affective” qualities (Cyr et al., 2006).
enabled cellular phone. Our exploratory analysis On a mobile interface Choi et al. (2006) similarly
using both quantitative and qualitative data reveals found that collectivist Koreans and Japanese
some interesting differences for culture and age preferred colorful screen design, while Finns
between user groups related to our three mobile preferred simpler, less colorful screen design.
design dimensions and satisfaction. Equally in- Results from the current investigation parallel
teresting is that there are no gender differences these earlier studies. In the area of visual design,
for the mobile device as tested. Canadians focused on Visual Ease (utilitarian
In the realm of information design, over- elements of design) while Chinese commented
all Canadian and Chinese users felt there was more on the ‘beauty’ of the design. Related to
adequate information presented on the mobile gender, men in both cultures and age groups were
device, but each group was attentive to differ- more concerned with ‘Visual Ease’ while women
ent types of information. Canadians focused on were more interested in ‘Visual Beauty’ and more
‘Information Breadth’ (more choice of alterna- aesthetic elements of the interface.
tive restaurants) and the utility of information

262
Perceptions of Mobile Device Website Design

Almost all participants found the mobile inter- ations themselves may be culturally bound. As
face relatively easy to use and navigate. The sites Yeo (2001) points out, participants from various
are not deep, are laid out in a logical way, and the cultures are prone to provide false statements
joystick is easy to master. Chinese participants had during usability evaluations to allow the designer
little to say about navigation, while Canadians to ‘save face’. Research is needed to develop us-
women and older users generally commented on ability and research methodologies that accurately
challenges of navigation the interface. This finding reflect personal opinions and preferences across
is aligned to other work in which women found cultures. Further, designers tend to ignore the role
stationary websites more difficult to navigate (Cyr culture may play in the design of the interface
& Bonanni, 2005), or where physical limitations (Sheridan, 2001), in particular within the Web
may have implications for interface design with context (Jagne et al., 2005). While many studies
older users (Hawthorn, 2000). have been inconclusive, cultural factors deserve
Consistent with findings for the stationary In- further investigation (Kwon & Chidambaram,
ternet, it is not surprising that in this investigation 2000). At the very least, researchers and design-
design elements impact satisfaction in a mobile ers will ideally seek to better understand design
service context. Further, the current work is con- elements that promote cultural attractiveness.
sistent with Chae et al. (2002) who found that It is noteworthy that only one mobile applica-
information quality of mobile services contributed tion (a restaurant selection) and only one WAP site
to satisfaction. However, Chae et al. examined (lonelyplanet.com) are used in this experiment.
different elements of the mobile interface than The site offered some visual treatment, but has
in our investigation. Collectively, these findings a narrow structure and simple layout. A more
support the assertion that regardless of technology complex site with a deeper hierarchy may reveal
or device, the interface is often considered the more pronounced differences across diverse
most important component of the entire system groups. Future research should consider multiple
to the end-user (Sarker & Wells, 2003), and plays designs across multiple mobile applications. A
an important role in user attitudes (Bidgoli, 1990). positive feature of this investigation is the applica-
As such, further research related to effective tion of design elements as outlined by the design
mobile design has commercial implications for community (Garrett’s 2003 categorizations for
m-commerce and is aligned to investigations of information design, visual design, and navigation
effective website design in e-commerce (Agarwal design) to a mobile treatment. Further, in a nascent
& Venkatesh, 2002; Fogg & Tzeng, 1999; Hui & area of study as represented here, a strong point
Triandis, 1985; Morgan & Hunt, 1994; Schultz, of the investigation is the use of both quantita-
2003). tive and qualitative methodologies. In addition to
This is an initial exploratory study with a surveys, interview data was evaluated using atlas.
limited sample size, representing only two cul- ti software to provide a systematic evaluation of
tures. Although participants are representative of words or phrases into categories relevant to this
the desired cultural groups, they may not fully investigation. The use of atlas.ti afforded deeper
represent the socio-economic group within their insights into the user experience than survey data
country. It is recommended that follow-up studies alone can provide.
draw samples from larger populations, and from This research is an important first step in un-
additional culture groups. Also, usability evalu- derstanding the impact of individual differences

263
Perceptions of Mobile Device Website Design

on the design and satisfaction of mobile services. Becker, S. A. (2002). An exploratory study on
This relatively unexplored area is worthy of future web usability and the internationalization of US
attention. The quest for further knowledge as to e-businesses. Journal of Electronic Commerce
how culture, gender and age impact technology Research, 3, 265–278.
use and satisfaction implies appreciation that
Bernard, M. (2002). Criteria for optimal web
these relationships are dynamic, and subject to
design (Designing for usability).
continuous transformation (Simon, 2001). En-
hanced understanding of subtle dimensions related Bhattacherjee, A. (2001). Understanding infor-
to specific user groups will eventually enable mation systems continuance: An expectation-
fulfillment of expectations for optimal anytime confirmation model. MIS Quarterly, 2, 351–369.
and anywhere services. doi:10.2307/3250921
Bidgoli, H. (1990). Designing a user-friendly in-
terface for a decision support system. Information
REFERENCES
Technology, 12, 148–154.
Aarnio, E. A., Heikkila, J., & Hirvola, S. (2002). Bigne, E., Ruiz, C., & Sanz, S. (2005). The impact
Adoption and use of mobile services empirical of internet user shopping patterns and demograph-
evidence from a Finnish survey. In Proceedings ics on consumer mobile buying behavior. Journal
of 35th International Conference of System Sci- of Electronic Commerce Research, 6, 193–209.
ences, Hawaii, USA.
Brennan, M. (2000). Dot-com flavor of the week
Agarwal, R., & Venkatesh, V. (2002). Assessing a or hand-held revolution?
firm’s web presence: A heuristic evaluation pro-
cedure for measurement of usability. Information Brown, K. (2001). Using computers to de-
Management Research, 13, 168–121. liver training: Which employees learn and
why? Personnel Psychology, 54, 271–296.
Anckar, B., & D’Incau, D. (2002). Value creation doi:10.1111/j.1744-6570.2001.tb00093.x
in mobile commerce: Findings from a consumer
survey. Journal of Information Technology Theory Butler, D. (2000). Gender, girls, and computer
& Application, 4, 43–64. technology: What’s the status now? Clearing
House (Menasha, Wis.), 73, 225–229.
Anderson, R. E., & Srinivasan, S. S. (2002). E-
satisfaction and e-loyalty: A contingency frame- Carlson, P. J., Kahn, B. K., & Rowe, F. (1999).
work. Psychology and Marketing, 20, 123–138. Organisational impacts of new communication
doi:10.1002/mar.10063 technology: A comparison of cellular phone
adoption in France and the United States. Jour-
Balasubramanian, S., Konana, P., & Menon, nal of Global Information Management, (July):
N. (2003). Customer satisfaction in virtual en- 19–29.
vironments: A study of online investing. Man-
agement Science, 49, 871–889. doi:10.1287/ Chae, M., & Kim, J. (2004). Do size and structure
mnsc.49.7.871.16385 matter to mobile users? An empirical study of the
effects of screen size, information structure, and
Barber, W., & Badre, A. N. (2001). Culturability: task complexity on user activities with standard
The merging of culture and usability. In Proceed- web phones. Behaviour & Information Technol-
ings of 4th Conference on Human Factors and the ogy, 23, 165–181. doi:10.1080/0144929041000
Web. New Jersey: Basking Ridge. 1669923

264
Perceptions of Mobile Device Website Design

Chae, M., Kim, J., & Ryu, H. (2002). In- Coursaris, C., Hassanein, K., & Head, M. (2003).
formation quality for mobile internet ser- Understanding the mobile consumer. In S. Nansi
vices: A theoretical model with empirical (Ed.), Wireless communications and mobile com-
validation. Electronic Markets, 12, 38–46. merce (pp. 132–165). Idea Group Inc.
doi:10.1080/101967802753433254
Cronbach, L. J. (1971). Test validation in educa-
Chau, P. K., Au, G., & Tam, K. Y. (2000). Impact tional measurement. In R. L. Thorndike, (Ed.), 2nd
of information presentation modes on online ed. (pp. 443–507). Washington, D.C: American
shopping: An empirical evaluation of a broadband Council on Education.
interactive shopping service. Journal of Organiza-
Crosby, L. B., DeVito, R., & Pearson, J. M. (2003).
tional Computing and Electronic Commerce, 10,
Manage your customers’ perception of quality.
1–22. doi:10.1207/S15327744JOCE100101
Review of Business, 24.
Chen, S. C., & Dhillon, G. S. (2003). Interpreting
Cyr, D. (2008). Modelling website design across
dimensions of consumer trust in e-commerce.
cultures: Relationships to trust, satisfaction and
Information Technology and Management, 4,
e-loyalty. Journal of Management Information
303–318. doi:10.1023/A:1022962631249
Systems, 24(4), 47–72. doi:10.2753/MIS0742-
Childers, T., Carr, C., Peck, J., & Carson, S. (2001). 1222240402
Hedonic and utilitarian motivations for online
Cyr, D., & Bonanni, C. (2005). Gender and web-
retail shopping behavior. Journal of Retailing, 77,
site design in e-business. International Journal
511–535. doi:10.1016/S0022-4359(01)00056-2
of Electronic Business, 6, 565–582. doi:10.1504/
Chin, W. W. (1998). The partial least squares ap- IJEB.2005.008536
proach to structural equation modeling. In G.A.
Cyr, D., Bonanni, C., Bowes, J., & Ilsever, J.
Marcoulides (Ed.), Modern methods for business
(2005). Beyond trust: Website design preferences
research (pp. 295–336). Mahwah: Lawrence Erl-
across cultures. Journal of Global Information
baum Associates.
Management, 13, 24–52.
Choi, B., Lee, I., & Kim, J. (2006). Culturability
Cyr, D., Bonanni, C., & Ilsever, J. (2004). De-
in mobile data services: A qualitative study of the
sign and e-loyalty across cultures in electronic
relationship between cultural characteristics and
commerce. In Proceedings of 6th International
user-experience attributes. International Journal
Conference on Electronic Commerce.
of Human-Computer Interaction, 20, 171–206.
doi:10.1207/s15327590ijhc2003_2 Cyr, D., Head, J., Larios, H., & Pan, B. (2006).
Exploring human images in website design across
Choi, B., Lee, I., Kim, J., & Jeon, Y. (2005). A
cultures: A multi-method approach. In Proceed-
qualitative cross-national study of cultural influ-
ings for the Fifth Pre-ICIS HCI Research in MIS
ences on mobile data service design. In Proceed-
Workshop 2006, Milwaukee, Wisconsin. Nomi-
ings of Computer-Human Interaction, Portland,
nated for Best Paper Award.
Oregon, USA, 2-7 April.
Cohen, J. (1988). Statistical power analysis for
the behavioral sciences. 2nd ed. New York: Aca-
demic Press.

265
Perceptions of Mobile Device Website Design

Cyr, D., & Trevor-Smith, H. (2004). Localization Ellis, R. D., & Kurniawan, S. H. (2000). Increasing
of web design: An empirical comparison of Ger- the usability of online information for older users:
man, Japanese, and U.S. website characteristics. A case study in participatory design. International
Journal of the American Society for Information Journal of Human-Computer Interaction, 12,
Science and Technology, 55, 1–10. doi:10.1002/ 263–276. doi:10.1207/S15327590IJHC1202_6
asi.20075
Evers, V., & Day, D. (1997). The role of culture in
Deaux, K., & Kite, M. E. (1987). Thinking about interface acceptance. In S. Howard, J. Hammond,
gender. In B. B. Hess & M. M. Ferree (Eds.), & G. Lindegaard (Eds.), Proceedings of Human
Analyzing gender: A handbook of social science Computer Interaction, INTERACT ’97. London:
research (pp. 92–117). Newbury Park, CA: Sage Chapman and Hall.
Publications.
Farkas, K., & Farkas, J. B. (2000). Guidelines for
Dillon, R. J., & McKnight, C. (1990). The designing web navigation. Technical Communica-
effect of display size and text splitting on tion, Third Quarter, (pp. 341–358).
reading lengthy text from the screen. Behav-
Flavián, C., Guinalíu, M., & Gurrea, R. (2005).
iour & Information Technology, 9, 215–227.
The role played by perceived usability, satisfaction
doi:10.1080/01449299008924238
and consumer trust on website loyalty. Informa-
Doll, W. J., & Torkzadesh, G. (1988). The mea- tion & Management, 42, 719–729. doi:10.1016/j.
surement of end-user computing satisfaction. MIS im.2004.07.001
Quarterly, 12, 259–274. doi:10.2307/248851
Fogg, B. J., Soohoo, C., & Danielson, D. (2002).
Doney, P. M., Cannon, J. P., & Mullen, M. R. (1998). How people evaluate a web site’s credibility? Re-
Understanding the influence of national culture on sults from a large study. Persuasive Technology
the development of trust. Academy of Management Lab, Stanford University.
Review, 23, 601–620. doi:10.2307/259297
Fogg, J., & Tseng, S. (1999). Credibility and
Duchnicky, R. L., & Kolers, P. A. (1983). Read- computing technology. Communications of the
ability of text scrolled on visual display terminals ACM, 14, 39–87.
as a function of window size. Human Factors,
Fornell, C., & Larcker, D. F. (1981). Evaluat-
25, 683–692.
ing structural equation models with unob-
Egan, D. E. (1998). Individual differences in served variables and measurement error. JMR,
human-computer interaction. In M. Helander Journal of Marketing Research, 18, 39–50.
(Ed.), Handbook of human-computer Interaction doi:10.2307/3151312
(pp. 543–568). Amsterdam: Elsevier Science
Garrett, J. J. (2003). The Elements of user experi-
Publishers.
ence: User-centered design for the web. Indiana,
Egger, F. N. (2001). Affective design of e-com- USA: New Riders Publications.
merce user interfaces: How to maximize perceived
Gefen, D., & Straub, D. W. (1997). Gender dif-
trustworthiness. In Proceedings of International
ferences in the perception and use of email: An
Conference of Affective Human Factors Design.
extension to the technology acceptance model. MIS
London Press.
Quarterly, 21, 389–400. doi:10.2307/249720

266
Perceptions of Mobile Device Website Design

Gommans, M., Krishan, K. S., & Scheddold, K. Hu, J., Shima, K., Oehlmann, R., Zhao, J., Take-
B. (2001). From brand loyalty to e-loyalty: A mura, Y., & Matsumoto, K. (2004). An empirical
conceptual framework. Journal of Economic and study of audience impressions of B2C web pages
Social Research, 3, 43–58. in Japan, China and the U.K. Electronic Com-
merce Research and Applications, 3(2), 176–189.
Goodman, J., Brewser, S. A., & Gray, P. (2005).
doi:10.1016/j.elerap.2003.09.004
How can we best use landmarks to support older
people in navigation? Behaviour & Information Hui, C. H., & Triandis, H. C. (1985). Mea-
Technology, 24, 3–20. doi:10.1080/0144929051 surement in the cross-cultural psychology: A
2331319021 review and comparison of strategies. Journal
of Cross-Cultural Psychology, 16, 131–152.
Goodman, J., & Lundell, J. (2005). HCI and
doi:10.1177/0022002185016002001
the older population [Editorial]. Interacting
with Computers, 17, 613–620. doi:10.1016/j. Jagne, J., Smith-Atatkan, S., Duncker, E., & Cur-
intcom.2005.09.001 zon, P. (2005). Cross-cultural factors of physical
and e-shopping. In Proceedings of the Eleventh
Hair, F., Anderson, R. E., Tathan, R. L., & Black,
International Conference on Human Computer
W. C. (1998). Multivariate data analysis. Engle-
Interaction.
wood Cliffs, N.J.: Prentice Hall.
Jiang, Z., & Benbasat, I. (2003). The effects of
Harris, P., Rettie, R., & Kwan, C. C. (2005). Adop-
interactivity and vividness of functional control in
tion and usage of m-commerce: A cross-cultural
changing web consumers’ attitudes. In Proceed-
comparison of Hong Kong and the United King-
ings 24th International Conference of Information
dom. Journal of Electronic Commerce Research,
Systems, Seattle, USA.
6, 210–224.
Joines, J. L., Scherer, C., & Scheufele, D. (2003).
Hawthorn, D. (2000). Possible implications
Exploring motivations for consumer web use and
of aging for interface design. Interacting with
their implications for e-commerce. Journal of
Computers, 12, 507–528. doi:10.1016/S0953-
Consumer Marketing, 20.
5438(99)00021-1
Jones, M. Y., Stanaland, A. J., & Gelb, B. D. (1998).
Hoffman, D. L., & Novak, T. P. (1996). Marketing
Beefcake and cheesecake: Insights for advertisers.
in hypermedia computer-mediated environments:
Journal of Advertising, Summer.
Conceptual foundations. Journal of Marketing,
60, 50–68. doi:10.2307/1251841 Kateranttanakul, P., & Siau, K. (1999). Measuring
information quality of web sites: Development of
Hofstede, G. (1980). Culture’s consequences:
an instrument. In Proceedings of International
Comparing values, behaviors, institutions and
Conference of Information Systems, Charlotte,
organisations across nations, 2nd ed. Thousand
North Carolina (pp. 279–285).
Oaks, CA: Sage Publication.
Kelley, C. L., & Charness, N. (1995). Issues in
Holland, J., & Baker, S. M. (2001). Customer
training older adults to use computers. Behav-
participation in creating site brand loyalty. Journal
iour & Information Technology, 14, 107–120.
of Interactive Marketing, 15, 34–45. doi:10.1002/
doi:10.1080/01449299508914630
dir.1021

267
Perceptions of Mobile Device Website Design

Kerlinger, F. N. (1964). Foundations in behav- Lohse, G. L., & Spiller, P. (1998). Electronic shop-
ioral research. New York: Holt, Rinehart, and ping. Communications of the ACM, 41, 81–88.
Winston. doi:10.1145/278476.278491
Kim, K., Kim, J., Lee, Y., Chae, M., & Choi, Y. Lohse, G. L., & Spiller, P. (1999). Internet retail
(2002). An empirical study of the use contexts and store design: How the user interface influences
usability problems in mobile Internet. In Proceed- traffic and sales. Journal for Computed-Mediated
ings of 35th Annual Hawaii International Confer- Communication, 5.
ence on System Sciences, Los Alamitos, CA.
MACRO. (2004). Mobile phone usage among the
Kim, M. K., Park, C., & Jeong, D. H. (2004). The teenagers and youth in Mumbai. Report.
effects of customer satisfaction and switching bar-
Maltby, Chudry, F., & Wedande, G. (2003). Cyber
rier on customer loyalty in Korean mobile telecom-
dudes and cyber babes: Gender differences and
munication services. Telecommunications Policy,
internet financial services. Journal of Financial
28, 145–159. doi:10.1016/j.telpol.2003.12.003
Services Marketing, 8, 152–165. doi:10.1057/
Knight, M., & Pearson, J. (2005). The changing palgrave.fsm.4770115
demographics: The diminishing role of age and
Marcus, A., & Gould, E. W. (2000). Cultural di-
gender in computer usage. Journal of Organiza-
mensions and global web user interface design.
tional and End User Computing, 17, 49–65.
Interaction, (July/August): 33–46.
Koufaris, M. (2002). Applying the technology
McKinney, V., Yoon, K., & Zahedi, F. M. (2002).
acceptance model and flow theory to online con-
The measurement of web-customer satisfaction:
sumer behaviour. Information Systems Research,
An expectation and disconfirmation approach.
13, 205–222. doi:10.1287/isre.13.2.205.83
Information Systems Research, 13, 296–315.
Kubeck, E., Delp, N. D., Haslett, T. K., & McDan- doi:10.1287/isre.13.3.296.76
iel, M. A. (1996). Does job-related training perfor-
Morgan, R., & Hunt, S. D. (1994). The
mance decline with age? Psychology and Aging,
Commitment-trust theory of relationship
11, 92–107. doi:10.1037/0882-7974.11.1.92
marketing. Journal of Marketing, 58, 20–38.
Kumar, S., & Zahn, C. (2003). Mobile communi- doi:10.2307/1252308
cations: Evolution and impact on business opera-
Morris, G. M., Venkatesh, V., & Ackerman, P. L.
tions. Technovation, 23, 515–520. doi:10.1016/
(2005). Gender and age differences in employee
S0166-4972(02)00120-7
decisions about new technology: An extension to
Kwon, H. S., & Chidambaram, L. (2000). A test the theory of planned behavior. IEEE Transac-
of the technology acceptance model: The case tions on Engineering Management, 52, 69–84.
of cellular phone adoption. In Proceedings of doi:10.1109/TEM.2004.839967
33rd Hawaii International Conference on System
Nielsen, J. (2001). Designing for web usability.
Sciences.
Indianapolis: New Riders Publications.
Lee, Y., Kim, J., Lee, I., & Kim, H. (2002). A
Nielsen, J., & Del Galdo, E. M. (1996). Interna-
cross-cultural study on the value structure of
tional User Interfaces. Wiley Computer Publish-
mobile internet usage: Comparison between Ko-
ing, John Wiley & Sons, New York.
rea and Japan. Journal of Electronic Commerce
Research, 3, 227–239.

268
Perceptions of Mobile Device Website Design

Nosek, B. A., Banaji, M. R., & Greenwalk, A. G. Perry, M., O’Hara, K., Sellen, A., Harper, R.,
(2002). Harvesting implicit group attitudes and be- & Brown, B. A. T. (2001). Dealing with mobil-
liefs from a demonstration website. Group Dynam- ity: Understanding access anytime, anywhere.
ics, 6, 101–115. doi:10.1037/1089-2699.6.1.101 Transactions on Computer-Human Interaction,
8, 323–347. doi:10.1145/504704.504707
Nysveen, H., Pedersen, P. E., & Thorbjørnsen, H.
(2005). Intentions to use mobile services: Ante- Pitt, L. F., Watson, R. T., & Kavan, C. B. (1995).
cedents and cross-service comparisons. Journal of Service quality: A measure of information sys-
the Academy of Marketing Science, 33, 330–346. tems effectiveness. MIS Quarterly, 19, 173–187.
doi:10.1177/0092070305276149 doi:10.2307/249687
Oliver, R. L. (1980). A cognitive model of the Putrevu, S. (2001). Exploring the origins and
antecedents and consequences of satisfaction information processing differences between men
decisions. JMR, Journal of Marketing Research, and women: Implications for advertisers. Academy
17, 460–469. doi:10.2307/3150499 of Marketing Science Review, 10.
Oliver, R. L. (1999). Whence consumer Rainer, R. K., Laosethakul, K., & Astone, M.
loyalty? Journal of Marketing, 63, 33–44. (2003). Are gender perceptions of computing
doi:10.2307/1252099 changing over time? Journal of Computer Infor-
mation Systems, 43, 108–114.
Palmer, J. W. (2002). Website usability, de-
sign, and performance metrics. Information Rau, P. P., Chen, J., & Chen, D. (2006). A study of
Systems Research, 13, 151–167. doi:10.1287/ presentations of mobile web banners for location-
isre.13.2.151.88 based information and entertainment information
websites. Behaviour & Information Technology,
Parasuraman, A., Zeithaml, V. A., & Berry, L. L.
25, 253–261. doi:10.1080/01449290500222009
(1988). SERVQUAL: A multiple-item scale for
measuring customer perceptions of service quality. Resiel, J. F., & Shneiderman, B. (1987). Is bigger
Journal of Retailing, 64, 12–40. better? The effects of display size on program
reading. In G. Salvendy (Ed.), Social, ergonomic
Park, C., & Jun, J. K. (2003). A cross-cultural
and stress aspects of work with computers (pp.
comparison of internet buying behavior: Effects
113–122). Elsevier Science Publishers.
of internet usage, perceived risks, and innova-
tiveness. International Marketing Review, 20, Rivard, S., & Huff, S. (1988). Factors of success
534–553. doi:10.1108/02651330310498771 for end user computing. Communications of the
ACM, 31, 552–561. doi:10.1145/42411.42418
Pearson, J. M., Crosby, L., Bahmanziari, T., &
Conrad, E. (2003). An empirical investigation Rodgers, S., & Harris, M. A. (2003). Gender and
into the relationship between organisational cul- e-commerce: An exploratory study. Journal of
ture and computer efficacy as moderated by age Advertising Research, 43, 322–329.
and gender. Journal of Computer Information
Rohm, J., & Swaminathan, V. (2004). A typology
Systems, 43, 58–70.
of online shoppers based on shopping motiva-
tions. Journal of Business Research, 57, 748–757.
doi:10.1016/S0148-2963(02)00351-X

269
Perceptions of Mobile Device Website Design

Rosen, D. E., & Purinton, E. (2004). Website de- Simon, S. J. (2001). The impact of culture and
sign: Viewing the web as a cognitive landscape. gender on web sites: An empirical study. The
Journal of Business Research, 57, 787–794. Data Base for Advances in Information Systems,
doi:10.1016/S0148-2963(02)00353-3 32, 18–37.
Salthouse, T. A. (1992). Reasoning and spatial Singh, N., Xhao, H., & Hu, X. (2003). Cultural
abilities. In F. I. M. Craik & T. A. Salthouse adaptation on the web: A study of American com-
(Eds.), The handbook of aging and cognition, panies’ domestic and Chinese websites. Journal of
3rd ed. (pp. 167–212). Hillsdale, NJ: Lawrence Global Information Management, 11, 63–80.
Erlbaum Associates.
Straub, D. W. (1989). Validating instruments
Sanchez-Franco, M. J. (2006). Exploring the influ- in MIS research. MIS Quarterly, 12, 147–170.
ence of gender on the web usage via partial least- doi:10.2307/248922
squares. Behaviour & Information Technology, 25,
Straub, D. W., Loch, W., Aristo, R., Karahanna,
19–36. doi:10.1080/01449290500124536
E., & Strite, M. (2002). Toward a theory-based
Sarker, S., & Wells, J. D. (2003). Understand- measurement of culture. Journal of Global Infor-
ing mobile handheld device use and adop- mation Management, 10, 13–23.
tion. Communications of the ACM, 46, 35–40.
Stroetmann, V. N., Husing, T., Kubitschke, L., &
doi:10.1145/953460.953484
Stroetmann, K. A. (2002). The attitudes, expecta-
Schmidt, S. H., & Frick, O. (2000). WAP – de- tions and needs of elderly people in relation to
signing for small user interfaces. In Proceedings e-health applications: Results from a European
of Conference on Human Factors in Computing survey. Journal of Telemedicine and Telecare, 8,
Systems. 82–84. doi:10.1258/135763302320302154
Schrott, G., & Gluckler, J. (2004). What makes Sun, H. (2001). Building a culturally-competent
mobile computer supported cooperative work corporate web site: An explanatory study of
mobile? Towards a better understanding of co- cultural markers in multilingual web design. In
operative mobile interactions. International Jour- Proceedings of SIGDOC’01, 21-24, October (pp.
nal of Human-Computer Studies, 60, 737–752. 95–102).
doi:10.1016/j.ijhcs.2003.11.006
Szymanski, D. A., & Hise, R. T. (2000). E-
Schultz, L. (2003). Effects of graphical elements satisfaction: An initial examination. Journal of
on perceived usefulness of a library. Masters Retailing, 76, 309–322. doi:10.1016/S0022-
thesis. 4359(00)00035-X
Sheridan, E. F. (2001). Cross-cultural web site Tarasewich, P. (2003). Designing mobile com-
design: Considerations for developing and strate- merce applications. Communications of the ACM,
gies or validating local appropriate online content. 46, 57–60. doi:10.1145/953460.953489
MultiLingual Computing, 12.
Teo, H., Oh, L., Liu, C., & Wei, K. (2003). An
Simmers, S., & Anandarajan, M. (2001). User empirical study of the effects of interactivity on
satisfaction in the internet-anchored workplace: web user attitude. International Journal of Human-
An exploratory study. Journal of Information Computer Studies, 58, 281–305. doi:10.1016/
Technology Theory and Application, 3, 39–61. S1071-5819(03)00008-9

270
Perceptions of Mobile Device Website Design

Teo, T., Lim, V., & Lai, R. (1999). Intrinsic and York, J., & Pendharkar, P. (2004). Human-com-
extrinsic motivation in internet usage. Omega puter interaction issues for mobile computing in
International Journal of Management Science, 27, a variable work context. International Journal
25–37. doi:10.1016/S0305-0483(98)00028-0 of Human-Computer Studies, 60, 771–797.
doi:10.1016/j.ijhcs.2003.07.004
Tractinsky, N. (1997). Aesthetics and apparent
usability: Empirically assessing cultural and Zhang, X., Keeling, K. K., & Pavur, R. J. (2000).
methodological issues. In CHI Proceedings. Information quality of commercial web site
homepages: An explorative analysis. In Proceed-
van der Heijden. (2003). Factors influencing the
ings of International Conference on Information
usage of websites: The case of a generic portal in
Systems, Brisbane, Australia (pp. 164–175).
the Netherlands. Information & Management, 40,
541–549. doi:10.1016/S0378-7206(02)00079-4
Venkatesh, V., Morris, G. M., & Ackerman, P.
ENDNOTES
L. (2000). A longitudinal field investigation of
gender differences in individual technology adop- 1
Throughout this paper the terms ‘portable’,
tion decision-making processes. Organizational ‘handheld’, ‘mobile’ and ‘small-screen’
Behavior and Human Decision Processes, 83, refer to essentially the same type of device,
33–60. doi:10.1006/obhd.2000.2896 typically called a ‘smart phone’. Also, the
Venkatesh, V., Ramesh, V., & Massey, A. P. term ‘mobile Internet’ refers to accessing the
(2003). Understanding usability in mobile com- Web directly on a small-screen, handheld,
merce. Communications of the ACM, 46, 53–56. or mobile device that may be connected to
doi:10.1145/953460.953488 the Internet.
2
The work by Geert Hofstede (1980) outlines
Weilenmann, A., & Larsson, C. (2000). Collab- 5 cultural dimensions: (1) Power distance -
orative use of mobile telephones: A field study of extent to which a society accepts unequal
Swedish teenagers. In Proceedings of NordiCHI distributions of power in organizations and
2000, Stockholm, Sweden, 23–25 October. institutions. (2) Uncertainty avoidance - how
Yang, K. (2005). Exploring factors affecting the societies accommodate high levels of uncer-
adoption of mobile commerce in Singapore. Telem- tainty and ambiguity in the environment.
atics and Informatics, 22, 257–277. doi:10.1016/j. (3) Masculinity-Femininity - in feminine
tele.2004.11.003 societies there is an emphasis on quality of
life and relationships; cultures that focus
Yeo, A. (2001). Global software development on material success and assertiveness are
lifecycle: An exploratory study. In CHI 2001 considered more masculine in orientation.
Conference on Human Factors in Computing (4) Individualism-Collectivism - in an indi-
Systems (pp. 104–111). ACM Press. vidualist society individuals are expected to
Yoon, S. (2002). The antecedents and conse- consider personal interests over interests of
quences of trust in online-purchase decisions. the group and individual decision-making is
Journal of Interactive Marketing, 16, 47–63. valued; in a collectivist culture the good of
doi:10.1002/dir.10008 the group is more likely to be considered.

271
Perceptions of Mobile Device Website Design

(5) Time Orientation - whether the focus in


on short-term vs. long-term considerations.
For a further elaboration of Hofstede’s cul-
tural dimensions, refer to Hofstede (1980)
or Simon (2001).

272
Perceptions of Mobile Device Website Design

Appendix A. Screen Shots

Figure 2.

273
Perceptions of Mobile Device Website Design

Figure 3.

274
Perceptions of Mobile Device Website Design

Appendix B. Design Survey

Following are the statement used in the survey. Each was answered on a 7-point Likert scale from strongly
disagree to strongly agree.Information Design [Sources: Cyr et al. 2005; 2004]

ID-1: I find the information logically presented.

ID-2: All service options, service attributes and restaurant information are well presented.

ID-3: I find the information to be well organized.

ID-4: The presentation of information is effective.Navigation Design [Sources: Cyr et al. 2005; 2004]

ND-1: This browser provides good navigation facilities to information content.

ND-2: I can easily navigate the CityGuide site.

ND-3: I find the CityGuide site easy to use.Visual Design [Sources: Cyr et al. 2005; 2004; van der
Heijden 2003]

VD-1: The screen design (i.e. colors, boxes, menus, etc) is attractive.

VD-2: This site looks professionally designed.

VD-3: The graphics are meaningful.

VD-4: The overall look and feel of the site is visually appealing.Satisfaction [Sources: Cyr et al. 2005;
2004]

S-1: This site appeals to me emotionally.

S-2: This service completely fulfills my expectations.

S-3: This service satisfies my needs well.

S-4: Using this service is satisfactory overall.

275
Perceptions of Mobile Device Website Design

Appendix C: Atlas.ti Concept Map

Note: C_M = Canadian male, C_F = Canadian female, CH_M


= Chinese male, CH_F = Chinese female.

Figure 4.

This work was previously published in Mobile and Ubiquitous Commerce: Advanced E-Business Methods, edited by M. Head;
E. Li, pp. 173-200, copyright 2009 by Information Science Reference (an imprint of IGI Global)

276
277

Chapter 2.2
Paralingual Web Design and
Trust in E-Government
Roy H. Segovia
San Diego State University, USA

Murray E. Jennex
San Diego State University, USA

James Beatty
San Diego State University, USA

ABSTRACT multilingual populations, where issues of trust


among minority speakers and majority speakers
Can Web design improve the way governments may exist, as they indicate that paralingual Web
serve their constituencies through the use of in- design can help reduce these trust issues.
formation technology [i.e., e-government]? This
article proposes that the use of paralingual Web
design can overcome possible trust issues in e- INTRODUCTION
government with bilingual populations. An exper-
iment was conducted where active e-government Electronic government (e-government) is the use
Web pages were converted to paralingual format of Information and Communication Technologies
and then site visitors were surveyed regarding their (ICT), including the Internet, by government
trust in the content and readability. The results organizations to facilitate providing information
of the experiment show that trust was improved and services to their constituents. E-government
for the minority language speakers, while the Web sites provide everything from basic infor-
majority language speakers remained neutral mation about governmental bodies and issues,
with neither group indicating significant decrease to online services such as registering vehicles
in readability. These findings have important and applying for employment and for permits.
implications for societies with large bilingual or More recent e-government services include e-

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Paralingual Web Design and Trust in E-Government

consultation, which is citizen participation and to fit the constituent market and in the case of
response to forthcoming consultations and deci- a constituent market that speaks more than one
sions on matters of public interest (Jadu, 2005). language, allowing for these multiple languages.
The impetus to implement e-government can be The concern driving this experiment is that there
attributed to cost control and improved service to may be a trust issue affecting the success/adop-
citizens. Another driver is government’s growing tion of e-government should these systems fail to
awareness of the need to attain more democratic take into account the bi or multi lingual aspects
governance (Coleman and Gotze, 2001; OECD, of their constituents.
2001), coupled with a widespread public inter- A premise of information systems, IS, is that
est in the potential of ICT to empower citizens for an IS to be successful the intended system
and to increase government accountability and users must “use” the system where Rai et al.
transparency (Hart-Teeter, 2003). An example (2001) consider “use” to be the consumption of
is the United States E-Government initiative the outputs of the IS by the users as measured in
targets use of improved Internet-based technol- terms such as frequency of use, amount of time
ogy to make it easier for citizens and businesses of use, numbers of access to the IS, usage pattern,
to interact with the government, save taxpayer etc. General thinking is that the more an IS is
dollars, and streamline citizen-to-government used, the more successful the IS. Two of the more
communications (USOMB, 2005). These many widely accepted IS success/acceptance models,
drivers make it likely that e-government will be the DeLone and McLean (1992 and 2003) IS
a lasting ICT application leading e-government Success Model and the Davis (1989) Technology
system designers to look for tools and methodolo- Acceptance Model, TAM, incorporate “use” as
gies that will ensure their acceptance and use by a measure of success (DeLone and McLean) or
the intended users. successful adoption (TAM) through constructs
This article introduces one such potential de- such as Intent to Use, Perceived Usefulness, and
sign methodology, paralingual Web design, and Perceived Ease of Use.
uses an experiment to test this design methodol- Several authors (Gefen, et al., 2002; Tan, et
ogy to see if it has potential for improving system al., 2005, Tan, et al., 2008, Warkentin, et al.,
acceptance and success. Paralingual is a Web 2002) suggest that use of e-government is influ-
design methodology for presenting information in enced by the trust that potential users have with
more than one language. Paralingual Web design e-government. This article hypothesizes that this
involves placing content in the desired languages trust in e-government, and thus subsequent use,
but instead of having separate pages for each lan- can be increased in bi and multi lingual societies
guage as is common in a bi or multilingual Web by using paralingual Web design. This allows
design, the bi or multilingual content is placed readers who are bilingual to easily see both ver-
side by side on the same page. The inspiration sions and readily determine if the same informa-
for this article is the trend towards localization tion is being said in each version. It is expected
in ecommerce systems and a concern that there that trust will be increased through this citizen
may be a localization issue for e-government validation process.
when the target population is bi or multilingual. The contribution of this research is showing
Localization is defined by the Localization In- designers of e-government Web pages how the
dustry Standards Association (LISA, 2008) as the Web pages can be designed to improve trust in a
process of modifying specific products or services bi-lingual constituency. While this research did
for specific markets. In the case of e-government not test this design approach in a multi-lingual
this involves tailoring e-government Web sites environment it is expected this design can also

278
Paralingual Web Design and Trust in E-Government

be applied to e-government Web pages for these government and elected officials. Furthermore,
constituencies. this mistrust was more intense and persistent
than expected, outstripping the levels that have
been measured by polls and focus groups (Rosell,
LITERATURE REVIEW Gantwerk, and Furth, 2005).
In addition to the trust issues above, there are
This article draws from three main bodies of known issues with trust in e-government Web
literature, the trust, paralingual, and IS accep- sites. This is clearly the effect of the general mis-
tance/success literatures. These literatures are trust by citizens in their government bodies, as
summarized below and provide the theoretical mentioned previously. The principal reason given
foundation for the article. The trust literature is for mistrust of the Web is an artifact of the internet
presented first as it provides the issue of concern itself. Namely, the internet is now perceived to
for the article. The paralingual literature is second be beyond the control of the hosts and providers
to provide the background for why the proposed in terms of security and trust. Despite the use of
design methodology is a good solution for the lock icons, digital signatures, passwords, privacy
trust issue. The IS acceptance/success literature policy statements, and other security techniques,
is provided third as it helps provide the framing internet users feel that hosts and providers have
for the experiment. lost control of the digital data transport medium
Trust can be defined as “the subjective as- as well as the software infrastructure that sup-
sessment of one party [trustor] that another party ports it, impeding the growth of e-government
[trustee] will perform a particular transaction (Mercuri, 2005.) To counter this, the International
according to his or her confident expectations, Telecommunication Union (ITU) is providing sup-
in an environment characterized by uncertainty” port for national e-government projects including
(Ba and Pavlou, 2002, p. 245.) For e-government enhancing security and trust in the use of public
this means users trusting that the e-government networks (Khalil-babnet, 2005).
service is providing correct information, that Improving trust in government and e-govern-
data will be protected, and that transactions will ment is a critical issue. In the study by Viewpoint
be conducted in a secure manner and recorded Learning, citizens voiced a strong desire to find
appropriately. constructive solutions to problems facing the state
Trust in government has historically been (Rosell, Gantwerk, and Furth, 2005). In a geo-
problematic in the United States of America as graphical area with a high proportion of bilingual
constituent citizens are known to have a high speakers, usage of e-government Web sites may
level of distrust in their governing bodies. Trust in be improved in the same way as has been shown
government has been declining for more than three effective in electronic commerce (ecommerce).
decades and has been the topic of a substantial That is, with regard to language issues, research-
amount of research in political science (Levi and ers have found that customers are far more likely
Stoker, 2000; Hibbing and Theiss-Morse, 2002). In to buy products and services from Web sites in
the state of California, a recent study exposed an their own language, even if they can read English
unexpectedly high level of distrust in government well. Furthermore, attention to site visitors’ needs
by California citizens. During 2004, a series of should be an important consideration in Web de-
dialog-oriented seminars were held by Viewpoint sign because such attention can help a site build
Learning in various locations in California. One trust with customers (Schneider, 2003). Gassert
of the seven major findings of the study was that (2004) suggests building trust through knowledge
an underlying issue was profound mistrust of by using ICT for better education and informa-

279
Paralingual Web Design and Trust in E-Government

tion. Additionally, LaVoy (2001) supports the use 2006.) There are similar requirements regarding
of e-government as a way of improving trust by Minimum Wage, Pregnancy Disability Leave and
improving accountability. Finally, Gefen, et al. the California Family Rights Act (OSHA4LESS.
(2002) view trust in government as the main driver COM, 2006)
for e-government adoption. Their analysis show The term “paralingual” is used to define the
data privacy concerns create the biggest barrier layout of information using two sets of text in
to adoption of e-government. While this form of different languages on the same page, such as in
e-government service, online tax service, consists a Web page. The term was coined as an extension
of real transactions, the trust issue dealt with in of the word “bilingual.” Para is a Greek prefix
this research comes even before citizens attempt that means beside, near, or alongside (Wikipedia,
such transactions. Namely, the citizens must be 2006). Therefore, paralingual refers to two lan-
given a reason to simply trust in the information guages adjacent to each other on the same page.
that is on a government Web site. Paralingual Web pages are almost non-existent
One of the earliest known examples of written on the Web. Although many Web sites are now
multi-language information is the Rosetta Stone. multi-language Web sites, the common layout for
According to Wikipedia (2006a), this archaeologi- these Web sites is to separate the languages to
cal artifact is a granite stone with writing in three separate pages. This commonality is reflected in
different written scripts dated to about 200 BCE the standards on localization (LISA, 2008.). This
(Before Common Era, essentially the same as localization can be found in e-government. Cun-
BC). It contains Greek, Demotic Egyptian, and liffe et al (2002) reports a case study of developing
Egyptian hieroglyphics. The message in the three a bilingual Web site in English and Welsh for users
scripts is the same and is a decree by the Egyp- in Wales. This study focused on Web site design
tian ruler Ptolemy V regarding taxes and temple for just two languages and recognized that there
construction. The purpose of having the message are many bilingual areas in the world (Cunliffe
written in three languages adjacent to each other et al, 2002). One of the most important aspects
was to solve a difficult linguistic problem.. of designing for bilingual Web site content is to
A contemporary example of multi-language provide rich interconnectivity between materials
dissemination of important information is a in the two languages (Cunliffe et al, 2002). While
requirement in the California Labor Code about not the main focus of the study, Cunliffe, et al.
employers posting worker information in English (2002) does discuss the options for placement of
and Spanish. According to the Labor Code, several two languages on the same page.
important documents, such as the Notice of Work- Paralingual Web design is expected to affect
ers Compensation Coverage, must be posted in e-government use two ways: increased use due
Spanish and English whenever there are employ- to increased trust and possible decrease in use
ees of Spanish descent (California Labor Code, due to impacts on ease of use and usefulness. As

Figure 1. Technology acceptance model (Davis, 1989)


Perceived
Usefulness
External Attitude Toward Behavioral System
Variables Using Intention to Use Usage

Perceived Ease
of Use

280
Paralingual Web Design and Trust in E-Government

discussed above it is expected to help improve DeLone and McLean (2003) revision of the
trust. The two models mentioned in the Intro- IS Success Model also helps predict the impact
duction help to predict the probable impact of of paralingual Web design. This is a causation
paralingual Web design. TAM was developed model that implies that system quality, information
by Davis (1989) as an explanation of the general quality, and service quality will lead to increased
case determinants of computer acceptance that are use or increased intent to use which will lead to
capable of explaining user behavior across a broad benefits and success of the system (see Figure
range of systems, technologies, and user popula- 2.) The intent to use construct is important for
tions. The model includes use as a determinant this article as it is similar to the ease of use and
but indicates that use is determined by ease of usefulness constructs from TAM, especially when
use and perceived usefulness, attitude, and inten- intent to use is operationalized using the Perceived
tion to use (see Figure 1). TAM is a derivative of Benefit Model (Thompson, Higgins, and Howell,
Fishbein and Ajzen’s (1975) Theory of Reasoned 1991). Additionally, trust is reflected in the infor-
Action (TRA) model. TRA focuses on situation mation quality dimension as users must be able
specific personal beliefs and attitudes, and the to trust the information in the system for there
effects of the beliefs of others who can influence to be quality. Paralingual Web design provides
the individual. The fundamental premise of TRA the trust in the information quality and aids in
is that individuals will adopt a specific behavior perceived usefulness in intent to use. Ultimately,
if they perceive it will lead to positive outcomes the IS Success Model predicts that improved trust
(Compeau and Higgins, 2001). However, adop- will help improve information quality which will
tion is also influenced by two factors, Perceived increase intent to use/actual use leading to benefits
Usefulness and Perceived Ease of Use. Perceived and system success.
Usefulness reflects that an individual’s perception The conclusion from the literature review is
of usefulness influences their intention to use that Paralingual Web design may be a design tool
the technology primarily through the creation that will support building trust in content and
of a positive attitude. This is consistent with the process by e-government users. As mentioned
TRA, which holds that attitude (an individual’s earlier, Geffen, et al. (2002) consider this essential
positive or negative feelings about performing a for e-government adoption as it support perceived
behavior) influence behavioral intention. Geffen, usefulness. The only concern is an impact to
et al. (2002) found that trust impacts Perceived perceived ease of use resulting in the experiment
Usefulness with increased trust improving Per- design discussed in the next section.
ceived Usefulness. Perceived Ease of Use reflects
the user’s assessment of how easy a system is to
learn and use. TAM includes ease of use as a RESEARCH METHODOLOGY
separate belief construct based on the concept of
self-efficacy (an individual’s judgment of his/her This study sought to determine whether para-
ability to organize and execute tasks necessary to lingual Web site design can improve trust in
perform a behavior). Ultimately, TAM predicts e-government Web sites while maintaining ease
that if paralingual Web design can improve trust of use and usefulness for both the reader and the
with the users while not reducing perceptions of provider government. In a bi or multilingual en-
ease of use or usefulness then it can be expected vironment, the ability to communicate concepts
that paralingual Web design will be accepted by across diverse cultures and languages has become
users. increasingly important, especially when issues
of trust are involved. Further, e-government has

281
Paralingual Web Design and Trust in E-Government

Figure 2. DeLone and McLean’s (2003) Revisited IS success model


Information
Quality

Use/Intention
to use
System
Quality
Net
Benefits
User
Satisfaction
Service
Quality

rapidly increased in usage, making it even more English only. However, due to its proximity to the
essential for Web designers to conscientiously border and its large Hispanic/Latino population
strive to ensure that concepts have the same it is reasonable to assume there is a substantial
meaning across cultures. This study focuses on bilingual population.
paralingaul issues in a highly bilingual populated To create the experiment, three informational
location in the United States, that of San Diego Web pages were converted to paralingual format
and Tijuana, with a total combined population consisting of English and Spanish text placed
of over 5 million as of 2004 and over 64 million horizontally adjacent to each other. Informational
annual border crossings. The subject municipality pages were chosen as the National City Web site
is National City, a city of 54,260 (2000 census) is primarily informational and does not perform
located approximately 10 miles from the United any financial transactions. This is considered ac-
States border with Mexico. The municipality ceptable as the majority of e-government Web sites
population is 60% Hispanic/Latino (all or part), in 2004 were informational rather than financial
39% white (all or part), and 30% other (all or part transaction focused (ICMA, 2004). Translation
where other includes Asian, Hawaiian/Pacific of the English content in the original selected
Islander, Black, and American Indian) (Note that Web pages was performed taking into account
the sum is greater than 100% due to respondents the following:
reporting belonging to more than one race) (Na-
tional City, 2008). Actual percentage of English- • Variations in the style and vocabulary of
Spanish bilingual residents is not reported but is Spanish. A style to reflect the local style
understood to be large. The subject municipality was chosen. This style is mostly Mexican in
was selected because of its expected bilingual its structure and vocabulary, so these were
population and its willingness to participate in kept in perspective at all times.
the experiment. It should be noted that simply • The level of writing was kept at approxi-
because the municipality population is heavily mately a high school level of comprehension
Hispanic or Latino does not mean that the local (the same as the English version).
language is predominately Spanish. This is a US • The Spanish translation was written to con-
city and the citizens are US citizens. English is form strictly to correct language structure,
the majority language spoken in National City as syntax, and spelling. This is demonstrated by
evidenced by the National City Web site being in the correct application of diacritical marks,
such as accents, tildes, and umlauts.

282
Paralingual Web Design and Trust in E-Government

The translation task was performed by a na- tion on the National City Web site because
tive Spanish speaker with professional training it is in English and Spanish side by side.
and experience as a translator. The translated
content was then evaluated and modified by a The seventh item was also asked at the re-
Spanish language professor with a specialization quest of National City and queried how aware
in translation studies. the respondent was of the multilingual nature of
A survey was generated using survey monkey National City:
to gauge the opinions of visitors to the experi-
mental Web pages. The survey consisted of eight 7. Are you aware that other residents in National
items. Items 1-4 dealt with demographics of the City speak more than one language? (Yes,
respondents and were asked at the request of No, Not sure)
National City:
The final item operationalized ease of use
1. What is your age range? (18-24, 25-34, 35- and usefulness by querying perceptions of site
44, 45-54, 55-64, 65 or over) readability. Only readability is measured as no
2. Are you a resident of National City? (Yes, transactions are performed with the paralingual
No) Web sites:
3. What language or languages do you use for
communication (speaking, reading, writ- 8. Please respond to this statement: It was easy
ing)? (English only, Spanish only, mostly for me to read the pages with English and
English with some Spanish, about half Spanish next to each other: (No, it was very
English and half Spanish, mostly Spanish difficult, No, it was somewhat difficult, It
with some English, some other combination was neither difficult nor easy, Yes, it was
of languages) (Note that this question was somewhat easy, Yes, it was very easy)
used to group responses)
4. Have you visited the National City official Municipality officials encouraged participa-
Web site before now? (Yes, No) tion by constituents and residents in the vicinity
through a series of bilingual public announce-
Items 5 and 6 operationalized trust. Item 5 ments encouraging individuals to visit the modi-
queries improved trust due to the information fied Web pages and to complete a brief survey
on the Web page while item 6 queries improved documenting their opinions on the Web site. The
trust due to the paralingual format. Two items respondents to the survey had the choice of filling
were used to ensure that improvement in trust out the survey in English or Spanish, presumably
was due to paralingual format and not to just selecting their primary language of communica-
reading the information (both use a 7 point Likert tion. The respondents who answered the online
scale with 1 being Strongly Disagree and 7 being surveys represent a portion of Web users who
Strongly Agree): visited the National City Web site and chose to
view at least one of the paralingual pages. This
5. Please respond to this statement: I have a constitutes a self-selected sample. The visitors to
greater trust now than before in my under- the Web site may have been responding to one
standing of the National City Web site. of the various methods of advertisement of the
6. Please respond to this statement: I have a research, or may have been incidental visitors to
greater trust now than before in the informa- the Web site who then decided to follow a link
from the home page to one of the paralingual

283
Paralingual Web Design and Trust in E-Government

pages. While there are problems with self selected measure for examining the correlation between
samples, it is acceptable for this experiment as the matched responses to Questions 5 and 6 within
goal was to get real users to respond to the Web each language group. When deemed appropriate
pages and the inherent difficulties in recruiting a and meaningful, means and standard deviations
random sample on a public Web page. Data was have been calculated for various data sets as well.
collected over a three month period with 133 All statistical tests were performed via SPSS and
responses being collected. Data were grouped validated using other software. All results are
based on the respondent’s answer to the question reported in the format recommended by Jaccard
asked about the respondent’s primary language and Becker (1990).
for communication. The range of response choices All collected surveys were used for the
was from “English only” to “Spanish only”; an analysis. There was no exclusion or limitation of
additional choice was “some other combination surveys based on any criteria. However, it should
of languages.” In order to generate analytical be noted that some respondents did not fill out all
results that are more representative of the lan- survey questions. For example, a small portion
guage component of the sample data, the English of both English and Spanish respondents did not
sample data included only those who indicated fill out the second half of the surveys. Thus, the
“English only,” “Mostly English,” or “some other data used in different components of the analysis
combination of languages.” The Spanish sample consist of different sample sizes, depending on the
data included those that indicated “Spanish only,” questions being examined. The software used for
“mostly Spanish,” or “half English and half Span- these analytical tests accounted for the missing
ish.” This resulted in 97 English responses and data and thus calculated test values based only on
36 Spanish responses. the number of questions actually answered.
Statistical data analysis was used to analyze
the collected data. Beatty (2000), Siegel (1956),
McClave, Benson, and Sincich (2008), Jaccard RESULTS and DISCUSSION
and Becker (1990), and others have stressed the
importance of selecting appropriate statistical For all tests, the alpha value was chosen to be α
procedures that correspond to the methods used = 0.05. For ease and uniformity of comparison
to assign numerical values, as well as the type and between all tests, the z value will be used. For
level of data being analyzed. Since the question- the chosen α = 0.05 and 2-tailed tests, the critical
naires employed in the current study are based z value is z = ±1.96.
on a traditional Likert format, with anchor points
ranging from “strongly disagree” to “strongly Mann-Whitney U on Grouping by
agree,” nonparametric statistics were deemed ap- Language Choice
propriate. The nonparametric tools used include
the Mann-Whitney U test, the Wilcoxon T test, The Mann-Whitney U test was used to compare
and Spearman’s rho. The Mann-Whitney U test is the central tendencies of responses to item 5: I have
an appropriate tool for comparing central tenden- a greater trust now than before in my understand-
cies between independent responses to Question ing of the National City Web site, based on the 97
5 and 6 across the two language groups, while English-based responses and the 36 Spanish-based
the Wilcoxon matched pairs T test is an appropri- responses. This item was designed to measure
ate tool for comparing central tendencies across differences between respondents’ improvement
the matched pairs of data. With regard to tests of trust in their understanding of the information
of association, Spearman’s rho is an appropriate on the paralingual page only. The Mann-Whitney

284
Paralingual Web Design and Trust in E-Government

U and its corresponding z-value (-1.907) resulted speakers improved their trust because the text
in no statistical difference in perceptions between was in Spanish, this does not necessarily sup-
the English and Spanish respondents. port that a paralingual Web design is necessary
A second Mann-Whitney U test was used to to improve trust and that simply being bi-lingual
compare the central tendencies of responses to in some format will increase trust of the minority
item 6: I have a greater trust now than before in speakers.
the information on the National City Web site
because it is in English and Spanish side by side, Spearman’s rho on Grouping by
again based on the 97 English-based responses Language Choice
and the 36 Spanish-based responses. This item
was designed to measure differences between Spearman’s rho was used to examine the relation-
the respondents’ improvement of trust in regard ship between responses on English item 5 (n=97)
to the information on the paralingual page, based and 6 (n=97) to determine how well the English
on having the information in the two languages respondents’ answers correlated between the two
on the same page. This test yielded a z-value of items regarding trust. The test measures the level
-4.406 (p < .01), indicating that there were sig- of correlation between these two items. The two-
nificant differences between the responses of the tailed test yielded ρ = 0.504 (p < .01), which was
English and Spanish respondents with regard to statistically significant.
this question. Spearman’s rho was also calculated for Span-
These results partially support the hypothesis ish items 5 (n=36) and 6 (n=36). The calculated
that use of paralingual Web pages will increase correlation, ρ = 0.563 (p<.01), is significant for a
trust in the page content and government spon- 2-tailed test as calculated by SPSS.
sor. Since trust levels are different only when the Spearman’s rho was performed to see the
paralingual information is considered, and the degree of correlation between items 5 and 6 in
Spanish respondents showed higher medians than each group. The rho value for English items 5 and
the English respondents in both items 5 and 6, the 6 is 0.504, so the correlation is “significant”. For
Spanish respondents show an increased level of Spanish items 5 and 6 the rho value is 0.563, also
trust based on the paralingual content. significant. Thus each of the two groups answered
consistently in the trust questions.
Wilcoxon T Test on Grouping by
Language Choice Readability Reflecting Ease of Use
and Usefulness of the Paralingual
The Wilcoxon matched-pairs T test was used Web Pages
to compare the central tendencies on data sets
similar to those tested with the Mann-Whitney Item 8 of the survey is a measure of the read-
U test, items 5 and 6. The calculated z value for ability as reflecting ease of use and usefulness of
English items 5 and 6 is z = -3.188. This indicates the paralingual pages. The choices for responses
that there is a statistical difference in the means had a range of five from “it was very difficult”
of the answers to each item. For Spanish items to “it was very easy.” The percentages of the
5 and 6, the calculated z value is z = -1.034 and English respondents (n=97) who answered “it
we fail to reject the null hypothesis, therefore we was somewhat easy” or “it was very easy” is
conclude that there is no statistical difference in 61.3 % and the percentage of Spanish respon-
the means of the answers to each item. dents (n=36) answering similarly is 85.7%. This
This result seems to indicate that Spanish is interpreted to imply that having the page in

285
Paralingual Web Design and Trust in E-Government

paralingual format did not diminish readability expected. It was found that the paralingual format
and thus ease of use and usefulness, important to improved trust for the minority speaker but not
predicting acceptance of paralingual Web pages the majority speaker. Upon reflection this is an
by users. However, it is also implied that English expected finding.
respondents were more likely to find an impact An additional finding with respect to ease of
to ease of use and usefulness. This may suggest use and usefulness was noted. It was found that
that these respondents will be less accepting of in general respondents did not find the paralin-
a change to paralingual format as they perceive gual format hard to read, however, it was noted
paralingual format to be less useful. that the majority speakers (English) were less
enthusiastic than the minority (Spanish) speakers
Alternative Analytical Calculations about the paralingual Web design. This finding
has implications for the adoption of paralingual
Performing nonparametric tests is the appropri- Web design in that it may show that there will
ate method for analysis of ordinal data. These be resistance to adoption of a paralingual Web
results have been shown in the previous section. approach by the majority speakers. The implica-
However, means and standard deviations are tion to policy makers is that there needs to be
more commonly understood and therefore used additional research done with respect of citizen
more commonly to describe data. Table 1 is a attitudes towards bilingual government prior to
summary of the means and standard deviations implementing a paralingual Web strategy. This is
for items 5, 6, and 8. particularly important in different regions of the
United States where it could be expected that the
generally favorable bilingual acceptance attitudes
CONCLUSION found in California may not exist.
The conclusion of this article is that paralingual
This article is primarily intended to provide evi- Web design is useful for e-government in areas
dence to support government decision makers, with significant bilingual populations. However,
e-government researchers, and e-government Web there are limitations to this approach as it appears
designers in applying paralingual Web page design that there is a risk of backlash and rejection from
for improving trust in government in regions where the majority speaking population. The implication
there is a high proportion of bilingual residents. An for policy makers is that paralingual Web design
experiment was performed to test the hypothesis should be used when there are known trust issues
that paralingual Web design will improve trust in between majority and minority speakers that
the content of the e-government Web page without translate into trust issues with government and
significantly affecting ease of use and usefulness. e-government initiatives.
This hypothesis was confirmed, but not quite as

Table 1. Means and standard deviations of survey data


Item Value Range English: Mean/StdDev Spanish: Mean/StdDev
5 1-7 4.2209/1.785 4.8182/1.286
6 1-7 3.5930/1.811 5.0606/1.540
8 1-5 3.88/1.259 4.366/0.994

286
Paralingual Web Design and Trust in E-Government

Limitations Web design involving three or more languages


may not be practical or may induce substantial
This experiment has a small sample with limited ease of use and usefulness issues. Research needs
items testing improved trust and ease of use and to be done to see if this is a practical approach or
usefulness. The conclusion that paralingual Web if more traditional approaches of having multiple
design improved trust for the minority Spanish versions of the same Web page each in a different
speakers is supported by the statistical analysis language is a preferable approach.
but it cannot be totally discounted that trust may A final area for future research is in e-govern-
have been increased simply because the content ment policy. The conclusion that majority speakers
was in Spanish. Additionally, this experiment may resist paralingual implementations is very
only looked at informational Web pages and the important to policy makers. Research into what
conclusions may not apply to financial or other factors may influence majority speakers to accept
transactional Web pages. Finally, only one city and adopt paralingual implementations or which
was looked at, one minority language used (Span- factors may influence majority speakers to reject
ish,) and the sample population was self selected paralingual implementations is critical to policy
meaning that the results may not be reflective of makers for crafting appropriate e-government
all populations, cultures, and languages. strategies and policies. It is expected that para-
lingual Web design will only be appropriate in
Areas for Future Research regions where there is significant lack of trust
by the minority speakers but this needs to be
There are several areas for future research of which confirmed.
the first are those areas that address the limitations
to this research. This includes further studies
using other languages and locations; obtaining REFERENCES
a large sample size; and using transaction based
Web pages in addition to informational pages. 2005 The Year of Languages, (2005). Special
It is expected that there may be ease of use and project of the American Council on the Teaching
usefulness issues associated with paralingual Web of Foreign Languages (ACTFL). Available from
design used for transactional Web pages which http://www.yearoflanguages.org/i4a/pages/index.
could affect adoption of the pages and requiring cfm?pageid=3419 accessed March 21, 2006.
that the trust improvement from paralingual
Ba, S. and Pavlou, P.A., (2002). Evidence of the
Web design be balanced against ease of use and
Effect of Trust Building Technology in Electronic
usefulness.
Markets: Price Premiums and Buyer Behavior,
An additional area for future research is in
MIS Quarterly, (26:3), 2002, pp. 243-268.
using paralingual Web design in a multilingual
environment. In this case multilingual implies Beatty, J. R. (2000). Statistical Methods, Sixth
more than two languages and is a topic of relevance Edition. McGraw-Hill, New York.
in many countries in Europe and Asia (Finland
Becerra, H., (2006). Welcome to Maywood, Where
and Switzerland are two examples). While there
Roads Open Up for Immigrants. Los Angeles
are historical examples of multilingual designs,
Times, March 21, 2006, p. A-1.
none are in Web page layouts (the Rosetta Stone
is an example of a three language representation). Bharat, K. and Broder, A. (1998). A technique for
Given the limitations on screen areas, especially measuring the relative size and overlap of public
for mobile and/or handheld devices, a paralingual Web search engines. In 7th WWW,.

287
Paralingual Web Design and Trust in E-Government

California Labor Code, Section 3550(d). Accessed unifr/egov-trust/slides/html/title.html accessed


on March 24, 2006 from http://www.leginfo.ca.gov/ November 29, 2005.
cgi-bin/displaycode?section=lab&group=03001-
Gefen, D., Warkentin, M., Pavlou, P.A., and Rose,
04000&file=3550-3553.
G.M., (2002). EGovernment Adoption. Eighth
Compeau, D. and Higgins, C., (1995). Computer Americas Conference on Information Systems,
Self-Efficacy: Development Of A Measure And Association for Information Systems.
Initial Test. MIS Quarterly, 19(2), pp. 189-211.
Girion, L., (2006). Language Becoming an Issue
Coleman, S., and Gøtze, J, (2001). Bowling Togeth- for Health Insurers. Los Angeles Times, March
er: Online Public Engagement in Policy Delibera- 20, 2006, p. C-1.
tion. Hansard Society, London. Retrieved October
Grefenstette, G and Nioche, J., (2000). Estimation
4, 2005 from http://bowlingtogether.net.
of English and non-English Language Use on the
Castelli, W.A., Huelke, D.F., and Celis, A., (1969). WWW. Xerox Research Centre Europe. Available
Some basic anatomic features in paralingual from http://arxiv.org/pdf/cs.CL/0006032 accessed
space. Oral Surgery, Oral Medicine, And Oral March 19, 2006.
Pathology, 27(5), pp. 613-621.
Grönlund, A. and Horan, T. A. (2004). Introducing
Cunliffe, D., Jones, H., Jarvis, M., Egan, K., E-Gov: History, Definitions, and Issues. Com-
Huws, R., and Munro, S., (2002). Information munications of the Association for Information
Architecture for Bilingual Web Sites. Journal of Systems, 15: 713-729.
the American Society for Information Science
Gulli, A. and Signorini, A., (2005). The Index-
and Technology, 53(10), p. 866.
able Web is More than 11.5 Billion Pages. WWW
Davis, F., (1989). Perceived Usefulness, Perceived 2005, May 10–14, 2005, Chiba, Japan.
Ease Of Use, And User Acceptance Of Information
Hart-Teeter, (2003). The New E-government Equa-
Technology. MIS Quarterly, 13, pp. 319-339.
tion: Ease, Engagement, Privacy and Protection.
DeLone, W.H. and McLean, E.R., (1992). Infor- A report prepared for the Council for Excellence
mation Systems Success: The Quest for the De- in Government, Retrieved November 27, 2005
pendent Variable. Information Systems Research, from http://www.excelgov.org/usermedia/images/
3, pp. 60-95. uploads/PDFs/egovpoll2003.pdf.
DeLone, W.H. and McLean, E.R., (2003). The De- Hibbing, J.R. and Theiss-Morse, E. (2002). Stealth
Lone and McLean Model of Information Systems Democracy: Americans’ Beliefs About How
Success: A Ten-Year Update. Journal of Manage- Government Should Work. Cambridge University
ment Information Systems, 19(4), pp. 9-30. Press, Cambridge.
Fishbein, M., and Ajzen, I., (1975). Belief, At- International City/County Management Associa-
titude, Intention And Behavior: An Introduction tion (ICMA) (2004). Electronic Government Sur-
To Theory And Research. Addison-Wesley, vey 2004. accessed February 22, 2008 from: http://
Reading, MA. icma.org/upload/bc/attach/%7B9BA2A963-DDCC-
40B7-836D-F1CFC17DCD98%7Degov2004Web.
Gassert, H. (2004). “How to Make Citizenz Trust
pdf,
E-Government,” University of Fribourg, E-Gov-
ernment Seminar, Information Systems Research Jaccard, J. and Becker, M.A., (1990). Statistics for
Group. Available from http://edu.mediagonal.ch/ the Behavioral Sciences. Wadsworth Publishing

288
Paralingual Web Design and Trust in E-Government

Company, Belmont, CA, USA. accessed April 9, 2006.


Jadu, (2005). Enterprise Content Management for National City, (2008). National City Web Site.
Public and Private Sector. Available from http:// Available from http://www.ci.national-city.ca.us/
www.jadu.co.uk/egov/jadu_egov_econsultation. main.asp accessed February 22, 2008.
php accessed November 26, 2005.
National Performance Review, (1993). From Red
Khalil-babnet, M. WSIS Prepcom-2: Cyberse- Tape To Results: Creating A Government That
curity an Issue for All. Available from http:// Works Better And Costs Less. Washington, D.C.:
www.babnet.net/en_detail.asp?id=935 accessed Government Printing Office.
November 29, 2005.
Organization for Economic Co-operation and De-
La Prensa-San Diego. Available from http:// velopment (OECD), (2001). Citizens as Partners,
laprensa-sandiego.org/rates/rates.html accessed Information, Consultation and Public Participa-
November 27, 2005. tion in Policy-Making.
LaVoy, D.J., (2001) Trust and Reliability. Public Osborne, D. and Gaebler, T., (1992). Reinventing
Manager, Fall 2001, 30(3), p 8. Government: How The Entrepreneurial Spirit Is
Transforming The Public Sector. Reading, MA:
Levi, M. and Stoker, L., (2000). Political trust
Addison-Wesley.
and trustworthiness. Annual Review of Political
Science 3: 475-507. OSHA4LESS.COM, “Sales Book - lew 2.pdf”,
2006. Received as attachment to electronic mail
Localization Industry Standards Associa-
message, March 24, 2006.
tion, LISA, (2008). Localization. Accessed
February 23, 2008 from http://www.lisa.org/ Paralinguism in the Theatres and the International
Localization.61.0.html. Theatre Festivals. Sasho Ognenovski. Intercul-
tural Communication, ISSN 1404-1634, 1999,
Macias, E. and Temkin, E., (2005). Trends And
August, issue 1. Available from http://www.immi.
Impact Of Broadband In The Latino Community.
se/intercultural/ accessed November 19, 2005.
Tomás Rivera Policy Institute, 2005. Available
from http://www.trpi.org/PDFs/broadband.pdf Rai, A., Lang, S., and Welker, R., (2002). Assessing
accessed March 21, 2006. the Validity of IS Success Models: An Empiri-
cal Test and Theoretical Analysis. Information
Mercuri, R.T. (2005). Trusting in Transparency.
Systems Research, 13(1), pp. 50-69.
Association for Computing Machinery. Com-
munications of the ACM, 48(5), p. 15. Rosell, S., Gantwerk, H., and Furth, I., (2005).
Listening To Californians: Bridging The Discon-
McClave, J.T., Benson, P.G., and Sincich, T.,
nect. Viewpoint Learning, Inc. Retrieved January
(2008). Statistics for Business & Economics,
15, 2006 from http://www.viewpointlearning.com/
10/E. Prentice Hall, Upper Saddle River, New
pdf/HI_Report_FINAL.pdf.
Jersey, USA.
Sandoval, V.A. and Adams, S.H., (2001). Subtle
Moller, R.M., (2000). Profile of California Com-
Skills for Building Rapport: Using Neuro-
puter and Internet Users. California Research
Linguistic Programming in the Interview Room.
Bureau. California State Library. Available from
Available from http://www.fbi.gov/publications/
http://www.library.ca.gov/crb/00/01/00-002.pdf

289
Paralingual Web Design and Trust in E-Government

leb/2001/august2001/aug01p1.htm accessed No- United States Census Bureau, Census 2000.


vember 19, 2005. “QT-PL. Race, Hispanic or Latino, and Age:
2000. Data Set: Census 2000 Redistricting Data
Schneider, G.P., (2003). Electronic Commerce,
(Public Law 94-171) Summary File Geographic
Fourth Annual Edition. Thomson Course Tech-
Area: National City city, California.” Available
nology, Boston.
from http://factfinder.census.gov/servlet/QT-
Secretariat, Treasury Board of Canada, (2006). Table?_bm=y&-geo_id=16000US0650398&-
Communications Policy of the Government qr_name=DEC_2000_PL _U_ QTPL&-ds_
of Canada. Available from http://www.tbs-sct. name=D&-_lang=en&-redoLog=false, accessed
gc.ca/pubs_pol/sipubs/comm/comm1_e.asp#04 March 2, 2006.
accessed March 21, 2006.
United States Department of Labor, The Family
Siegel, S., (1956). Nonparametric Statistics for and Medical Leave Act of 1993, 29 CFR 825.300,
the Behavioral Sciences. McGraw-Hill Book 2006. Available from http://www.dol.gov/dol/
Co., New York. allcfr/ESA/Title_29/Part_825/29CFR825.300.
htm accessed March 24, 2006.
Tan, C.W., Pan, S.L., and Lim, E.T.K., (2005).
Towards the Restoration of Public Trust in Elec- United States Office of Management and Budget
tronic Governments: A Case Study of the E-Filing (USOMB), (2005), E-Gov: Powering America’s
System in Singapore”, Proceedings of the 38th An- Future With Technology. Retrieved October 5,
nual Hawaii International Conference on System 2005 from http://www.whitehouse.gov/omb/egov/
Sciences, IEEE Computer Society Press, index.html.

Tan, C.-W., Benbasat, I., and Cenfetelli, R.T., Warkentin, M., Gefen, D., Pavlou, P.A., and
(2008). Building Citizen Trust towards e-Govern- Rose, M., (2002). Encouraging Citizen Adoption
ment Services: Do High Quality Web sites Mat- of e-Government by Building Trust. Electronic
ter? Proceedings of the 41st Hawaii International Markets, 12(3), 157-162.
Conference on System Sciences, IEEE Computer
Wikipedia, (2006). Paralanguage. Avail-
Society Press.
able f rom ht t p://en.wi k ipedia.org /wi k i /
Thompson, R.L., Higgins, C.A., and Howell, J.M., Paralanguage#Linguistics accessed September
(1991). Personal Computing: Toward a Conceptual 19, 2006.
Model of Utilization. MIS Quarterly, March, pp.
Wikipedia, (2006a). Rosetta Stone. Available
125-143.
from http://en.wikipedia.org/wiki/Rosetta_stone

This work was previously published in International Journal of Electronic Government Research, Vol. 5, Issue 1, edited by V.
Weerakkody, pp. 36-49, copyright 2009 by IGI Publishing (an imprint of IGI Global).

290
291

Chapter 2.3
Designing Medical
Research Web Sites
Jonathan P. Grady
University of Pittsburgh, USA

Michael B. Spring
University of Pittsburgh, USA

Armando J. Rotondi
University of Pittsburgh, USA

Abstract INTRODUCTION

This chapter discusses the design of Web sites The Internet, specifically the World Wide Web, is
to be used as the basis of medical research. It is being tested by the medical community as a po-
broken down into three sections: Part 1 discusses tentially more efficient and effective mechanism
the various issues that have to be addressed in the to serve the needs of patients and caregivers. The
design of a Web site that will be used to assess some Web can be used to quickly and cost-effectively
intervention based on the Web site. Part 2 discusses disseminate information, collect data and target
the design of such a Web site and the development communications to specific people or groups of
of a tool to facilitate this process. Part 3 presents people. May et al., (2005) describe a range of ef-
the results of preliminary usability analysis for the forts in the United Kingdom to provide services via
tool to assist medical researchers in constructing telemedicine. In the United States, researchers have
Web sites that can meet the needs and requirements begun to study the efficacy of web-based medical
of medical intervention studies. The results of the interventions. Studies have been performed in recent
preliminary interviews, prototype walkthroughs, years examining the effectiveness of internet-based
and preliminary usability studies are presented interventions with patients suffering from a variety
laying the groundwork for future development and of medical conditions, including heart transplants
more formal usability studies. (Dew et al., 2004), smoking cessation (Lenert et
al., 2003), osteoarthritis (McAlindon, Formica,
DOI: 10.4018/978-1-60566-356-2.ch053

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Designing Medical Research Web Sites

Kabbara, LaValley, & Lehmer, 2003), clinical • Medical researchers may not understand
depression (Christensen, Griffiths, & Jorm, 2004), the capabilities or the limitations of data
traumatic brain injury (Rotondi, Sinkule, & gathering related to the website and de-
Spring, 2005), and schizophrenia (Rotondi et al., signers may not be aware of the data analy-
2007). While these studies had different degrees sis needs of the intervention. For example,
of success achieving their goals, they generally how will session information be main-
concluded that 1) the Internet is a potentially tained, how will IDs be assigned, and how
useful tool for medical interventions, and 2) the will treatment groups be identified?
success of an Internet-delivered treatment depends • The designers may not have sufficient
largely on the design of the website and other knowledge of the Health Insurance
electronic communications used in conjunction Portability and Accountability Act of 1996
with the treatment. (HIPAA) which sets rigorous standards for
The cost to design, implement and maintain protecting patients’ personal information
a website to be used as the basis for medical (Neale & Schwartz, 2004); data collection,
intervention research can be high. Projects at the storage, and sharing mechanisms must
University of Pittsburgh by Rotondi and Spring be carefully designed to adhere to these
budgeted equipment, staff and maintenance costs standards.
for a medical research website well in excess of
$100,000. Marshall and Haley (2000) report the Websites used in a variety of interventions
cost of setting up the site infrastructure alone at reported in the literature had many common ele-
$35,000. Mounting a website for medical research ments. This suggested that it might be possible to
takes from three months to a year, depending on develop tools to aid the development of reliable
the amount of usability testing that needs to be and functional medical intervention websites at
done. At the same time, an under-budgeted or lower cost. With this in mind, we set out to develop
poorly designed website can result in sub-optimal a system that would:
usability for subjects, data that cannot be used to
answer research questions, exposure of sensitive 1. Eliminate common pitfalls in design
patient data, etc. Poorly designed research websites 2. Enhance data collection to better meet re-
are created because: search objectives.
3. Reduce cost.
• The medical researcher does not under- 4. Ensure that the website is maximally usable
stand the nature of website design. (A re- by the target population.
searcher may not anticipate the need to 5. Ensures that the website is as secure as
translate their professional model of the possible.
subject matter into a model manageable by
the consumer.) The effectiveness of a system can be assessed
• The individual or team hired to design the based on:
website may not have knowledge about
the nature of the subjects or the treatment. 1. Time required to build a website.
(The designer may not be aware of physical 2. Monetary cost of building a website.
or cognitive impairments that will require 3. Ability to meet HIPAA requirements for
simplified navigational systems or reduced security and privacy of medical data.
page complexity.) 4. Usability of the website from both the end-
users’ and researchers’ perspectives.

292
Designing Medical Research Web Sites

The rest of the chapter is structured as fol- the screen layouts of each Web page, 5) selected
lows. First, we describe the needs of web-based a middleware product for connecting the site’s
medical interventions, as well as HIPAA and its pages to the DBMS, and 6) coded the pages, us-
implications for medical websites. This section ing JavaScript validation code on each page to
also provides a survey of recent efforts to stan- verify user input. After completing the database
dardize the collection and presentation of medical and website, the authors 7) set up a machine to
information. Second, we present the framework for house both the database and web server software,
designing Internet-based interventions based on 8) obtained a unique IP address and domain name,
the critical characteristics identified from recent 9) applied security measures to the infrastructure,
literature and introduce Webmaster-R (Webmaster and 10) trained the staff who will enter data. The
for Research), a system built around the frame- authors provided a detailed outline of suggested
work that is designed to augment a researcher’s security practices, including security procedures
ability to mount a medical intervention website. for the website, servers, client machines, and
Third, we present some formative evaluation data facilities housing the servers.
related to the efficacy of the Webmaster-R system Spring and Rotondi at the University of Pitts-
using a heuristic evaluation and walkthrough with burgh collaborated on the development of several
researchers from the University of Pittsburgh’s sites to support research on internet-based medical
School of Medicine. The last section discusses interventions. This chapter is in part a report of a
directions for future work. series of interviews conducted in July, 2005 with
Spring, Rotondi, and other researchers involved in
the early stages of website development to gather a
BACKGROUND preliminary list of needs. The list of critical needs
generated from these interviews included:
Bush et al. (2004) proposed a framework for
Internet and Web access that medical researchers • Navigation and design elements appropri-
should consider when designing an online study. ate for audience
The authors divided these issues into two groups, • Methods to ensure information is presented
connectivity and human interface. Connectivity at the right level of detail for the intended
encompassed all issues related to the capabilities audience.
of the hardware and software used in conjunction • Tools to optimize data collection and
with Internet access, as well as the availability of analysis.
access based on geography, locale, and degree • Naming conventions for various elements
of convenience. Human interface involved all of an internet-based intervention, includ-
issues related to demographics, skill sets, and ing the usernames of participants and the
special needs of potential participants. Bush et names of web pages.
al. then applied the highly detailed framework • Ability to segment participants and isolate
to the recruitment of subjects for a pilot study interactions based upon these segments.
communicating breast cancer risk. Marshall and • The site developed complies with HIPAA
Haley (2000) presented a ten-step process for guidelines.
building a secure website for collaborative re-
search from the ground up. The authors focused Based upon this input, a prototype system was
on the site’s database, including 1) developing the developed, whose design and preliminary evalu-
database schema, 2) selecting a DBMS package, ations are explained in greater detail in Sections
and 3) creating a separate table of variables for 3 and 4 respectively.
each investigator. Next, the authors 4) designed

293
Designing Medical Research Web Sites

Figure 1. Comparison of the identified types of medical websites

A Website Taxonomy information dissemination, 2) patient interven-


tions, or 3) research on interventions. Figure 1
Websites serve different purposes. We classify summarizes a few of the differences. The nature
sites generally as 1) providing information, 2) of the types of sites is described below.
supporting interaction, 3) allowing transactions,
4) supporting transformation. Generally, the Information Websites
evolution of a website moves from information
only to one that provides the first three functions. Medical information websites provide information
The last stage of website evolution is one where it on one or more medical conditions. No relation-
transforms the nature of the service or business. For ship with a physician is assumed. Because these
example, in phase 3, the web provides an alterna- websites do not provide any treatment, they are not
tive mechanism for transactions. In phase 4, new bound by HIPAA guidelines. If discussion groups
services may be provided which are not possible and chat rooms are publicly accessible, users are
in non-web formats. Our work here is primarily informed that their postings are not private. One
directed to sites that provide for information and of the largest and well-known medical information
interaction, with some minimal consideration of websites is WebMD’s Consumer Portal (http://
transactions. The use of websites to transform www.webmd.com), which provides public access
how medical interventions are delivered is likely to treatment information, discussion boards, and
to come about only after significant research data expert chats concerning a wide variety of medical
has been collected on the efficacy of parallel conditions, all targeted to a general audience.
interventions.
We further classify medical websites in terms
or whether they are intended to support: 1) general

294
Designing Medical Research Web Sites

Medical Intervention Websites Common Characteristics of


Medical Intervention Websites
Medical intervention websites are designed for
physicians to provide information and/or treatment Many of the early Web-based medical interven-
to specific consumers who have specific medical tions use the web to deliver treatment and pro-
conditions. While the information provided on vide guidance to patients, family members and
medical intervention websites may be similar to the caregivers of those suffering from a medical
medical information sites, it is targeted at specific condition. These interventions have become more
individuals and may contain references to an feasible over the last 10 years as the penetration
individual’s specific condition. Thus, measures of Internet connectivity in U.S. households has
are taken to ensure that interactions among site increased. There is increasing support for the use
users remain secure and private. Because client- of the web for a wide range of treatments (Ikemba
physician relationships are established, medical et al, 2002.) Our work on intervention websites
intervention websites must adhere to HIPAA has focused on two common functions:
guidelines. Therefore, these sites tend to serve a
much smaller audience whose members must be Information Dissemination
granted access to the sites resources explicitly.
Interactions on the site (e.g., discussion groups) are The websites include a repository of condition-
monitored and accessible only to group members specific medical information designed for patients
and authorized individuals. Data collected from and their families, as well as links to selected
the website must be secured, and all personally external resources. Presenting the information
identifiable information eliminated from general at an appropriate level with the right amount of
exposure. detail was found to be especially critical. Rotondi
et al. (2007) used a concrete, text-rich design with
Medical Intervention a shallow navigation structure that has been found
Research Websites to be more appropriate for those suffering from
cognitive impairments, such as schizophrenia.
Similar in structure to medical intervention Christensen et al., (2004) included a depression
websites, medical intervention research websites literacy website designed with an eighth-grade
are specifically designed for collecting data on reading level. Lenert et al. (2003) attributed the
client interaction as clinical trials are conducted high drop-out rate in their multi-stage smoking
through the website. Medical intervention re- cessation program to relevant medical informa-
search websites share the same characteristics of tion being presented at an inappropriate level of
medical intervention websites, but also include detail (namely confusing navigation structure and
features – such as log and data analysis tools - to text-heavy design).
aid researchers in the analysis of the efficacy of
the intervention. The studies cited in the next Communication Mechanisms
section all employed secured, private medical
intervention research websites available only to Communication between individuals and profes-
specific audiences dealing with specific medical sionals is an important aspect of many research
conditions. intervention studies. The public intervention
websites (e.g., WebMD and Schizophrenia.com)
provide public discussion boards for individuals
to share experiences related to the medical condi-

295
Designing Medical Research Web Sites

tion or topic. The research-oriented intervention tions on the privacy and security of electronic
websites included a mechanism for doctors to patient-physician communications. The original
send messages to groups of patients, and most text of the HIPAA Security Rule required all email
included a hyperlink or contact form for individual communications between patients and physicians
patients to send messages to doctors. Gray et al. to be encrypted. Because no universal solution
(2000) included monitored discussion boards for email encryption exists, this rule would have
for interaction among patient families, as well significantly discouraged the use of email by
as video conferencing capabilities for real-time both patients and doctors, impeding communi-
communication between families and medical cation between the two. The current text of the
staff members. Communication was not anony- Rule now states that email encryption should be
mous in the discussion groups, but privacy of performed if it is feasible, but allows organiza-
communication was paramount in doctor-family tions to address alternative methods. Schmidt
interactions. Because the site fostered a great deal advocates examination of security measures (ex.
of support among families and caregivers, the PGP, Web-based Encryption) to avoid accidental
Internet-based intervention was well-received. disclosure but suggests having patients authorize
Lenert et al.’s (2003) smoking cessation interven- the use of unencrypted email.
tion had limited interaction capabilities, other than Baer, Sariou, and Koutsky (2002) provided a
doctor-patient and site-patient communication via detailed explanation of their data collection and
email, which may have also led to the program’s storage methods of sensitive health information
high dropout rate. from university students. Men and women were
assigned access to different URL’s, where all com-
HIPAA Concerns munication (including login) was encrypted using
SSL 3.0. All form data provided by subjects were
The Health Information Portability and Account- validated using a JavaScript-based validator before
ability Act (HIPAA), which took full effect in being posted. Validated data were encrypted using
2003, specifies procedures for handling the per- PGP v.2.6.2 and then written to the server in text
sonally identifiable information (PII) of patients files. Only authorized administrators decrypted
in a secure, private, and confidential manner. the data in the text file, using a separate computer
HIPAA places restrictions on the PII that can be disconnected from all networks.
collected, stored, and transmitted via websites The need for website authentication and access
and email communications. The Act also requires controls was critical to the integrity of experi-
organizations to disclose their procedures for ments, especially when identifying the subjects
handling PII in their privacy policies. While many of experimental groups using the website during
organizations have updated their privacy policies trials, and making sure only they were the ones
to meet new guidelines, a recent study found the using the website. McAlindon et al. (2003), while
new policies have actually become harder to read using their website to also recruit potential subjects
and understand (Anton et al., 2007.) Given the for their osteoarthritis trials, used offline consent
ramifications of HIPAA’s guidelines, as well as the forms and medical records to authenticate subjects
long-existing needs to preserve patient confiden- before accepting them into the study. Gray et al.’s
tiality and the integrity of research experiments, a (2000) Internet-based intervention for high-risk
number of recent studies have focused on privacy infants utilized RSA’s SecureID handheld device,
and security implementations within the context allowing users to log in to the website with a one-
of Internet-based interventions. time hash generated from the device along with
Schmidt (2003) examined HIPAA’s implica- their memorized PIN.

296
Designing Medical Research Web Sites

FRAMEWORK FOR Infrastructure


MEDICAL INTERVENTION
RESEARCH WEBSITES Infrastructure includes the hardware running the
intervention website, as well as the facilities and
The frameworks suggested by Bush et al. (2004) network supporting the site. With the risks of
and Marshall and Haley (2000) provide reasonable privacy, security, and confidentiality breaches of
starting points for medical intervention research- patient data, setting up a properly-designed and
ers. However, these solutions require a significant robust infrastructure requires careful consider-
working relationship between researchers and ation. Infrastructure concerns include:
web designers. As Rotondi and Spring reported
in interviews, the cost of this kind of web site • Physical Facilities: Equipment should be
development and maintenance exceeded $200,000 housed in locked rooms with access lim-
in their studies. Examination of the frameworks ited to a small number of people. Servers
suggests that the difficult problems exist in three should be connected to uninterruptible
areas: infrastructure, site development, and data power supplies with surge protection to
analysis. There are interrelations among these prevent loss of data and minimize server
three areas, but they can to some extent be treated outages.
separately. The area of greatest cost variance, and • Network: The bandwidth and Quality-of-
the area most amenable to automation is the sec- Service(QoS) needs of a medical interven-
ond – site development. The primary focus of our tion website will vary depending upon the
work has been on addressing this second area by web components, the number of users ac-
developing a tool called Webmaster-R (Webmas- cessing the site, frequency of access, and
ter for Research). While the tool is conceptually the amount of data transmitted per visit.
similar to generalized tools such as Dreamweaver Employing a bandwidth-intensive compo-
and FrontPage, the tool is also designed to insure nent such as video conferencing, as Gray
that the site meets the requirements for security et al. (2000) did in their high-risk infant
and privacy of medical data as outlined in the study, makes Quality-of-Service a poten-
Health Insurance Portability and Accountability tially critical aspect of a medical inter-
Act, meets the needs of the researchers for ease vention website. Service disruptions may
of construction and maintenance, and insures that cause undue anxiety to the patients (or
data is appropriately collected and organized for caregivers), diminishing the effectiveness
research study data analysis. This work has been of the intervention and suggesting the need
funded in part by the Dean of the School of Infor- for redundant servers, disks, and network
mation Sciences at the University of Pittsburgh. access points.
Below we discuss some of the infrastructure • General Server Security: The number of
considerations, which are generally outside the users with access to the data on the serv-
scope of the current work. Next, we introduce the er should be kept to a minimum. Default
website development framework considerations. passwords should be changed to ones that
Finally, we examine some of the data analysis cannot be easily guessed or discovered us-
considerations, which are not yet addressed, but ing brute-force attacks. Anti-virus, firewall,
which are within the scope of this research. and intrusion detection software should be
installed and configured to run automati-
cally. All unnecessary network services
(ex. Telnet, FTP, IRC) should be disabled.

297
Designing Medical Research Web Sites

Network connections to the server for pur- provides an indication of the complex issues that
poses of development should be restricted a medical researcher should insure are being ad-
to authorized domains. dressed by those responsible for installation and
• Logging, auditing: Auditing and log anal- management of the infrastructure.
ysis is critical in determining not only who Related to the Operating System (assuming
is accessing the site, but also in providing Microsoft Windows):
statistics on how often subjects are visiting
the site and what activities they engage in • Apply all security patches before connec-
during their visits. Logging was important tion to the Internet, from a CD or other re-
in Lenert et al.’s (2003) smoking cessation movable media.
intervention for determining how far sub- • Do NOT set up the server as part of a
jects progressed in the multi-stage program Windows domain.
before quitting the trial, as those subjects • Set the administrator password to at least 8
most often did not respond to e-mail fol- characters in length.
low-ups. When assigning user logins, em- • Use a password that is difficult to guess,
ploy naming conventions that make iden- not based on a dictionary word, using a
tifying different categories of authorized mixture of numerical, upper and lower
users in the log files easier. case characters.
• Backup procedure: Data critical to the • Create separate logins for each user, and
intervention – database tables, web pages, give them access only to the resources they
and site structures – should be backed up need.
each night on tape media. RAID redundan- • Install anti-virus software (such as
cy should be employed to prevent loss of Symantec Anti-Virus).
data in case of a fatal hard drive failure. All • Set the Windows Firewall to “on”.
backup procedures should be documented. • Obtain (for free) Microsoft’s Baseline
• Web and DBMS servers: All pages, Security Analyzer for routine security
including the root page of the website, checks.
should be accessed only through authenti- • Check event and system logs on a regular
cated login, ideally using SSL encryption basis (daily or weekly).
(thus requiring a certificate issued from a • Set Windows Automatic Updates to notify
certificate authority such as VeriSign or you when patches are available, download
Thawte.) The URL of the research web- them, but NOT to install them. Verify and
site should be advertised only to autho- only then install them.
rized users. Permissions should be set on
directories and/or specific pages to prevent Related to the Web Server (assuming Micro-
access from unauthorized users within the soft IIS):
research study (e.g., to prevent partici-
pants from gaining access to administra- • Turn on only those services (e.g.. FTP,
tive pages.) Security procedures should be mail) that you need.
documented, and staff members should be ° If using IIS 5.0, turn off all unnecessary
trained on website security techniques. services.
° If using IIS 6.0, check that all other services
While this chapter focuses on the develop- are turned off (by default, they should be.)
ment of the website itself, the following checklist • Ensure logging is enabled

298
Designing Medical Research Web Sites

• Ensure that Directory Traversal is turned communication (ex. medical diagnosis, testing,
OFF. prescriptions), and 4) transformation, which
• Disable Internet Printing refers to unique services available only through
• Give write privileges only to administra- web services. We examine the nature of the first
tors or other users authorized to edit the three below.
content of the website. Information components provide access to
the medical content and related web resources
Securing Your Database Server (assuming located internally or externally, as well as any
Microsoft SQL Server): means of searching, indexing, or managing the
site’s content and resources. The presentation of
• Run the database server under an account information should center on the characteristics
without administrator privileges and needs of the target audience. Information
• Assign a password to the “sa” account. should be presented at the appropriate level and
• Do NOT allow direct connections to the appropriate detail for the understanding of the
database server. intended audience, and mechanisms should be
• Devise a backup plan for all databases put in place to ensure the appropriateness of the
• Do NOT grant full access to your databas- content (ex. checking reading level). Navigation
es to all users. Limit users’ access to only structures should coincide with a user’s mental
those resources that they need. model of the presented information. Mechanisms
• Ensure that all scripts and code pro- should be provided for collaborating researchers
grammed by developers are secure. to insert and review resources. Resources should
° Verify input data types (ISNUMERIC) be indexed with search capabilities, so users can
° For string data, replace single quotes with easily locate relevant material. Other Web re-
two single quotes sources, such as documents on external websites,
° Create reusable input validation modules should be included only if they are relevant to
° Use stored procedures to abstract data ac- the treatment.
cess. Interaction entails any aspect of person-to-
This is only a sample list. More definitive and person or person-to-group communication facili-
up-to-date lists can be obtained by your organiza- tated by the website. In a medical intervention
tional security offices or from organizations such environment, interactions can have many distinct
as CERT (http://www.cert.org/). forms: Doctor-site, Patient-doctor, Caregiver-
doctor, Caregiver-patient, and Patient-patient.
Website Design Principles Doctor-site interaction refers to “broadcast” com-
munications that are posted by a medical researcher
As with websites generally, the components of onto the website, such as responses to questions
the website can be classified into four categories: or event announcements, targeted to the general
1) information, referring to the medical content audience. The remaining interactions, however,
and related resources of the site, as well as any are all individual-individual, and require special
means of searching, indexing, or managing the consideration in terms of HIPAA compliance and
site’s content and resources; 2) interaction, which confidentiality:
describes person-to-person communication facili-
tated by the website, 3) transaction, which refers • Isolation of discussions: Create sepa-
to any medical treatment performed via electronic rate chats or discussion boards, granting

299
Designing Medical Research Web Sites

permissions only to members of specific Transactions – namely diagnosis, testing, and


groups to facilitate treatment and preserve prescriptions – are the most difficult to implement
patient confidentiality. via internet-based interventions, and are rarely
• Privacy of communications: Interactions attempted in the cited literature. Mechanisms
should only be disclosed either accord- for supporting these transactions must allow the
ing to the site’s privacy policy, or by ex- physician the ability to perform the requested
plicit consent of the patient or caregiver. transaction without violating medical ethics. Phy-
Mechanisms should be put in place (such sicians who diagnosis conditions electronically
as options on an HTML form) that allow must be able to do so with the relative ease and
participants to control disclosure of their certainty as through traditional means, especially
communications. the ability to authenticate the patient and verify
• Anonymity of communication: According the presence of a condition. Web-based testing
to HIPAA guidelines, virtually all person- should only be performed within acceptable risk
ally identifiable information (PII) must be levels, and if the outcome measures of the tests can
stripped from communication. When possi- be transmitted electronically. As is the case with
ble, data collection tools should be set up to diagnosis, authenticating patients and verifying
exclude PII and only use usernames, mak- the need for requested medications are paramount
ing identifying individuals difficult should when prescribing medications via electronic
information be disclosed unintentionally. means. McAlindon et al.’s (2003) osteoarthritis
• Real-time communication: IRC and clinical trials tested the effects of glucosamine,
video-conferencing capabilities should be a safe nutritional product, on osteoarthritis of
included if synchronous communication is the knee; with a testing procedure that had a
required as part of the intervention. very low risk of complications, McAlindon et
al. felt comfortable conducting the experiment
Other elements of communication that require remotely. They also stress that osteoarthritis of the
modifications to infrastructure procedures should knee is relatively easy to diagnose remotely, so
also be considered: they relied heavily on recruitment and diagnosis
through electronic means (although, as mentioned
• Archiving of communication: earlier, participation in the study depended on the
Communication relevant to the research, subjects’ medical records and their signing of an
particularly any interaction involving pa- offline consent form.) Finally, they also argue that
tients, should be stored following HIPAA testing through electronic means is feasible if the
guidelines, eliminating PII. Archives outcome measures of the trials can be transmitted
should be organized and stored on a secure electronically, such as their verified osteoarthritis
server, preferably in a secure directory on test that relies on a computer-based self-evaluated
the web server where it can be recorded for questionnaire.
backup.
• Flow-through mailing of communica- Automated Website Design
tion: Mechanisms should be created that
allow for automatic classification and re- Initial meetings were held with M.B. Spring and
direction of electronic communication re- A.J. Rotondi to review their experiences in devel-
ceived by the site’s mail server to the ap- oping the websites for NIH studies. They provided
propriate researchers or staff members. access to the software developed for these sites,
as well as information about the design process,

300
Designing Medical Research Web Sites

mistakes and oversights that were made, and resource libraries, using sophisticated document
other problems encountered. Combining a general indexing techniques. Threaded discussion boards
design framework developed by Spring intended use simpler database search techniques.
for conference and proceedings management and Interaction - Components were developed to
the custom-designed software for websites that provide functionality commonly found on medical
served as the basis for previous medical interven- intervention research websites to facilitate com-
tion research studies, a system was developed that munication among participants and staff. Inter-
poses a series of questions to principal investiga- active modules available through Webmaster-R
tor in terms that they should understand. Based include:
on those answers a fully functional website is
developed that can be manually enhanced if the • Threaded discussion boards
PI so desires. Users specify the design elements • Blogs, either as personal journals or an-
of their site and the components they wish to nouncements from staff
include. These preferences are then dynamically • IRC chat
merged with a series of predefined templates to • Video streaming
produce the site’s pages. The system design and • Contact forms
how it addresses the theoretical framework are
discussed below. Researchers can create multiple instances of
Information - A series of components were the above modules, and specify which groups
developed to address the presentation and navi- of participants have access to specific instances.
gational needs of a website, as well as the ap- Additionally, researchers have the ability to create
propriateness and management of content. The their own online questionnaires and specify how
system’s Website Options allow the researcher to 1) they will be presented to users, and 2) how
create the presentation and navigational elements they will be measured for quantitative analysis.
of a website, including: Secure mail facilities are included with appropriate
modules (i.e. discussion boards, contact forms)
• Cascading style sheets to direct communication from participants or the
• Global website navigation menu website to appropriate staff members.
• Website logo Transaction - Components of Webmaster-R
• Color and font schemes for text, links, and can be used for transactions, but in general, the
backgrounds framework stops short of explicitly advocating
transaction-type components because of the com-
Tools are also available to help researchers plex implications, as well as ethical considerations.
determine the appropriateness of content in terms Even in cases where the medical researcher is also
of level and detail. Researchers can compute a the primary care physician for the intervention’s
simple Flesch-Kincaid reading level score for participants, the risks of misidentifying a patient,
documents, or they can evaluate a document’s misdiagnosis, or providing an erroneous prescrip-
readability based on the cognitive abilities of its tion often outweighs the benefits of performing
intended audience. a transaction via electronic means. The obstacles
All content added to Webmaster-R is in- to building a general-purpose set of tools for di-
dexed and linked appropriately according to the agnosing any ailment and prescribing any drug
researcher’s preferences. Search functions are electronically via this type of system are great.
built into information-intensive modules, such as With cautions, PI’s can use components such as

301
Designing Medical Research Web Sites

self-report surveys and mail – included under Implementation


the interaction class to approximate transaction
interactions. Webmaster-R was implemented as a Java-based
application, producing standard XHTML or JSP
Content Management & documents as output from pre-defined templates.
Data Analysis Tools For output that requires database-driven com-
ponents, Webmaster-R includes a series of pre-
Content management facilities through a protected configured Java servlets able to communicate with
administrative section are automatically included any Microsoft SQL Server database. Included are
with every website created by Webmaster-R. Items servlets that automatically write out the neces-
in resource and question-and-answer libraries can sary database tables for the website, all of whose
be added, removed, or edited either directly online schema conform to HIPAA guidelines. At the time
or via web upload. Researchers may organize the of this writing, a typical Webmaster-R session of
content in both libraries by categories that may be an hour will result in the creation of more than
added or re-labeled online. Access to real-time chat 250 files, most of which are programs that process
rooms may be granted to groups or specific users data, distributed in 25 folders representing almost
through the created website, and full transcripts a Megabyte of data.
of all chat room discussions are available at any Webmaster-R is available for free to anyone in
time. With the consent of participants, researchers the medical research community to use.
can also add and edit profiles, bios, and photos of
participants and staff.
Data and log analysis tools are provided to help PROTOTYPE EVALUATION
researchers audit their site’s usage, track usage for
individual users, and identify potential problems After development of a prototype version of
with the sites usability. Website logs are parsed Webmaster-R, meetings were held with two re-
to remove entries that represent support files (i.e., searchers from the University of Pittsburgh who
images, external stylesheets and HTML framesets) were in the early stages of programs that would
and filtered to remove visits by staff members. entail research on web based medical interven-
Tools are available to help researchers identify tions. The researchers provided more insights into
users’ sessions, as well as search queries issued the budget process, usability issues encountered,
by participants. Researchers can then create cus- technical limitations overcome, and techniques
tomized summaries or detailed reports via a web used to comply with HIPAA. This section reports
interface. For example, a researcher can produce on a series of prototype walkthroughs and an initial
a report summarizing the number of postings and usability study with a research team developing
read messages within a discussion group over a a new medical intervention research website.
given timeframe, or how many times a particular After Figure 2, Figure 3 presents a summary of
user accessed specific threads over the duration of the feedback evaluating the effectiveness of the
the study. Full automation of the data collection framework. Five metrics were used:
has not yet been implemented, but all the tools
and procedures have been documented that allow • Saves money?
researchers to collect and analyze both web logs • Saves time?
and database tables. • HIPAA compliant?
• Acceptable usability - application?
• Acceptable usability - resulting websites?

302
Designing Medical Research Web Sites

Figure 2. The functional components of Webmaster-R

Figure 2 shows the functional components ties are added. For example, answering a question
of Webmaster-R. The PI is able to incrementally such as “Do you want periodic reports of website
develop the components, adding and changing activity?” will activate code fragments and scripts
the components at any time. that will do database queries automatically at
Users can preview their work in HTML as they specified intervals and mail the results to given
work, and also have the option of publishing the email addresses. Answering yes to a simple ques-
site to their local computer (for later uploading) or tion will add code and structure that would take
directly to a web server, provided they have a net- at least several hours and, in many cases, several
work connection to the server. As the components days of programming if done from scratch.
are added, a system of files is built up recording A typical view of Webmaster-R is provided in
the decisions. At any time, the basic configuration Figure 3 showing the kinds of information that
can be modified or changed. As the PI answers the PI is expected to provide.
questions that have been designed to make sense
to them, complex pre-configured common facili-

Figure 3. Screenshot of Webmaster-R

303
Designing Medical Research Web Sites

Prototype Walkthroughs that some of the program’s user interface elements


were confusing, particularly the names of some
The first of three prototype walkthroughs was of the input fields and buttons. They suggested
conducted in July, 2005 with three researchers from improving the interface by providing more visual
the University of Pittsburgh. The walkthrough con- cues as to what website content a particular field
sisted of a demonstration of an early Webmaster-R will generate – for example, pre-filling each com-
prototype, with feedback gathered during and after ponent’s text fields with descriptive text rather
the demonstration. The Webmaster-R prototype than presenting empty fields.
demonstrated at this meeting had fully functional The suggestions were incorporated into
Website Option tools, but only one functional Webmaster-R’s user interface, and development
Website component, the Home Page. continued on the component parts of Webmaster-
Participants were asked the following questions R. In November, 2005 a second walkthrough was
regarding the application: conducted with one of the researchers interviewed
in the initial meeting. Website components for
1. In what areas of research would you utilize discussion boards, resource libraries, and group
a medical intervention website? member profiles had been completed and were
2. What special needs of your audience need demonstrated. The participant indicated that the
to be addressed in the design of the site? demonstrated components, as well as the rest of
3. Are there aspects of Webmaster-R that you the program would be useful, but that a stream-
feel would be helpful to you? ing video component would also be important
4. What do you think is missing from to her research. The demonstrated components
Webmaster-R? would help her reduce both development time
5. Does Webmaster-R meet your research and monetary costs, while still complying with
needs? HIPAA regulations. She also indicated that a mini-
6. Would you consider using Webmaster-R to mal set of documentation (10-20 pages) should
develop a website? accompany the application as a user’s guide. This
walkthrough was followed by the initial usability
The initial impressions of Webmaster-R were test of the Webmaster-R prototype. Following the
generally favorable. Two researchers indicated meeting, online help documents were created and
that they would consider using the application to added to the application.
build a medical intervention research website, and In the third meeting held in December, 2005,
that it would reduce the time needed to implement two researchers from the WPIC Behavioral
the site. One participant indicated he could not Medicine Program were presented with a com-
because the websites generated by the program plete prototype of Webmaster-R, including design
would not meet the usability needs of consumers changes incorporated from previous feedback. The
with cognitive impairments. Participants found the researchers were given the option of using the
choices of website components suitable for their application to prototype the website themselves
information and interaction needs, although they rather than hire outside consultants to build the
could not fully judge the application’s ability to site. The researchers indicated the application
reduce development costs or comply with HIPAA would be very useful for building a website and
without seeing the finished components. They would reduce their development costs in terms of
also cited the program’s ease of use and ability to both time and money. They found the set website
generate a website from a small set of preferences components sufficient for their research. Both
as strengths. Participants also indicated, however, researchers still found some of the program’s

304
Designing Medical Research Web Sites

terminology confusing, and felt some of the result- of each usability test was to assess Webmaster-
ing pages (particularly in the Library component) R’s user interface – not necessarily to test if the
would be unsuitable for their target audience – i.e. user could construct a production quality website
adolescents. Following this meeting, the appli- in one sitting.
cation’s interface and templates were altered to Based on the test results and the participants’
address the researchers’ feedback. The changes feedback, the tool shows promise but is not yet
were implemented prior to the second usability a complete solution to researchers’ needs. We
test. A set of documents was also created to help identified three critical areas where the tool needs
the researchers with issues outside the scope of the to be enhanced:
program, including 1) instructions for installing
the program, 2) generating and saving additional • Provide more component choices: In her
web pages using a word processor, and 3) using second usability session, the STI researcher
document templates to enforce a consistent style requested a number of website components
across all documents. that were unavailable in Webmaster-R. Her
requested site functionality included com-
Usability Tests ponents based on “Web 2.0” technologies
– technologies that provide a more inter-
After further development of the prototype active and seamless experience for visi-
based on the feedback from the aforementioned tors. Two such components – a personal
interviews, three usability test sessions were blog for patients and a real-time chat re-
conducted – two with a researcher from the quiring only a standard web browser, have
Starzl Transplant Institute (STI) and one with since been incorporated into Webmaster-R.
a research assistant from the WPIC Behavioral Other requested components, such as an
Medicine Program (WPIC). Both sessions with online questionnaires and video stream-
the STI researcher were conducted as cognitive ing, are more challenging to include in an
walkthroughs (Wharton et al., 1994) requiring the automated tool because of the complexity
researcher to show the steps she would take to 1) of the final products. For questionnaires, it
begin a website project, 2) add website compo- was relatively easy to construct a tool to
nents to the project, and 3) configure a discussion build questionnaires based on a series of
board for all group members using a prototype of questions using a Likert scale. The pros-
Webmaster-R. The session with the WPIC research pect of an instrument with complex logic
assistant was conducted similar to Battleson et al. and varying response types was beyond the
(2001) using the think-aloud protocol on a more scope of this current effort. For video pre-
advanced version of the Webmaster-R prototype. sentations, the tool should not only be able
The research assistant, who was not familiar with to incorporate items from various multime-
HTML, was asked to start a new website project dia formats, but it should also build pre-
and complete the following sub-tasks: enter the sentations that will work for participants
project’s technical settings, design the site’s layout, with older, slower computers and low-
define the website’s administrator and contact bandwidth Internet connections. Building
information, create a navigation menu, add and a streaming media server from scratch was
configure five website components, supply text deemed beyond the scope of this effort.
for the Home Page component, and add at least • Focus on end-user design options:Another
one hyperlink to the Home Page’s text. The goal unsolved challenge in Webmaster-R is the

305
Designing Medical Research Web Sites

Figure 4. Summary of the evaluations of the effectiveness of the framework

lack of mechanisms that ensure a website assistant, online help pages were provided
will have an appropriate design for its in- to help overcome blocks; however, these
tended audience. For example, the WPIC pages were never accessed during the ses-
research assistant was able to build a func- sion. Visual cues need to be added to the
tioning website within 90 minutes, but the interface that make clear what pieces of
text-heavy appearance of the constructed information will be viewed “publicly” on
site would have been inappropriate for the website by consumers and caregivers
younger audiences. Other researchers have versus information that it intended for in-
expressed an interest in building sites that ternal purposes only.
meet the requirements of Section 508 of
the Rehabilitation Act to be used by people CONCLUSION
with cognitive and physical impairments.
One potential automated solution to this This chapter introduces a framework for classi-
problem is to ask researchers a series of fying the critical needs of a medical intervention
questions describing the characteristics of research website. Within this framework four main
their audience, and then using one of a set categories have been identified for describing
of pre-defined design templates that is most these critical needs: infrastructure, information,
appropriate for the described audience. interaction, and transaction. Based upon this
• Make the tool more usable for research- underlying framework, the Webmaster-R system
ers: Our goal is to make Webmaster-R easy was built to test the framework and assess the
for a researcher to use without having to feasibility of automating the web development
wade through volumes of documentation. process for medical intervention research websites.
Although improvements to the interface Preliminary evaluations conducted on the system
made the tool more intuitive for the STI indicate that researchers find the tool helpful and
researcher in the second session compared able to meet most of their research objectives. The
to the first, she found some of the terminol- formative evaluation indicates that the design tools
ogy in the directions and labels too techni- are not yet flexible enough to meet some specific
cal. In the session with the WPIC research objectives. Notably, the tools do not yet insure

306
Designing Medical Research Web Sites

that the resulting material is simple enough for of telehealth services in the United Kingdom.
younger audiences, or consumers with cognitive Their work, and this current effort approach the
impairments. On the other hand, in 5 different issue of developing supporting evidence of the
situations, we have been able to reduce the time efficacy of web based interventions from dif-
required to mount a complex automated research ferent perspectives. At the same time, the goal
website, with most of the work done by the medi- is the same – the provision of some degree of
cal researcher his or her self, from three or four medical treatments via new mechanisms. From
months to one to two weeks. Further, the actual this perspective, the ultimate goal of this work is
time required to obtain a polished final website the development of tools and systems that make
has been reduced from three to four months of medical services available and accessible to those
technical work to less than 40 hours. who need them.
While these results are promising, much work
still needs to be done. The current studies suggest
that such a tool can aid in the rapid and inexpensive REFERENCES
development of websites that can be used as the
basis of clinical trials. At this point in time, there Anton, A., Earp, J., Vail, M., Jain, N., Gheen, C.,
is no guarantee that the resulting sites will be & Frink, J. (2007). An analysis of Web site privacy
mounted so as to guarantee HIPPA compliance. policy evolution in the presence of HIPAA. IEEE
It is not clear whether it is possible to insure this. Security and Privacy, 5(1), 45–52. doi:10.1109/
In addition, while a number of scripts have been MSP.2007.7
written to aid in the analysis of the data collected Baer, A., Sariou, S., & Koutsky, L. (2002). Obtain-
via the website, we have treated this process ing sensitive data through the Web: An example of
individually to date so as to answer as many of design and methods. Epidemiology (Cambridge,
the questions researchers have as possible. We Mass.), 13, 640–645. doi:10.1097/00001648-
have not found many questions that could not be 200211000-00007
answered by analysis of the data, but it is not clear
that this process could be fully automated at this Battleson, B., Booth, A., & Weintrop, J. (2001).
point. Finally, each researcher we have worked Usability testing of an academic library Web
with has asked for additional functionality that we site: A case study. Journal of Academic Librari-
have added to the basic model. This process seems anship, 27(3), 188–198. doi:10.1016/S0099-
to accelerate as the sophistication of researchers 1333(01)00180-X
and subjects increases and as the functionality it
Bush, N., Bowen, D., Wooldridge, J., Ludwig,
is possible to realize through browser technology
A., Meischke, H., & Robbins, R. (2004). What
increases. While we have found it easy to extend
do we mean by Internet access? A framework for
the framework to accommodate new functional-
health researchers. Preventing Chronic Disease,
ity, each option makes the system more complex
1, 1–17.
from a design perspective. Ultimately, validation
of the system ultimately requires the same kind Christensen, H., Griffiths, K., & Jorm, A. (2004).
of empirical validation as the studies it endeavors Delivering interventions for depression by using
to support - i.e., research studies of interventions the Internet: Randomized controlled trial. British
based on this tool more successful than research Medical Journal, 328, 265–268. doi:10.1136/
studies using ad hoc site development. bmj.37945.566632.EE
In the introduction we referred to a paper by
May et al. (2005), which examines the provision

307
Designing Medical Research Web Sites

Dew, M., Goycoolea, J., Harris, R., Lee, A., Zomak, Neale, A., & Schwartz, K. (2004). A primer of
R., & Dunbar-Jacob, J. (2004). An Internet-based the HIPAA privacy rule for practice-based re-
intervention to improve psychosocial outcomes searchers. The Journal of the American Board of
in heart transplant recipients and family caregiv- Family Practice, 17(6), 461–465. doi:10.3122/
ers: Development and evaluation. The Journal of jabfm.17.6.461
Heart and Lung Transplantation, 23(6), 745–758.
Rotondi, A. J., Sinkule, J., Haas, G. L., Spring,
doi:10.1016/j.healun.2003.07.002
M., Litschge, C., & Newhill, C. (2007). Design-
Gray, J., Safran, C., Davis, R., Pompilio-Weitzner, ing Web sites for persons with cognitive deficits:
G., Stewart, J., Zaccagnini, L., & Pursley, D. Design and usability of a psychoeducational
(2000). Baby CareLink: Using the Internet and intervention for persons with severe mental ill-
telemedicine to improve care for high-risk in- ness. Psychological Services, 4(3), 202–224.
fants. Pediatrics, 6, 1318–1324. doi:10.1542/ doi:10.1037/1541-1559.4.3.202
peds.106.6.1318
Rotondi, A. J., Sinkule, J., & Spring, M. (2005).
Ikemba, C., Kozinetz, C., Feltes, T., Fraser, C., An interactive Web-based intervention for persons
McKenzie, E., Shah, N., & Mott, A. (2002). Inter- with TBI and their families: Use and evaluation
net use in families with children requiring cardiac by female significant others. The Journal of
surgery for congenital heart disease. Pediatrics, Head Trauma Rehabilitation, 20(2), 173–185.
109, 419–422. doi:10.1542/peds.109.3.419 doi:10.1097/00001199-200503000-00005
Lenert, L., Munoz, R., Stoddard, J., Delucchi, Website: WebMD Consumer Portal (http://www.
K., Bansod, A., Skoczen, S., & Perez-Stable, E. webmd.com).
(2003). Design and pilot evaluation of an Inter-
Wharton, C., Rieman, J., Lewis, C., & Polson,
net smoking cessation program. Journal of the
P. (1994). The cognitive walkthrough method: A
American Medical Informatics Association, 10,
practitioner’s guide. In J. Nielsen & R. L. Mack
16–20. doi:10.1197/jamia.M1128
(Eds.), Usability inspection methods, (pp. 105-
Marshall, W., & Haley, R. (2000). Use of a 140). New York: John Wiley.
secure Internet Web site for collaborative medi-
cal research. Journal of the American Medical
Association, 284, 1843–1849. doi:10.1001/
Key Terms and definitions
jama.284.14.1843
May, C., Finch, T., Mair, F., & Mort, M. (2005). Content Management: The processes and
Toward a wireless patient: Chronic illness, scarce tools that are used to introduce and maintain the
care, and technological innovation in the United content used by an organization or a website.
Kingdom. Social Science & Medicine, 61, 1485– Data Analysis: Related to a website, data
2494. doi:10.1016/j.socscimed.2005.03.008 analysis involves the interpretation of logged data
maintained generally for a website and specifically
McAlindon, T., Formica, M., Kabbara, K., LaVal- related to the needs of researchers.
ley, M., & Lehmer, M. (2003). Conducting clinical Intervention Research: In the case of medi-
trials over the Internet: Feasibility study. British cine, research that is designed to assess the efficacy
Medical Journal, 327, 484–487. doi:10.1136/ of a particular intervention.
bmj.327.7413.484 Navigation System: A series of website fea-
tures that allow a user to move from one part of

308
Designing Medical Research Web Sites

a website to another without error and to keep Usability: The measure of the ability of users
track of where they are. to navigate a website and use it to accomplish
Security: Concerns about authentication, some goal with a minimum of mistakes.
access control and data privacy. With respect Website Design: The process of gathering
to a website, it relates to who is allowed to do requirements, analyzing needs and designing a
what on a website and to the assurance that data website that meets those needs.
is kept private, accessible, and its integrity is
maintained.

This work was previously published in Handbook of Research on Information Technology Management and Clinical Data
Adminstration in Healthcare, edited by A. Dwivedi and J. Grady, pp. 852-869, copyright 2009 by Information Science Refer-
ence (an imprint of IGI Global).

309
310

Chapter 2.4
Designing Web Information
Systems for a Framework-
Based Construction
Vítor Estêvão Silva Souza
Universidade Federal do Espírito Santo, Brazil

Ricardo de Almeida Falbo


Universidade Federal do Espírito Santo, Brazil

Giancarlo Guizzardi
Universidade Federal do Espírito Santo, Brazil

Abstract Introduction

In the Web Engineering area, many methods and The World Wide Web (also referred to as WWW
frameworks to support Web Information Systems or simply Web) was created as a means to publish
(WISs) development have already been proposed. documents and make them available to people in
Particularly, the use of frameworks and container- many different geographical locations. However, the
based architectures is state-of-the-practice. In this advent of the Common Gateway Interface (CGI),
chapter, we present a method for designing frame- in 1993, allowed for authors to publish software
work-based WISs called FrameWeb, which defines instead of documents and for visitors to execute
a standard architecture for framework-based WISs them, producing dynamic results.
and a modeling language that extends UML to build The evolution of Web development technol-
diagrams that specifically depict framework-related ogy and the emergence of high-level languages
components. Considering that the Semantic Web (such as PHP, ASP, JSP, etc.) and platforms (such
has been gaining momentum in the last few years, as Microsoft .NET and Java Enterprise Edition)
we also propose an extension to FrameWeb, called allowed for more complex applications to be built
S-FrameWeb, that aims to support the development on the Web. Soon enough, a handful of large B2C
of Semantic WISs. (business-to-consumer, such as online stores) and
B2B (business-to-business, such as supply chain
DOI: 10.4018/978-1-60566-278-7.ch011

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Designing Web Information Systems for a Framework-Based Construction

management systems) applications were being works are WebML (Ceri et al., 2000), WAE (Con-
deployed on the Internet. allen, 2002), OOWS (Fons et al., 2003), UWE
Thus, the concept of Web Applications (We- (Koch et al., 2000), and OOHDM (Schwabe &
bApps) was born. WebApps consist of a set of Rossi, 1998), among others.
Web pages or components that interact with the Parallel to the academic research, the industry
visitor, providing, storing and processing informa- and the developer community have also proposed
tion. WebApps can be informational, interactive, new technologies to provide a solid Web infra-
transactional, workflow-based, collaborative work structure for applications to be built upon, such
environments, online communities, marketplaces as frameworks and container-based architectures.
or web portals (Ginige & Murugesan, 2001). Using them we can improve productivity at the
In this chapter, however, we focus on a spe- coding phase by reusing software that has already
cific class of Web Applications, called Web-based been coded, tested and documented by third par-
Information Systems (WISs). WISs are just ties. As their use becomes state-of-the-practice,
like traditional information systems, although methods that focus on them during software design
deployed over the Internet or on an Intranet. could provide a smoother transition from models
These systems are usually data-centric and more to source code.
focused on functionality rather than content and This has motivated us to develop a WebE
presentation. Examples are online stores, coopera- design method that focuses on frameworks. The
tive environments, and enterprise management Framework-based Design Method for Web En-
systems, among many others. gineering (FrameWeb) (Souza & Falbo, 2007)
Although many Software Engineering prin- proposes a basic architecture for developing We-
ciples have long been established before the cre- bApps and a UML profile for a set of design models
ation of the Web, first-generation WebApps were that brings concepts used by some categories of
constructed in an ad-hoc manner, with little or no frameworks, which are applied in container-based
concern for them. However, with the increase of architectures as well.
complexity of the WebApps, which is especially Meanwhile, many researches have been direct-
true for WISs, the adoption of methodologies and ed to the construction of what is being considered
software processes to support the development the future of the WWW: the Semantic Web. Coined
team becomes crucial. by Berners-Lee et al. (2001), the term represents an
Thus, a new discipline and research field was evolution of the current WWW, referred by some
born. Web Engineering (or WebE) can be defined as the “Syntactic Web”. In the latter, information is
as “the establishment and use of engineering prin- presented in a way that is accessible only to human
ciples and disciplined approaches to the develop- beings, whereas in the former data is presented
ment, deployment and maintenance of Web-based both in human-readable and machine-processable
Applications” (Murugesan et al., 1999, p. 2). Press- formats, in order to promote the development of
man (2005) complements this definition stating software agents that would help users carry out
that WebE borrows many conventional Software their tasks on the Web.
Engineering fundamental concepts and principles However, for Berners-Lee’s vision to become
and, in addition, incorporates specialized process a reality, Web authors and developers must add
models, software engineering methods adapted to semantic annotations to their Web Applications.
the characteristics of this kind of application and This is neither an easy nor a small task and sup-
a set of enabling technologies. port from tools and methods is needed. Thus,
In this field, a lot of methods and modeling an extension of FrameWeb was proposed. The
languages have been proposed. Some well known Semantic FrameWeb (S-FrameWeb) (Souza et al.,

311
Designing Web Information Systems for a Framework-Based Construction

2007) incorporates into the method activities and As with conventional software engineering,
guidelines that drive the developer in the defini- a WebE process starts with the identification of
tion of the semantics of the WISs, resulting in a the business needs, followed by project plan-
“Semantic Web-enabled” application. ning. Next, requirements are detailed and mod-
The objective of this chapter is to discuss the eled, taking into account the analysis and design
current research state regarding Web Engineering perspective. Then the application is coded using
(methods and modeling languages), frameworks tools specialized for the Web. Finally, the system
and the Semantic Web, and to present FrameWeb, is tested and delivered to end-users (Pressman,
and its extension S-FrameWeb, as a new method 2005).
based on best practices for the development of Considering that, in general, the platform in
Web-based Information Systems. We close the which the system will run is not taken into account
chapter by presenting future trends and opportu- before the design phase of the software process, de-
nities for this research. veloping a WebApp would be just like developing
any other application up to that phase. However,
many differences between Web Engineering and
Background Conventional Software Engineering have been
identified by researchers and practitioners (Ahmad
Web Engineering was born from the need to et al., 2005), such as sensitivity to content, short
apply Software Engineering principles to the time frames for delivery, continuous evolution,
construction of WebApps, adapting them to the focus on aesthetics, etc (Pressman, 2005).
application’s size, complexity and non-functional This has motivated researchers to propose
requirements. A lot of research on methods and different methods, modeling languages and
modeling languages has already been conducted, frameworks for Web Engineering. The amount
providing an extensive background for new re- of propositions is quite vast, demonstrating that
search. academics and practitioners have not yet elected
Meanwhile, companies and independent devel- a standard concerning Web development. In this
opers create frameworks and propose container- subsection we briefly present some methods,
based architectures to promote reuse and improve while the following subsections focus on model-
productivity while maintaining good design ing languages and frameworks.
principles. Furthermore, research on the Semantic Web Application Extension (WAE) (Conallen,
Web has been pointing out some directions on 2002) defines an iterative and incremental software
what the Web may become in the future. process, centered on use cases and based on the
This section discusses the current state-of-the- Rational Unified Process (Krutchen, 2000) and
art and state-of-the-practice on Web Engineering, the ICONIX Unified Process (Rosenberg & Scott,
modeling languages for WebE, frameworks for de- 1999). It proposes activities such as business analy-
velopment of WebApps and the Semantic Web. sis, project planning, configuration management
and an iterative process that includes the usual
Web Engineering software development cycle from requirement
gathering to deployment.
Web Engineering (or WebE) uses scientific, OOWS (Object Oriented Web Solution) (Fons
engineering, and management principles and et al., 2003) is an extension of the OO-Method
systematic approaches to successfully develop, (Pastor et al., 2001) for WebApp specification and
deploy, and maintain high-quality WebApps development. It divides the software process in
(Murugesan et al., 1999). two main steps: conceptual modeling and solution

312
Designing Web Information Systems for a Framework-Based Construction

development. In the conceptual modeling step, into activities, which in turn defines sets of work
the system specification is obtained by using products to be built.
conceptual models. For that, OOWS introduces Díaz et al. (2004, p. 650) also define the hy-
new models for representing navigational and permedia paradigm as one that “relies on the idea
presentational characteristics of web applications. of organizing information as a net of interrelated
In the solution development step, the target plat- nodes that can be freely browsed by users selecting
form is determined, and a specific architectural links and making use of other advanced naviga-
style is chosen. Then, a set of correspondences tion tools, such as indexes or maps”. We consider
between abstraction primitives and the elements hypermedia methods quite different than methods
that implement each tier of the architectural style for the development of WISs, as they focus on
are applied in order to automatically obtain the content and navigational structures instead of
final system (Pastor et al., 2003). functionality and seem to be better suitable for
The UML-based Web Engineering (UWE) information-driven WebApps.
(Koch et al., 2000) is a development process for Although hypermedia development methods
Web applications with focus on systematic design, are not on our focus, it is worthwhile to cite
personalization and semi-automatic generation. It OOHDM (Object Oriented Hypermedia Design
is an object-oriented, iterative and incremental ap- Method) (Schwabe & Rossi, 1998), a well-known
proach based on the Unified Modeling Language method that is representative of hypermedia
(UML) and the Unified Software Development methods. It was born from the need to represent
Process (Jacobson et al., 1999). The notation used hypermedia structures such as links, text-based in-
for design is a “lightweight” UML profile. The terfaces and navigation, and more recently has also
process is composed by requirement analysis, been applied to Web development. For instance, an
conceptual navigation and presentation design, extension of this method, called OOHDM-Java2
supplemented with task and deployment model- (Jacyntho et al., 2002), was proposed, which
ing and visualization of Web scenarios (Koch & consists of a component-based architecture and
Kraus, 2002). an implementation framework for the construction
Lee & Shirani (2004) propose a component- of complex WebApps based on modular archi-
based methodology for WebApp development, tectures (e.g. Java EE). The OOHDM process is
which is divided in two major parts: component divided into five steps: requirements gathering,
requirements analysis and component specifica- conceptual design, navigational design, abstract
tions. Analysis begins identifying the required interface design and implementation.
component functions and is followed by a com- During our research we have also found sev-
parison with the functions available in existing eral other methodological approaches that target
components. The component specification phase specific contexts or scenarios, such as:
has three activities: rendering specification, inte-
gration specification and interface specification. • The Business Process-Based Methodology
The Ariadne Development Method (Díaz et (BPBM) (Arch-int & Batanov, 2003),
al., 2004) proposes a systematic, flexible, inte- which blends advantages of the structured
grative and platform-independent process for and object-oriented paradigms for identi-
specification and evaluation of WebApps and fying and designing business components.
hypermedia systems. This process is composed of The central idea of business component
three phases: conceptual design, detailed design modeling is reusability of elementary
and evaluation. Each phase is further subdivided units, which are business activities. An

313
Designing Web Information Systems for a Framework-Based Construction

elementary unit that represents an atomic guidelines for layout and navigation modeling
changeable business process can be imple- from requirements specification through design.
mented with a portable set of Web-based Models, like the navigation diagram, the class
software components; diagram and the component diagram (the last two
• The Internet Commerce Development specific for the web tier), use WAE to represent
Methodology (ICDM) (Standing, 2002), Web components such as screens, server pages,
which is focused on the development of client pages, forms, links and many more.
B2C e-commerce applications, emphasiz- The UML-based Web Engineering (UWE)
ing not only technical aspects, but also (Koch et al., 2000) also defines a UML profile.
strategic, business and managerial aspects. Based on class and association elements, it defines
new elements to describe Web concepts, such as
Some of the methods presented above also navigation, indexes, guided tours, queries, menus
propose a modeling language that better suits its and many others.
purposes, such as WAE and UWE . In the next Another method that defines a modeling lan-
subsection, some of them are briefly presented. guage based on UML is OOWS (Fons et al., 2003).
For the construction of the navigational model,
Modeling Languages for UML packages represent navigational contexts
Web Engineering and form a directed graph where the arcs denote
pre-defined valid navigational paths. Each context
Modeling languages define notations to be used on is further modeled using a class diagram to show
the creation of abstract models to solve problems. the navigational classes that form them.
The Unified Modeling Language (UML) (Booch Not all modeling languages are UML-based.
et al., 2005), for instance, is a modeling language WebML (Ceri et al., 2000) is an example. It allows
that defines on its metamodel standardized nota- developers to model WebApp’s functionalities in
tions for different kinds of models, such as class a high level of abstraction, without committing to
diagrams and use case diagrams. However, UML any architecture in particular. WebML is based on
does not define when and to which purpose each XML, but uses intuitive graphical representations
model should be used. that can easily be supported by a CASE tool. Its
Hence, methodologies usually present their XML form is ideal for automatic generation of
own modeling language or, as is most commonly source code, producing Web applications auto-
seen, use and extend UML, defining a UML Pro- matically from the models.
file. For this purpose, UML includes extension Methods and modeling languages aid de-
mechanisms, such as stereotypes (definition of velopers mostly during analysis and design of
a new model element based on an existing one), information systems. However, one can also find
tagged values (attachment of arbitrary textual in- tools that focus on the implementation phase. In
formation to elements using label/value pairs) and the next subsection, we discuss frameworks that
constraints (semantic specification for an existing have been extensively used for the development
element, sometimes using a formal language). of WISs.
Based on these extension mechanisms, Conal-
len (2002) proposed the Web Application Exten- Frameworks for Web Development
sions (WAE), which extends UML to provide Web-
specific constructs for modeling WebApps. WAE WISs have very similar architectural infrastruc-
also advocates the construction of a new model, ture. Consequently, after the first systems started
the User Experience (UX) Model, which defines to be built, several frameworks that generalize

314
Designing Web Information Systems for a Framework-Based Construction

this infrastructure were developed to be reused • Dependency Injection frameworks;


in future projects. In this context, a framework is • Aspect-Oriented Programming
viewed as a code artifact that provides ready-to-use frameworks;
components that can be reused via configuration, • Authentication & Authorization
composition or inheritance. When combined, these frameworks.
frameworks allow large-scale n-tier WISs to be
constructed with less coding effort. Front Controller Frameworks
Putting together several of these frameworks
can produce what we call a container-based ar- MVC stands for Model-View-Controller (Gamma
chitecture. A container is a system that manages et al., 1994). It is a software architecture that was
objects that have a well-defined life cycle. A developed by the Xerox PARC for the Smalltalk
container for distributed applications, such as the language in 1979 (Reenskaug, 1979) and has found
applications servers for the Java Enterprise Edi- great acceptance by Web developers. When ap-
tion (Shannon, 2003), manage objects and offer plied to the Web, the MVC architecture is adapted
services such as persistence, transaction manage- and receives the name “Front Controller” (Alur
ment, remoting, directory services, etc. et al., 2003, p.166). Both terms are used indistin-
The use of these frameworks or container- guishably by Web developers.
based architectures has a considerable impact in The Front Controller architecture is depicted
the development of a WIS. Since it is possible to in Figure 1. When structured in this architecture, a
find many frameworks for the exact same task, WebApp manages all requests from clients using
we categorized them according their objectives an object known as Front Controller. Based on a
into the following classes: customizable configuration, this object decides
which class will respond to the current request
• Front Controller (or MVC) frameworks; (the action class). Then, following the Command
• Decorator frameworks; design pattern (Gamma et al., 1994), it instantiates
• Object/Relational Mapping frameworks; an object of that class and delegates the control

Figure 1. General architecture of a Front Contoller framework

315
Designing Web Information Systems for a Framework-Based Construction

to it, expecting some kind of response after its tier and should delegate business and persistence
execution. Based on that response, the controller tasks to components on appropriate tiers.
decides the appropriate view to present as result, Only for the Java platform, for instance, there
such as a web page, a report, a file download, are more than 50 MVC frameworks. Some of
among other possibilities. the most popular are Struts1, Spring MVC2 and
One of these possibilities is using a template Tapestry3.
engine that defines a template language that is
usually more suitable for the view layer than the Decorator Frameworks
usual dynamic web technology (such as JSP, ASP
or PHP). The template language is usually simpler, Decorator frameworks automate the otherwise
making it possible for Web Designers without tedious task of making every web page of the
specific programming skills to build them. Also, site have the same layout, meaning: header,
they tend to help developers not to break the MVC footer, navigational bar, color schemes and other
architecture by restricting what can be done in the graphical layout elements produced by a Web
template language (e.g. can not directly connect design team. Figure 2 shows how a decorator
to a database from a template). framework works.
MVC Frameworks usually provide the front They work like the Decorator design pattern
controller, a super-class or interface for action (Gamma et al., 1994), providing a class that inter-
classes, several result types and a well defined cepts requests and wraps their responses with an
syntax for the configuration file. The template appropriate layout before it is returned to the client.
engine is a separate tool, but the framework usu- It also provides dynamic selection of decorators,
ally provides integration to it. Note that on n-tier making it easy to create alternate layouts, such
applications, this framework belongs to the Web as a “print version” of the page. Examples of this
kind of framework are SiteMesh4 and Tiles5.

Figure 2. The process of decoration of websites

316
Designing Web Information Systems for a Framework-Based Construction

Figure 3. Persistence of objects using an ORM framework

Object/Relational Mapping Frameworks The use of ORM frameworks is not restricted


to Web applications and has been in use for quite
Relational Database Management Systems (RD- some time now in all kinds of software. The most
BMS) have long been the de facto standard for popular Java ORM framework is Hibernate6.
data storage. Because of its theoretical founda- Other well-known frameworks are Java Data
tions (relational algebra) and strong industry, Objects7, Apache Object Relational Bridge8 and
even object oriented applications use it for object Oracle Toplink9.
persistence, giving rise to a “paradigm mismatch”
(Bauer & King, 2004): tables, rows, projection Dependency Injection Frameworks
and other relational concepts are quite different
from a graph of interconnected objects and the Object-oriented applications are usually built in
messages they exchange. tiers, each of which having a separate responsibil-
Among the many options to deal with this ity. According to Fowler (2007), when we create
problem, there is the Object/Relational Mapping classes that depend on objects of other classes
(ORM) approach, shown in Figure 3, which is the to perform a certain task, it is preferred that the
automatic and transparent persistence of objects to dependent class is related only to the interface of
tables of a RDBMS using meta-data that describe its dependencies, and not to a specific implemen-
the mapping between both worlds (Bauer & King, tation of that service.
2004). Instead of assembling a string with the Creational design patterns, such as Factory
SQL command, the developer provides mapping Method, Abstract Factory and Builder (Gamma et
meta-data for the classes and call simpler com- al., 1994), help implementing this good practice
mands, such as save(), delete() or retrieveById(). in programming, known today as “programming
An object-oriented query language can also be to interfaces, not implementations” (Schmidt,
used for more complex retrievals. 2007). For instance, if a service class depends
on a data access class, it does not need to know

317
Designing Web Information Systems for a Framework-Based Construction

Figure 4. Dependency injection using a framework

how the data access class will perform its duty, Aspect-Oriented Programming
but only what it will do and what method should Frameworks
be called for the job to be done.
Dependency Injection (DI) frameworks allows The Aspect-Oriented paradigm is based on the
the developer to program to interfaces and specify concept of separation of concerns: the idea is
the concrete dependencies as meta-data in a con- to separate different concerns of a system to be
figuration file. When a certain object is obtained treated separately, thus reducing the complexity
from the DI framework, all of its dependencies are of development, evolution and integration of
automatically injected and satisfied. An abstract software (Resende & Silva, 2005). Although it
example is shown in Figure 4: when the client asks concerns the whole development process, its big-
for an instance of SomeClass, the DI framework gest influence is at the coding phase, with Aspect
first satisfies SomeClass’ dependencies and deliv- Oriented Programming (AOP).
ers the object with all dependencies fulfilled – in Once a cross-cutting concern is identified
the example, an instance of DependencyClass. (e.g.: logging, transaction management), instead
These frameworks are also known as Inversion of repeating similar code in different points, the
of Control (IoC) frameworks, since the control functionality can be implemented in a single
(who creates the objects) is removed from the place, becoming an aspect. Then, the different
dependent classes and given to the framework. places where that aspect should be applied are
As well as ORM frameworks, DI frameworks are identified (these are called pointcuts) and, before
not used exclusively for WebApps, although they the code is executed, a process called weaving is
tend to integrate more seamlessly with applications conducted to automatically spread the aspect all
that run inside containers, just like a WebApp runs over the code.
inside a Web server. Lots of frameworks provide The weaving can be conducted by an AOP
this service, including Spring Framework, Pico- framework during runtime or by an AOP compiler
Container10, Apache Hivemind11, etc. during compilation time. Many infrastructure
concerns that are usual in Web applications are

318
Designing Web Information Systems for a Framework-Based Construction

good candidates for this separation, making AOP for the semi-automation of many tasks that are
frameworks very popular. One example, depicted conducted on the Web.
in Figure 5, is that of transaction management. An In order for the software agents to reason with
AOP framework can make all business methods the information on the Web (reasoning meaning
transactional with few configuration steps, avoid- that the agents are able to understand it and take
ing the effort of repeatedly implementing the same sensible actions according to a predefined goal),
logic in all of them. web pages have to be presented also in a machine-
Some well-known AOP frameworks for the readable form. The most usual way for this is
Java platform are AspectJ12, Spring Framework annotating the pages using formal knowledge
and JBoss AOP13. representation structures, such as ontologies.
An ontology is an engineering artifact used
Authentication & Authorization to describe a certain reality, plus a set of explicit
Frameworks assumptions regarding the intended meaning of
its vocabulary words (Guarino, 1998). Along
Another common concern of Web information with ontology representation languages such as
systems is that of guaranteeing the security of OWL (W3C, 2007a), they are able to describe
the information. This is usually done by two information from a website in formal structures
different procedures: authentication (verifying if with well-defined inference procedures that al-
an access key is valid to access the application) low software agents to perform tasks such as
and authorization (verifying the level of access consistency checking, to establish relationships
of the authenticated user and what she is allowed between terms and to systematically classify and
to do). infer information from explicitly defined data in
Being such an important task, frameworks were this structure.
created to guarantee its proper execution. They can Designing an ontology is not a straightforward
be configured to support many different “auth” task. There are many methodologies for their con-
methods, using, as usual, meta-data and configura- struction (Gomez-Perez et al., 2005) and attention
tion files. Some well-known auth frameworks for has to be given to the selection of concepts, their
the Java platform are Acegi Security for Spring14, properties, relationships and constraints. However,
Apache Cocoon Authentication15 and the Java after the ontology is built, the annotation of static
Authentication and Authorization Services16. Web pages with languages such as OWL becomes
In spite of frameworks being much used, there a simple task, especially with the aid of tools, such
is no Web Engineering method that explores their as OILEd17 and Protégé18.
use in the design phase of the software process. However, several websites have their Web
To fill this gap, we proposed FrameWeb, a Frame- pages dynamically generated by software retriev-
work-based Design Method for Web Engineering ing information from data repositories (such as
(Souza & Falbo, 2007), which is presented later relational databases) during runtime. Since these
in this chapter. pages cannot be manually annotated prior to
their presentation to the visitor, another approach
The Semantic Web has to be taken. Two approaches that have been
proposed are dynamic annotation and semantic
The Semantic Web is being proposed as an Web services.
evolution of the current WWW, in which infor- The former works by recognizing whether the
mation is provided both in human-readable and request belongs to a human or a software agent,
computer-processable formats, in order to allow generating the proper response depending on the

319
Designing Web Information Systems for a Framework-Based Construction

Figure 5. example of application of AOP using an AOP runtime framework

320
Designing Web Information Systems for a Framework-Based Construction

client: in the first case, a HTML human-readable Design phase. The last two activities are Abstract
Web page; in the second, a document written in Interface Design and Implementation.
an ontology specification language containing
meta-data about the information that would be Some Considerations
displayed in the HTML version. Although the
solution seems appropriate, many aspects still need In our research, we haven’t found a method focused
to be addressed, such as: how are the agents sup- on the use of frameworks for the construction of
posed to find the Web page? How will they know WISs nor for the development of Semantic Web
the correct way to interact with it? For instance, applications. In the next sections, we present
how will they know how to fill in an input form FrameWeb, our proposal for the design of frame-
to submit to a specific request? work-based WISs, and its extension, S-FrameWeb,
The latter approach is based on Web Services, which incorporate into the method activities and
which are software systems designed to support in- guidelines that drive the developer in the defini-
teroperable machine-to-machine interaction over tion of the semantics of the WISs, resulting in a
a network (W3C, 2007b). Web Services provide “Semantic Web-enabled” application.
a nice way for software agents to interact with
other systems, requesting services and process-
ing their results. If semantic information is added FrameWeb
to the services, they could become interpretable
by software agents. Meta-data about the service FrameWeb is a design method for the construc-
are written in a markup language, describing its tion of Web-based Information Systems (WISs)
properties and capacities, the interface for its based on frameworks. The main motivations for
execution, its requirements and the consequences the creation of this method were:
of its use (McIlraith et al., 2001). Many tasks are
expected to be automated with this, including 1. The use of frameworks or similar container-
service discovery, invocation, interoperation, se- based architectures has become the de facto
lection, composition and monitoring (Narayanan standard for the development of distributed
& McIlraith, 2002). applications, especially those based on the
As the research on the Semantic Web pro- Web;
gresses, methods are proposed to guide developers 2. There are many propositions in the area of
on building “Semantic Web-enabled” applications. Web Engineering, including methodolo-
An example of this is the Semantic Hypermedia gies, design methods, modeling languages,
Design Method (SHDM) (Lima & Schwabe, frameworks, etc. However, we haven’t
2003). Based on OOHDM (Schwabe & Rossi, found one that deals directly with the par-
1998), SHDM is a comprehensive model-driven ticularities that are characteristic of the use
approach for the design of Semantic WebApps. of frameworks;
SHDM’s process is divided in 5 activities. In 3. Using a method that fits directly into the
the first step, Requirements Gathering, require- architecture chosen for the implementation
ments are gathered in the form of scenarios, promotes a greater agility to the software
user interaction diagrams and design patterns. process, which is something that is desired
The next phase, Conceptual Design, produces a in most Web projects.
UML-based conceptual model, which is enriched
with navigational constructs in the Navigational In general, FrameWeb assumes that certain
types of frameworks will be used during the

321
Designing Web Information Systems for a Framework-Based Construction

Figure 6. A simple software process suggested by FrameWeb

implementation, defines a basic architecture for Also, as mentioned earlier, one of the motiva-
WISs and proposes design models that are closer tions for the creation of FrameWeb is the demand
to the implementation of the system using these for agility that surrounds Web projects. Thus,
frameworks. although the method brings more agility especially
Being a design method, it doesn’t prescribe a to the design and coding phases, developers are
complete software process. However, it suggests advised to follow principles of agility during
the use of a development process that includes requirements analysis, as the ones proposed by
the following activities, as presented in Figure 6: Agile Modeling (Ambler & Jeffries, 2002).
requirements elicitation, analysis, design, coding, The main contributions of the method are for
testing and deployment. For a more systematic the Design phase: (i) the definition of a basic
usage of the method, it also suggests that, during architecture that divides the system in layers
Requirement Elicitation and Analysis, use case with the purpose of integrating better with the
diagrams are used to model requirements and frameworks; (ii) a UML profile for the construc-
class diagrams are used to represent the concep- tion of four different design models that bring the
tual model. concepts used by the frameworks to the design
stage of the software process.

Figure 7. A simplified use case diagram for LabES Portal

322
Designing Web Information Systems for a Framework-Based Construction

Figure 8. Conceptual model for the User Control module of the LabES Portal

The Coding phase is facilitated by the use of impacts on Testing and Deployment, but these
frameworks, especially because design models are yet subject to study and research.
show components that can be directly related Throughout the next subsections we detail
to them. The use of frameworks can also have FrameWeb’s basic architecture and its UML pro-

Figure 9. Conceptual model of the Item Control module of the LabES Portal

323
Designing Web Information Systems for a Framework-Based Construction

file. Examples diagrams were taken from the de- tectures of the system as well as the development
velopment of a portal for the Software Engineering of structural and behavioral models built based
Lab (LabES) of the Federal University of Espírito on the models developed in the previous phases,
Santo State using FrameWeb. Figure 7 shows its but that now consider the specific characteristics
use case diagram, simplified for brevity. of the chosen implementation platform.
The “LabES Portal” was proposed to provide FrameWeb defines a logical architecture for
a better interaction with the Software Engineer- WISs based on the architectural pattern Service
ing community. This WIS has a basic set of ser- Layer (Fowler, 2002, p. 133). As depicted in
vices providing information about current LabES Figure 10, the system is divided in three main
projects, areas of interest, publications and other layers: presentation logic, business logic and data
material available for download. Figures 8 and access logic.
9 show the conceptual models produced during The first layer concerns the graphical user
Analysis. interfaces. The View package contains the Web
Basically, the portal makes a collection of items pages, style sheets, images, client-side scripts,
available. These items can be organized in projects templates and everything else related to the exhi-
and subprojects or belong to the lab in general. bition of information to the user. The Controller
Publications (papers, books, book chapters and package encompasses action classes and other files
academic works) and generic materials can be related to the Front Controller framework. These
published in the portal. Items are also related to two packages are mutually dependent, since View
users (responsible user, editing users) and areas elements send user stimuli to Controller classes
of interest. while these process the response using pages,
models and other View components.
Framework-Based The business logic is implemented in the
WebApp Architecture second layer, divided in two packages: Domain
and Application. The former contains classes that
The Design activity, traditionally executed after represent concepts of the problem domain identi-
requirement elicitation and analysis, has as purpose fied and modeled by the class diagrams during
the description of the logical and physical archi- analysis and refined during design. The latter

Figure 10. FrameWeb’s basic architecture for WISs

324
Designing Web Information Systems for a Framework-Based Construction

has the responsibility of implementing the use frameworks presented earlier in this chapter. Each
cases defined in the requirements specification, package contains classes or other elements that
providing a service layer independent of the user integrate with these frameworks and, to model all
interface. The Application classes deal directly these elements, FrameWeb proposes a modeling
with Domain objects to implement system func- language based on the UML, which is presented
tionality and, thus, this dependency is represented next.
in the diagram.
The Controller package, on the presenta- Modeling Language
tion layer, depends on the Application package
since it mediates the user access to the system During design, besides specifying the system ar-
functionalities. User stimuli coming from View chitecture, the artifacts that will be implemented
are transformed by the Controller’s classes in by the programmers on the coding phase should
method calls to classes in the Application pack- be modeled. Since FrameWeb is based on the
age. Controller and View have also dependency frameworks presented earlier, we felt the need
relationships with Domain, but this is tagged as for a modeling language that would represent the
weak to denote low coupling: Domain objects are concepts that are present in these frameworks.
used only for exhibition of data or as parameters Following the same approach as other modeling
on method invocations between one package and languages such as WAE and UWE, FrameWeb uses
another, i.e., the presentation layer does not have UML’s lightweight extensions to represent typical
the right to alter domain entities. Web and framework components, creating a UML
The third and last layer regards data access profile that is used for the construction of four kinds
and has only the Persistence package. This pack- of diagrams, which are presented in the following
age is responsible for the storage and retrieval of subsections: domain model, persistence model,
persistent objects in long-term duration media, navigation model and application model.
such as databases, file systems, naming services,
etc. In the case of FrameWeb, it expects the use Domain Model
of an ORM framework through the Data Access
Object (DAO) pattern (ALUR et al., 2003, p. 462). The domain model is a UML class diagram that
The DAO pattern adds an extra abstraction layer, represents domain objects and their persistence
separating the data access logic of the chosen per- mapping to a relational database. This model is
sistence technology in a way that the Application used by the programmers to implement the classes
classes do not know which ORM framework is of the Domain package. FrameWeb suggests its
being used, allowing for its replacement, if nec- construction in two steps:
essary. It also facilitates unit testing, as one can
provide mock DAOs for the Application classes 1. Adapt the conceptual model produced
to be tested alone. during the Requirement Analysis phase to
As we can see in Figure 10, the Application FrameWeb’s architecture and to the chosen
package depends on the Persistence package to platform of implementation. This requires
retrieve, store and delete domain objects as the choosing data types for attributes, defining
result of use case execution. Since the Persistence navigabilities of the associations, promoting
package works with Domain objects, a weak de- attributes to classes (if necessary), etc.;
pendency is also portrayed in the figure. 2. Add persistence mappings.
This architecture provides a solid base for
the construction of WISs based on the types of

325
Designing Web Information Systems for a Framework-Based Construction

Persistence mappings are meta-data that allow Table 1 describes the possible O/R mappings
ORM frameworks to convert objects in memory for the domain model. For each mapping, the
to tuples in Relational Data Base Management table presents the extension mechanism used and
Systems and vice-versa. Mappings are added to the what are its possible values or syntax. None of the
domain model using stereotypes and constraints mappings is mandatory and most of them have
that guide developers in the configuration of the sensible defaults, reducing the amount of elements
ORM framework during implementation. Despite that have to be modeled. The default values are
the fact that these mappings are more related to shown in the third column in boldface.
persistence than domain, they are shown in this The Domain Model for the User Control
model because the classes that are mapped and module of sLabES Portal is shown in Figure 11.
their attributes are shown here. According to the default values, all classes are

Table 1. Possible OR mappings for the Domain Model

Mapping Extension Possible Values


<<persistent>>
If the class is persistent, transient or mapped (not persistent itself, but its
Class stereotype <<transient>>
properties are persistent if another class inherits them)
<<mapped>>
table=name
Name of the table in which objects of a class will be persisted Class constraint
(default: class’ name)
<<persistent>>
If an attribute is persistent or transient Attribute stereotype
<<transient>>
null
If an attribute can be null when the object is persisted Attribute constraint
not null
Date/time precision: store only the date, only the time or both (time- precision = (date | time | time-
Attribute constraint
stamp) stamp)
If the attribute is the primary-key of the table Attribute stereotype <<id>>
How the ID attribute should be generated: automatically, obtained in a table, generation = (auto | table | iden-
Attribute constraint
use of IDENTITY column, use of SEQUENCE column or none tity | sequence | none)
If the attribute represents the versioning column. Attribute stereotype <<version>>
If an attribute should be stored in a large object field (e.g.: CLOB,
Attribute stereotype <<lob>>
BLOB)
column=name
Name of the column in which an attribute will be persisted Attribute constraint (defaults to the attribute’s
name)
Size of the column in which an attribute will be persisted Attribute constraint size=value
If the association should be embedded (instead of having its own table, the
Attribute stereotype <<embedded>>
associated child class’ attributes are placed in the parent’s table)
<<union>>
Inheritance mapping strategy: one table for each class using UNION, one
Inheritance stereotype <<join>>
table for each class using JOIN or single table for the entire hierarchy
<<single-table>>
Type of collection which implements the association: bag, list, set or
Association constraint collection = (bag | list | set | map)
map
Order of an association’s collection: natural ordering (implemented in order = (natural | column names
Association constraint
code) or order by columns (ascending or descending) [asc | desc])
Cascading of operations through the association: nothing, persists, merges, cascade = (none | persist | merge
Association constraint
deletions, refreshs or all | remove | refresh | all)
Association fetching strategy: lazy or eager. Association constraint fetch = (lazy | eager)

326
Designing Web Information Systems for a Framework-Based Construction

Figure 11. Domain Model for the User Control module of LabES Portal

persistent and class and attribute names are used Persistence Model
as table and column names respectively.
As we can see in the diagram, attributes have As mentioned before, FrameWeb indicates the use
received mappings such as nullability and size. of the DAO design pattern (ALUR et al., 2003, p.
The birthDate attribute was mapped as date-only 462) to the construction of the data access layer.
precision. The recursive association in Area was Thus, the persistence model is a UML class dia-
configured to be sorted naturally (will be imple- gram that represents DAO classes responsible for
mented in the programming language) and to the persistence of the domain classes. Therefore,
cascade all operations (e.g. if an area is deleted, it guides the implementation of the classes from
all of its subareas are automatically deleted). the Persistence package. FrameWeb suggests three
None of the classes have ID or version at- steps for its construction:
tributes because they are inherited from a utility
package, as shown in Figure 12. The mapped 1. Model the interface and concrete imple-
stereotype indicates that DomainObjectSupport mentation of the base DAO (an example is
and HibernatePersistentObject are not persistent shown in Figure 12);
entities, but their subclasses, which are entities, 2. Define which domain classes need basic
inherit not only their attributes but also their O/R persistence logic and create a DAO interface
mappings. All domain classes in LabES Portal are and implementation for each one;
said to extend HibernatePersistentObject, inherit- 3. For each DAO, evaluate the need of specific
ing, thus, the UUID19, the persistence ID and the database queries, adding them as operations
version attribute. in their respective DAOs.
The parameters I and V are generic, allowing
for the user to choose the type of ID and version The persistence model presents, for each
attributes. HibernateBaseDAO is a base class for domain class that needs data access logic, an
data access objects, described in the persistence interface and a concrete DAO that implements
model, discussed in the next subsection. the interface. The interface has to be unique and

327
Designing Web Information Systems for a Framework-Based Construction

Figure 12. Utility classes for persistence

defines the persistence methods for a specific that all public methods are defined in the interface
domain class. One concrete class is modeled for and implemented in the concrete class.
each persistence technology used. Figure 12 shows the interface and implementa-
To avoid repeating in each DAO operations tion using Hibernate ORM framework, designed
that are common in all of them (e.g.: save, delete, for the LabES Portal project. Both interface and
retrieve by ID, etc.), a Base DAO (interface and class and declared using generic types, leaving to
implementation class) is modeled in a utility their subclasses to specify which class is being
package. Automatically all DAO interfaces inherit persisted and what is the type for its ID attribute.
from the BaseDAO interface and the same hap- The Base DAO defines methods to retrieve all
pens with concrete implementations, without the persistent entities of a given class, retrieve an
need to explicitly state that in the diagram. Also, entity given its ID, save and delete an entity.
to avoid repeating methods in the interface and As stated before, all public methods modeled in
implementations, the designer can choose to dis- HibernateBaseDAO are inferred to be defined in
play them in one of the two only and it is inferred the BaseDAO interface.

Figure 13. Persistence model of the User Control module of LabES Portal

328
Designing Web Information Systems for a Framework-Based Construction

Figure 13 shows the modeling of four DAOs dicate hyperlinks while composition associations
from the LabES Portal project, for the persis- between pages and forms denote the presence of
tence of the classes in the User Control module. the form in that page.
AreaDAO and UserTypeDAO are simple, as they In HTML forms, attributes represent the form
inherit all basic operations from the Base DAO fields and their types follow the HTML standard
and don’t need to define any extra ones. The for types of fields (e.g.: input, checkbox, etc.) or
other two define extra operations. For example, the names of the JSP tags used by the framework
UserDAO defines an operation to retrieve all users (e.g., for Struts2, textfield, checkbox, checkbox-
that have a given area of interest. This is necessary list, etc.).
because there is no navigability from Area to User The action class is the main component of
(see Figure 11) and the “Manage Area” use case the model. Its dependency associations show the
needs to prevent an area from being deleted if it control flow when an action is executed. Table
is associated with any user. 3 lists the different meanings of this kind of as-
As we can see, the persistence model does sociation, depending on the components that are
not define any UML extensions to represent the connected by it. Dependencies that are navigable
concepts that are needed to implement the data towards an action class represent method calls,
access layer, but only some rules that make this while the others represent results from the action
modeling simpler and faster. execution.
The attributes of the action class represent in-
Navigation Model put and output parameters relevant to that action.
If there is a homonymous attribute in an HTML
The navigation model is a UML class diagram form being submitted to the action, it means that
that represents different components that form the data is injected by the framework in the action
the presentation layer, such as Web pages, HTML class (input parameter). Likewise, when one of the
forms and action classes from the Front Controller result pages/templates show an attribute with the
framework. Table 2 shows the UML stereotypes same name of an attribute of the action class, this
used by the different elements that can be repre- indicates that the framework makes this informa-
sented in a navigation model. This model is used tion available for the output.
by developers to build classes and components When an action is executed, the framework
of the View and Controller packages. will execute a default action method or allow/
For Web pages and templates, the attributes request the explicit definition of which method
of the classes represent information from the to execute. In the latter case, the designer must
domain that is supposed to be displayed in the specify which method is being executed using
page. Dependency relationships between them in- the constraint {method=method-name} in the

Table 2. UML stereotypes used in the navigation model

Stereotype What it represents


(none) An action class, to which the Front Controller framework delegates the execution of the action.
<<page>> A static or dynamic Web page.
<<template>> A template that is processed by a template engine and is transformed into a Web page.
<<form>> A HTML form.
<<binary>> Any binary file that can be retrieved and displayed by the browser (e.g.: images, reports, documents, etc.).

329
Designing Web Information Systems for a Framework-Based Construction

Table 3. Dependency associations between an action class and other elements

From To What it represents


Page / template Action class A link in the page/template that triggers the execution of the action.
Form Action class Form data are sent to the action class when the form is submitted.
Action class Page / template The page/template is shown as one of the results of the action class.
Action class Binary file A binary file is shown as one of the results of the action class.
An action class is executed as result of another. This process is known as “action chain-
Action class Action class
ing”.

dependency association. The same is true for as- the action class returns this keyword as re-
sociations that represent results. Naturally, these sult of the action execution, the framework
methods should be modeled in the diagram. will follow this flow and show the appro-
When modeling action chaining, it’s sometimes priate result page/template/binary file;
necessary to indicate the method that was executed • {resultType=type name} determines the
in the first action and the one that will be executed type of result, among those supported by
in the following. These can be specified with the the framework. Usually, at least the fol-
constraints outMethod and inMethod. lowing types of result are available: binary
For dependency associations that represent (display a binary file), chain (action chain-
results there are two other constraints that can ing), dispatch (dispatches the request), re-
be used: direct (redirects the request) and template
(processes a template using a template
• {result=result name} specifies a keyword engine).
that represents this control flow, i.e., when

Figure 14. Navigation Model for the use case “Autheticate User”

330
Designing Web Information Systems for a Framework-Based Construction

The difference between a dispatch and a redi- forms; and (b) it models composition between
rection is that the first makes the action’s output pages and forms with a more appropriate nota-
parameters available to the view, while the second tion. Nonetheless, designers are advised to build
does not. When a dependency association doesn’t sequence diagrams to represent complex flows
specify a type, it means it is a dispatch. The default when they see fit.
result is defined by the framework. Last but not least, FrameWeb suggests four steps
The designer is free to choose the granularity for the construction of a navigation model:
of the action classes, building one for each use
case scenario, one for each use case (encompass- 1. Study the use cases modeled during require-
ing many scenarios), one for multiple use cases, ments analysis to define the granularity of
and so forth. Moreover, he/she should decide if the action classes (using, preferably, names
it’s best to represent many actions in a single dia- that can relate the actions to the use cases/
gram or have a separate diagram for each action. scenarios they refer to);
Figure 14 is the navigation model for a use case 2. Identify how the data gets to the action
of LabES Portal. class, modeling input pages and forms and
The figure shows that in the initial page of the appropriate attributes on them and in the
the portal (represented by web::index), there is action class;
a form where login and password can be filled. 3. Identify what are the possible results and
When submitted, this data goes to the action class model the output pages/templates/binary
for the execution of the executeLogin() method, files, also adding attributes when appropri-
which would access the business logic layer to ate. We suggest that results that come from
perform the use case. If the information filled exceptions should not be modeled to avoid
is correct (result = success), the user is taken to polluting the diagram;
web::home, which represents the starting page for 4. Periodically check if the model is getting
authenticated users. Otherwise, the user will be too complex and consider dividing it into
taken back to web::index (result = input), showing two or more navigation models.
once again the login form and an error message.
If the user forgot his/her password, he/she Application Model
can click on a link in the initial page to go to the
web::remindpassword page, where his/her login The application model is a UML class diagram
would be informed and sent to the action class. that represents classes from the Application pack-
The executeRemindPassword() method requests age and their relationship with the Controller and
the business logic layer to send the password to Persistence packages. Besides guiding the imple-
the user’s email address and informs the user that mentation of application classes, this diagram
the message has been sent. To log out, the user also instructs developers on the configuration of
clicks on the appropriate link and is redirected the Dependency Injection framework, which is
back to the initial page. responsible for managing the dependencies among
During the conception of FrameWeb, there these three packages.
has been a discussion on whether the navigation The granularity of the application classes can
model would be better represented by a sequence be chosen by the developer in the same way as the
diagram, as it could represent better the control granularity of the action classes. The application
flow. Two main reasons led to the choice of the model also shares similarities with the persistence
class diagram: (a) it provides a better visualization model, as it does not define any UML extension
of the inner elements of action classes, pages and and uses the “programming to interfaces” prin-

331
Designing Web Information Systems for a Framework-Based Construction

Figure 15. Part of an Application Model of the User Control module of LabES Portal

ciple, indicating the modeling of an interface for FrameWeb suggests four steps for the construc-
each application class. tion of an application model:
When an application class is modeled, all ac-
tion classes that depend on it should be displayed 1. Study the use cases modeled during analysis
in the diagram, with the appropriate namespaces to define the granularity of the application
and relationships depicted. Analogously, all DAOs classes (using, preferably, names that can
required by the application class to execute the relate the classes to the use cases/scenarios
use case should have their interfaces shown in the they implement);
model, along with the relationship with the ap- 2. Add to the interfaces/classes the methods
plication class. Both relationships are represented that implement the business logic, giving
by directed associations and the multiplicity is not special attention to the name of the method
required, as it is always 1. (as before, with the name of the class), its
Figure 15 shows part of an application model parameter, the parameters types and its return
of LabES Portal, depicting the classes that imple- type;
ments the “Manage User” and “Authenticate User” 3. By reading the use case descriptions, identify
use cases and its relationships with controller which DAOs are necessary for each applica-
and persistence components. The methods of the tion class and model the associations;
classes represent each scenario of each use case 4. Go back to the navigation model (if already
and define the parameters that should be given built) and identify which action classes de-
for them. pend on which application class and model
Application classes manipulate domain objects their associations.
and, thus, depend on them. These relationships,
however, are not shown in the diagram to avoid By defining the standard architecture and a
increasing the complexity of the model. One can UML profile for the construction of these four
know about these relationships by reading the diagrams, FrameWeb provides the appropriate
description of each use case. tools for the design of framework-based WISs.

332
Designing Web Information Systems for a Framework-Based Construction

To promote the construction of “Semantic Web- Domain Analysis


enabled” WISs, an extension called S-FrameWeb
was proposed and it is presented in the next sec- The first step for bringing a WIS to the Seman-
tion. tic Web is formally describing its domain. As
discussed previously in this chapter, this can be
achieved by the construction of an ontology. S-
S-FrameWeb FrameWeb indicates the inclusion of a Domain
Analysis activity in the software process for the
The main goal of S-FrameWeb is to make WISs development of a domain ontology (we don’t use
“Semantic Web-enabled”. Being a framework- the term “domain model” to avoid confusion with
centered method, the chosen approach is to have FrameWeb’s Domain Model – FDM –, which is
the Front Controller framework produce dynamic a design model).
annotations by identifying if requests come from Domain Analysis is “the activity of identifying
human or software agents. In the former case, the the objects and operations of a class of similar sys-
usual Web page is presented, while in the latter, tems in a particular problem domain” (Neighbors,
an OWL document is returned. 1981; Falbo et al., 2002). When a software is built,
To accomplish this, S-FrameWeb extends the purpose is to solve a problem from a given
FrameWeb in the following manners: domain of expertise, such as medicine, sales or car
manufacturing. If the domain is analyzed prior to
• The activity of Domain Analysis should the analysis of the problem, the knowledge that is
be conducted in the beginning of the proj- formalized about the domain can be reused when
ect to build an ontology for the domain in another problem from the same domain needs a
which the software is based. If it already software solution (Falbo et al., 2002).
exists, it should be reused (and eventually S-FrameWeb does not impose any specific
modified); method for the construction of ontologies. It also
• Requirement Specification and Analysis go doesn’t require a specific representation language,
as usual, except for the fact that conceptual but suggests the use of OMG’s20 Ontology Defi-
models build during Analysis can now be nition Metamodel (ODM) (OMG, 2007), “a lan-
based on the domain ontology built in the guage for modeling Semantic Web ontologies in
previous activity; the context of MDA” (Đurić, 2004). ODM defines
• During design, FrameWeb’s Domain an ontology UML profile that allows developers
Model (FDM) receives semantic annota- to represent ontologies in UML class diagrams.
tions based on the domain ontology; In the development of the LabES Portal, the
• During implementation, the MVC frame- SABiO method (Falbo, 2004) was followed,
work has to be extended in order to per- resulting in the construction of an ontology for
form dynamic annotation. educational portals that deals with competency
questions such as: what are the roles of the people
Figure 16 shows the software process suggested in the educational institution?, what are the areas
by S-FrameWeb while Table 4 summarizes the of interest of these people and the institution?,
evolution of the models throughout that software how is the institution organized?, etc. The ontol-
process. ogy was divided into two separate diagrams: one
The following subsections go through the for the general structure of educational protals
suggested software process discussing it in more and another specific for publications. Figure 17
detail. shows the first one.

333
Designing Web Information Systems for a Framework-Based Construction

Figure 16. The software process suggested by S-FrameWeb (SOUZA et al., 2007)

The domain ontology serves as a basis for Analysis, it suggests the use of ODM for the
the construction of the application’s conceptual graphical representation of the conceptual model,
model (during Requirement Analysis), which as it eases its conversion to FDM and, later on, to
should derive some classes and associations from code (using OWL).
the ontology, adding and modifying elements as Figure 18 shows the conceptual model for the
needed, concerning the specific problem being User Control module of the LabES Portal.
solved. The stereotype <<OntClass>> indicates do-
main classes, <<ObjectProperty>> models asso-
Requirement Specification ciations between domain classes, <<DataType>>
and Analysis represents XML data types and <<DatatypeProp-
erty>> models associations between classes and
The activities of Requirement Specification and data types.
Analysis should be conducted by the develop- The reader accustomed with UML conceptual
ment team using its methodology of preference. models may notice that associations are repre-
S-FrameWeb, like FrameWeb, does not prescribe sented as classes in ODM. This is because in
any methods or languages to this phase of the OWL associations are independent from classes
software process. However, as during Domain and, for instance, can form their own subsumption

Table 4. Models produced by the software process suggested by S-FrameWeb (SOUZA et al., 2007)

Activity Artifact What the model represents


Domain Analysis Domain Ontology Concepts from the domain to which the software is being built. Mod-
eled in ODM, but converted to OWL for deployment.
Requirement Analysis Conceptual Model Concepts that are specific to the problem being solved. Modeled in
ODM.
System Design FrameWeb’s Domain Model Same as above plus OR mappings. Modeled using S-FrameWeb’s
(FDM) UML profile.
Coding OWL code OWL representation of FDM, without OR mappings.

334
Designing Web Information Systems for a Framework-Based Construction

Figure 17. Diagram of the structural part of the ontology for educational portals

Figure 18. The conceptual model for the User Control module of LabES Portal, in ODM

hierarchy. This could also happen with attributes, S-FrameWeb suggests the conceptual model is
for the same reasons. More on ODM’s semantics simplified, such as the one shown in Figure 19.
can be found at (OMG, 2007). Notice that this diagram is very similar to the one
In the cases where there is no need to repre- in Figure 8.
sent associations or attributes as UML classes,

335
Designing Web Information Systems for a Framework-Based Construction

Design in previous activities. Figure 20 shows the FDM


for the LabES Portal. We can see that, based on
As discussed before, FrameWeb proposes the the simplified version of the conceptual model,
creation of four kinds of models during design: association navigabilities were defined, data types
domain, persistence, navigation and application were chosen among those of the implementation
models. These models are still used with S- platform and that some O/R mappings were in-
FrameWeb, although the domain model (FDM) cluded. The result is very similar of that of Figure
should be adapted to a representation more suitable 11, due to the simplifications performed.
to the purposes of this semantic extension. There- The representation of this model in a language
fore, S-FrameWeb suggests a new UML profile for that mixes profiles from both ODM and FrameWeb
this diagram, mixing the profile defined by ODM attempts to facilitate the implementation phase,
with the one proposed by FrameWeb. when an OWL file representing the conceptual
This new modeling language consists basically model should be created and the ORM framework
of the one defined by ODM, with the following should be configured.
adaptations:
Implementation, Testing
1. Specification of association navigabilities and Deployment
for the implementation of the classes;
2. Addition of the O/R mappings for the con- During implementation, the classes that, integrated
figuration of the ORM framework; with the frameworks, provide a software solution
3. Use of the data types of the implementation to the problem at hand are developed. S-FrameWeb
platform instead of those defined by the XML adds a new task to this activity: the construction
Schema Definition (XSD) standard21; of OWL files representing the domain ontology
4. Simplification of ODM’s syntax when pos- and the application conceptual model (based on
sible (if not already done previously). the FDM). As stated before, this task is facilitated
by the use of ODM in both models.
Naturally, the construction of the FDM should The OWL files should be used by the Front
be based on the conceptual model already built Controller framework to implement dynamic

Figure 19. The conceptual model for the User Control module of LabES Portal, in its simplified ver-
sion

336
Designing Web Information Systems for a Framework-Based Construction

Figure 20. S-FrameWeb’s Domain Model for the User Control module of LabES Portal

annotation on the Web pages. S-FrameWeb pro- with “FrameWeb” in their names. Publications
poses an extension to this kind of framework that that are returned by the applications are placed
recognizes when a request comes from a human under the <results> tag, while objects associated
or from a software agent by analyzing a specific with them are placed under <instancesList> tag.
HTTP request parameter (e.g. owl=true). In the The association is made using the UUID of each
case of a software agent, the framework should object. See Figure 21.
respond with an OWL file that is based on the Since this result should be based on the ap-
domain ontology and the conceptual model, and plication ontology, it was necessary to use an
represents the data that would be shown in the ontology parser. For this purpose, we chose the
human-readable version of the page. Jena Ontology API, a framework that provides a
To experiment this approach in practice, a pro- programmatic environment for many ontology
totype of an extension for the Struts2 framework languages, including OWL. With Jena and Java’s
was built. Figure 21 shows this extension and reflection mechanisms, the OWL Result Class
how it integrates with the framework. The client’s reads all properties that are made available to the
web browser issues a request for an action to the Web page by the action, produces an OWL docu-
framework. Before the action gets executed, the ment containing their information and delivers it
controller automatically dispatches the request to the software agent.
through a stack of interceptors, following the Testing should be conducted in order to check
pipes and filters architectural style. not only the source code, but also the ontologies
An “OWL Interceptor” was developed and codified in OWL. In the context of S-FrameWeb,
placed as first of the stack. When the request is however, this is still open to research and study.
made, this interceptor verifies the HTTP parameter Deployment works as the same as other WISs, but
and, if present, creates a pre-result listener that will should also include the OWL files in a specific
deviate successful requests to the “OWL Result place in order to be used by the Front Control-
Class”, another custom-made component that is ler’s extension.
responsible for producing this result.
The listing below is an excerpt of an OWL
document produced by the search of publications

337
Designing Web Information Systems for a Framework-Based Construction

Figure 21.

Future Trends had to be extended. However, some developers


had difficulties on capturing the idea of some
Web Engineering is a relatively new field of re- frameworks, especially the MVC framework.
search. New methods, languages and frameworks All of them had some experience with the Java
are proposed to provide practitioners with tools platform, but most did not have any experience
that can facilitate and increase the productivity with Web development.
when developing WebApps. At the end of the development, the develop-
FrameWeb is a new tool, targeting WISs that ers were asked to provide feedback on the work
have their architecture based on frameworks. By done. This feedback can be summarized in the
suggesting a standard architecture and bringing following items:
concepts from the frameworks to the design
models, developers can translate models to code • Allowing to directly model aspects relat-
more easily and designer have more control on ed to the use of frameworks is the biggest
the outcome of the implementation. strength of FrameWeb;
FrameWeb was first applied in the develop- • Implementing in Java what was modeled
ment of the Portal of the Software Engineering during design was very much facilitated by
Lab – LabES. First, developers were trained in the clear understanding of the semantics
general concepts of Web Engineering, in the use of of the four models (domain, persistence,
FrameWeb and also in the following frameworks: navigation and application);
WebWork2, FreeMarker (template engine), Site- • The simplicity of the models facilitated
Mesh, Hibernate and Spring. the adoption of FrameWeb, except for the
In general, the development went smoothly. navigation model, which added some com-
The method allowed the developers to deliver plexity to the method.
the models mostly in time and few deadlines

338
Designing Web Information Systems for a Framework-Based Construction

Two other case studies were conducted. The there are several opportunities to improve the
local Java User Group ESJUG22 modeled a col- method. Future work may include:
laborative learning environment called JSchool23
using FrameWeb for the same set of frameworks • Further research on the impact of the use of
used in the LabES Portal project. This helped frameworks and FrameWeb on the activity
mature the method in its initial version. of Testing. The current work provides no
Another case study reimplemented the LabES discussion on the subject of testing;
Portal changing the Front Controller framework. • Proposals on layout and interaction mod-
This helped identify some extensions that should els. Complete methods for the design of
be added to FrameWeb in order to cope with WebApps should include models that mod-
some characteristics of different frameworks. el aesthetics and usability;
For instance, this work suggested the addition of • Conduction of more formal experiments
the <<formBean>> stereotype for the navigation with the method, evaluating more precise-
model to represent how the framework Struts ly the gains in the productivity of the de-
sends data from the web page to the action class. velopment team. Currently, only informal
It also reached the conclusion that the navigation experiments have been conducted and con-
model in FrameWeb is somewhat dependent on clusions have been reached by requesting
the instance of Front Controller frameworks used, developer’s opinions;
and not generic as it was assumed before. • Tools could be developed to help create the
More case studies should be conducted to models or to convert the models to code,
assess the effectiveness of the method and its automatically implementing much of the
appropriateness to different instances of frame- infrastructure code and configuration for
works. Many improvements can come from more the most used frameworks available;
practical experiences. • To make FrameWeb’s models more ge-
The use of framework-based architectures neric, the development of an ontology
is becoming the standard for implementation on Web Applications and frameworks to
of medium-to-large-sized WIS. Taking the Java guide the evolution of FrameWeb’s mod-
platform as example, the definition of standards as eling language. New concepts brought by
JavaServer Faces (JSF)24 for Web development and new frameworks could be included in the
the new Enterprise JavaBeans (version 3.0)25 for ontology and, thus, taken to the modeling
distributed components reinforce that conclusion. language;
JSF defines a MVC-like architecture, and EJB 3.0 • Continuation on the research on the
had all of its persistence model reconstructed based Semantic Web and in-practice experiments
on Hibernate ORM framework and also makes on the construction of a Semantic WIS us-
heavy use of Dependency Injection. ing S-FrameWeb;
The research on the Semantic Web points out • Deeper discussions on how to tackle spe-
to the future of the World Wide Web. Methods for cific Semantic Web issues such as: how
the development of WISs should prepare for, or will agents find the desired web page?,
even help build, this new paradigm. S-FrameWeb how will they know how to interact with
suggests a software process that facilitates the it?, how will they know if a concept “table”
development of Semantic WISs by automating refers to a piece of furniture or a system-
certain tasks concerning the generation of semantic atic arrangement of data usually in rows
annotations on dynamic Web pages. Nonetheless, and columns?, will a top-level ontology be
FrameWeb and S-FrameWeb are far from ideal: used for all the Internet?

339
Designing Web Information Systems for a Framework-Based Construction

• Evaluation on how to use Semantic Web S-FrameWeb complements FrameWeb, add-


Services with S-FrameWeb instead of the ing activities that promote the construction of
dynamic page approach and a comparison Semantic WISs. Given that the Semantic Web
of both solutions. vision will not come true unless Web authors
add semantic to their websites, S-FrameWeb is a
step in that direction, giving directives for WISs
Conclusion developers to follow in order to add Semantic to
Web Applications.
The amount of propositions in the Web Engineer-
ing area, including methods, frameworks and mod-
eling languages, is quite vast, demonstrating that References
academics and practitioners haven’t yet elected a
standard when it comes to Web development. W3C (2007a). OWL Web Ontology Language
Parallel to this, many frameworks and contain- Guide, fev. 2004.
ers for the implementation of WISs were created, W3C (2007b). W3C Glossary and Dictionary.
denoting the need for a basic infra-structure
that helps on the quick development of reliable Ahmad, R., Li, Z., & Azam, F. (2005). Web
software with low future maintenance costs. Engineering: A New Emerging Discipline. In
With several ready-to-use and extensively tested Proceedings of the IEEE Symposium on Emerg-
components, frameworks promote reuse and good ing Technologies (pp.445-450). Catania, Italy:
programming practices. IEEE.
The large utilization of these frameworks and
Alur, D., Malks, D., & Crupi, J. (2003). Core J2EE
containers by practitioners and the absence of a
Patterns: Best Practices and Design Strategies,
design method directed to them has motivated
2nd edition. Prentice Hall.
the proposal of FrameWeb, a method based on
frameworks for the design of WISs. The current Ambler, S., & Jeffries, R. (2002). Agile Modeling:
research on the Semantic Web, with many efforts Effective Practices for Extreme Programming
on bringing this idea to reality has impelled us and the Unified Process, 1st edition. John Wiley
to extend this method and create S-FrameWeb: a & Sons.
method based in frameworks to the construction
Arch-Int, S., & Batanov, D. N. (2003). Develop-
of semantic WISs.
ment of industrial information systems on the Web
Given all of the options available, FrameWeb
using business components. [Elsevier.]. Comput-
comes in as another one that targets a specific
ers in Industry, 50(2), 231–250. doi:10.1016/
architecture, one based on the use of frameworks.
S0166-3615(02)00122-7
In this case, FrameWeb excels for its agility, be-
cause models are directed towards the framework Bauer, C., & King, G. (2004). Hibernate in Action,
architectures and allow for quick understanding 1st edition. Manning.
of the implementation. It also doesn’t introduce
Berners-Lee, T., Hendler, J., & Lassila, O. (2001).
much complexity, allowing organizations to use
The Semantic Web. Scientific American, 284,
their own processes up to design with few adapta-
34–43.
tions, if any. Of all the proposed design models,
the navigation model is the only one we consider Booch, G., Rumbaugh, J., & Jacobson, I. (2005).
a little bit complex, making FrameWeb very easy The Unified Modeling Language User Guide. 2nd
to learn and use. edition. Addison-Wesley Professional.

340
Designing Web Information Systems for a Framework-Based Construction

Ceri, S., Fraternali, P., & Bongio, A. (2000). Ginige, A., & Murugesan, S. (2001). Web Engi-
Web Modeling Language (WebML): a model- neering: An Introduction. [IEEE.]. IEEE Multi-
ing language for designing Web sites. Computer Media, 8(1), 14–18. doi:10.1109/93.923949
Networks, 33 (1-6), 137-157, Elsevier.
Gomez-Perez, A., Corcho, O., & Fernandez-
Conallen, J. (2002). Building Web Applications Lopez, M. (2005). Ontological Engineering.
with UML, 2nd edition. Addison-Wesley. Springer.
Díaz, P., Montero, S., & Aedo, I. (2005). Mod- Guarino, N. (1998). Formal Ontology and In-
elling hypermedia and web applications: the formation Systems. In Proceedings of the 1st
Ariadne Development Method. [Elsevier.]. Infor- International Conference on Formal Ontologies
mation Systems, 30(8), 649–673. doi:10.1016/j. in Information Systems (pp. 3-15), Trento, Italy:
is.2004.09.001 IOS Press.
Đurić, D. (2004). MDA-based Ontology Infra- Jacobson, I., Booch, G., & Rumbaugh, J. (1999).
structure. Computer Science and Information The Unified Software Development Process, Ad-
Systems 1 (1), ComSIS Consortium. dison Wesley.
Falbo, R. A. (2004) Experiences in Using a Method Jacyntho, M. D., Schwabe, D., & Rossi, G. (2002).
for Building Domain Ontologies. In Proceedings A Software Architecture for Structuring Complex
of the Sixteenth International Conference on Soft- Web Applications. In Proceedings of the 11th
ware Engineering and Knowledge Engineering International World Wide Web Conference, Web
(pp. 474-477), Banff, Alberta, Canadá. Engineering Alternate Track, Honolulu, EUA:
ACM Press.
Falbo, R. A., Guizzardi, G., & Duarte, K. C.
(2002). An Ontological Approach to Domain Koch, N., Baumeister, H., Hennicker, R., &
Engineering. In Proceedings of the 14th Interna- Mandel, L. (2000). Extending UML to Model
tional Conference on Software Engineering and Navigation and Presentation in Web Applications.
Knowledge Engineering (pp. 351- 358). Ischia, In Proceedings of Modelling Web Applications in
Italy: Springer. the UML Workshop, York, UK.
Fons, J., Valderas, P., Ruiz, M., Rojas, G., & Koch, N., & Kraus, A. (2002). The Expressive
Pastor, O. (2003). OOWS: A Method to Develop Power of UML-based Web Engineering. In D.
Web Applications from Web-Oriented Concep- Schwabe, O. Pastor, G. Rossi e L. Olsina (Ed.),
tual Models. In Proceedings of the International Proceedings of the Second International Work-
Workshop on Web Oriented Software Technology shop on Web-Oriented Software Technology (pp.
(pp. 65-70), Oviedo, Spain. 105-119), CYTED.
Fowler, M. (2002). Patterns of Enterprise Ap- Krutchen, P. (2000). The Rational Unified Process:
plication Architecture. Addison-Wesley. An Introduction, 2nd edition. Addison-Wesley.
Fowler, M. (2007). Inversion of Control Contain- Lee, S. C., & Shirani, A. I. (2004). A component
ers and the Dependency Injection pattern. based methodology for Web application develop-
ment. Journal of Systems and Software, 71 (1-2),
Gamma, E., Helm, R., Johnson, R., & Vlissides,
177-187, Elsevier. Lima, F., & Schwabe, D.
J. (1994). Design Patterns: Elements of Reusable
(2003). Application Modeling for the Semantic
Object-Oriented Software. Addison-Wesley.
Web. In Proceedings of the 1st Latin American
Web Conference (pp. 93-102), Santiago, Chile:
IEEE-CS Press.

341
Designing Web Information Systems for a Framework-Based Construction

McIlraith, S. A., Son, T. C., & Zeng, H. (2001). Schwabe, D., & Rossi, G. (1998). An Object
Semantic Web Services. [IEEE.]. Intelligent Sys- Oriented Approach to Web-Based Application
tems, 16(2), 46–53. doi:10.1109/5254.920599 Design. Theory and Practice of Object Systems
4 (4), Wiley and Sons.
Murugesan, S., Deshpande, Y., Hansen, S., &
Ginige, A. (1999). Web Engineering: A New Dis- Shannon, B. (2003). JavaTM 2 Platform Enterprise
cipline for Development of Web-based Systems. Edition Specification, v1.4. Sun Microsystems.
In Proceedings of the 1st ICSE Workshop on Web
Souza, V. E. S., & Falbo, R. A. (2007). FrameWeb
Engineering (pp. 3-13). Australia: Springer.
- A Framework-based Design Method for Web
Narayanan, S., & McIlraith, S. A. (2002). Simula- Engineering. In [Faro, Portugal: ACM Press.].
tion, Verification and Automated Composition of Proceedings of the Euro American Conference
Web Services. In Proceedings of the 11th interna- on Telematics and Information Systems, 2007,
tional conference on World Wide Web (pp. 77-88), 17–24.
Hawaii, USA: ACM.
Souza, V. E. S., Lourenço, T. W., Falbo, R.
Neighbors, J. M. (1981). Software Construction A., & Guizzardi, G. (2007). S-FrameWeb – a
Using Components. Ph.D. Thesis. Department of Framework-based Design Method for Web Engi-
Information and Computer Science, University of neering with Semantic Web Support. In Proceed-
California, Irvine. ings of Workshops and Doctoral Consortium of
the 19th International Conference on Advanced
OMG. (2007). Ontology Definition Metamodel
Information Systems Engineering (pp. 767-778),
Specification.
Trondheim, Norway.
Pastor, O., Gómez, J., Insfrán, E., & Pelechano, V.
Standing, C. (2002). Methodologies for devel-
(2001). The OO-Method Approach for Informa-
oping Web applications. [Elsevier.]. Informa-
tion Systems Modelling: From Objetct-Oriented
tion and Software Technology, 44(3), 151–159.
Conceptual Modelling to Automated Program-
doi:10.1016/S0950-5849(02)00002-2
ming. [Elsevier.]. Information Systems, 26(7),
507–534. doi:10.1016/S0306-4379(01)00035-7
Pressman, R. S. (2005). Software Engineering:
Endnotes
A Practitioner’s Approach, 6th edition. McGraw
Hill. 1
http://struts.apache.org/2.x/index.html
2
Reenskaug, T. (1979). THING-MODEL-VIEW- http://www.springframework.org
3
EDITOR, an Example from a planning system. http://tapestry.apache.org
4
Xerox PARC Technical Note. http://www.opensymphony.com/sitemesh
5
http://struts.apache.org/struts-tiles
Resende, A., & Silva, C. (2005). Programação 6
http://www.hibernate.org
Orientada a Aspectos em Java. Brasport. 7
http://java.sun.com/products/jdo
8
Rosenberg, D., & Scott, K. (1999). Use Case http://db.apache.org/ojb/
9
Driven Object Modeling with UML: A Practical http://www.oracle.com/technology/prod-
Approach. Addison-Wesley. ucts/ias/toplink
10
http://www.picocontainer.org
Schmidt, D. (2007). “Programming Principles in 11
http://jakarta.apache.org/hivemind
Java: Architectures and Interfaces”, chapter 9. 12
http://www.eclipse.org/aspectj

342
Designing Web Information Systems for a Framework-Based Construction

13
http://labs.jboss.com/portal/jbossaop Universal Unique Identifier (UUID) was
14
http://www.acegisecurity.org taken from this article.
15 20
http://cocoon.apache.org Object Management Group – http://www.
16
http://java.sun.com/products/jaas omg.org/ontology/
17 21
http://oiled.man.ac.uk/ The XML Schema standard can be found at
18
http://protege.stanford.edu/ http://www.w3.org/XML/Schema. Its data
19
The relationship between an object’s identity types are described in a specific page, at
in memory and its primary key in the data- http://www.w3.org/TR/xmlschema-2.
22
base raises several issues that are discussed http://esjug.dev.java.net
23
in the article “Hibernate, null unsaved value http://jschool.dev.java.net
24
and hashcode: A story of pain and suffering” http://jcp.org/en/jsr/detail?id=127
25
from Jason Carreira (http://www.jroller.com/ http://jcp.org/en/jsr/detail?id=220
page/jcarreira?entry=hibernate_null_un-
saved_value_and). The idea of using a

This work was previously published in Innovations in Information Systems Modeling: Methods and Best Practices, edited by
T. Halpin, J. Krogstie, and E. Proper, pp. 204-238, copyright 2009 by Medical Information Science Reference (an imprint of
IGI Global).

343
344

Chapter 2.5
Focused Requirements
Engineering Method for Web
Application Development
Ala M. Abu-Samaha
Amman University, Jordan

Lana S. Al-Salem
SpecTec Ltd & MEP, Greece

Introduction The Web engineering (WE) literature provides


a limited number of methods and techniques
The requirements phase of the system/application that can be used to manage the RE process in a
development process typically involves the activities Web development context [e3-value framework
of requirements elicitation, analysis, validation, and (Gordijn, Akkermans, & van Vliet, 2000), SOARE
specification. The main goal of such a process is approach (Bleistein, Aurum, Cox, & Ray, 2004),
“to develop a requirements specification document e-prototyping (Bleek, Jeenicke, & Klischewski,
which defines the system to be procured and which 2002), AWARE (Bolchini & Paolini, 2004), and
can act as a basis for the system design” (Sawyer, SSM/ICDT (Meldrum & Rose, 2004)]. Despite
Sommerville, & Viller, 1996). Hence the underpin- the availability of such a limited number of Web
ning assumption of the requirements engineering requirements engineering (WRE) methods, many
(RE) process is to transform the operational needs researchers criticised such methods for their failure
of an organisation into complete, consistent, and to address the necessity to align the Web applica-
unambiguous system/application specifications tion’ requirements to the organisation’s business
through an iterative process of definition and vali- strategy. Hence, the recommendation of many
dation (Pohl, 1994). researchers (Al-Salem & Abu-Samaha, 2005a;
Bleistein 2005; Bleistein, Cox, & Verner, 2004;
DOI: 10.4018/978-1-60566-026-4.ch244 Vidgen, Avison, Wood, & Wood-Harper, 2002) is

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Focused Requirements Engineering Method for Web Application Development

to utilise a general WRE framework for the de- 2001). Hence, Web applications can be defined as
velopment of Web applications that can align the “applications that tend to be used to integrate and
application’s requirements to the organisation’s streamline an organisation’s business processes
business needs and its future vision. The objective beyond organisational (customers, agents, sup-
of such a WRE framework is to incorporate the pliers, others) and geographical borders; provide
elicitation/analysis of business strategy as part of an organisation with competitive products and
the application’s RE process. services that give it a strategic advantage over
This chapter presents a WRE method that its competitors in the marketplace; promote
extends Sommerville and Kotonya’s viewpoint- business innovation; and/or improve operational
oriented requirements definition (VORD) and efficiency” (Al-Salem & Abu-Samaha, 2005a).
Kaplan and Norton’s balanced scorecard (BSC) There is a pressing need in the WE discipline
to elicit the Web application’ requirements and to for RE approaches and techniques that (a) take
plan/analyze the business strategy, respectively. into account the multiplicity of user profiles and
In addition, eWARE (extended Web application the various stakeholders involved [a stakeholder
requirements engineering) deploys the concept of is defined as “anyone who can share information
“requirements alignment” to attain business objec- about the system, its implementation constraints
tives during the requirements discovery, elicitation, or the problem domain” (Potts, Takahashi, &
and formalisation process to identify the services of Anton, 1994)]; (b) eliciting overall functionality
the Web application that will achieve the business and the business environment of the Web appli-
objectives in order to improve the organisation’s cation; (c) specifying technical and nontechnical
profitability and competitiveness. The chapter is requirements of the Web application, and (d)
organised into a number of sections. The second aligning the Web application’ requirements to the
section of this chapter provides a background overall business strategy (Bleistein et al., 2005;
to Web applications in terms of definition and Ceri, Fraternali, Bongio, Brambilla, Comai, &
differentiating characteristics. The third section Matera, 2003; Ginige & Murugesan, 2001; Kautz
provides a discussion of eWARE method in terms & Madsen, 2003; Lowe, 2003; Meldrum & Rose,
of phases and activities. This section is divided 2004; Nuseibeh & Easterbrook, 2000; Vidgen et
into two subsections to cover the activities of the al., 2002). More importantly, a Web application
two prominent phases of the eWARE process in must be developed with an emphasis on how
more detail. The fourth and fifth sections provide the services of such an application can achieve
a discussion of possible future trends in WRE and the business vision and strategy and fulfil the
a number of concluding remarks. business processes (Haire, Henderson-Sellers,
& Lowe, 2001).

Background eWARE Process

Web applications provide organisations an unprec- eWARE process can be best perceived as a series
edented chance to stretch their existence beyond of activities grouped into three phases; strategy
the typical boundaries of an organisation to include articulation via BSC, Web application’ require-
customers, trading partners, and suppliers.Little ments elicitation via VORD, and prototype build-
attention has been paid to the process of RE for ing; Figure (1) presents the phases and activities
Web application development, in comparison to of eWARE. Such a process aims to develop a Web
other areas of the development process [model- application requirements specification (WRS)
ling, design, and coding] (Ginige & Murugesan, document that is aligned with business strategy

345
Focused Requirements Engineering Method for Web Application Development

Figure 1. The eWARE process

and detailed enough to be used for contractual viewpoints (VPs), their sub-VPs, and requirements
purposes. for each viewpoint (VP). Kotonya and Sommeriv-
The strategy articulation phase of eWARE can lle (1996) define a VP as anyone who may have
be best thought of as a structure of many layers. some direct or indirect influence on the system/
The vision of the organisation is at the top of application requirements. Goals and objectives
the structure, while the strategic objectives are of the different stakeholder groups need to be
presented in the next layer of the structure fol- identified to define success or failure measures
lowed by the Web application services. The next for each stakeholder. Moreover, nonfunctional
layer of the structure contains the measurements requirements {NFR) need to be identified; these
and targets for meeting the strategic objectives. include some of the “-ilities” of the Web applica-
The level of detail tends to increase as we move tion, such as reliability, supportability, maintain-
down the structure of the strategy. This articula- ability, affordability, and so forth. Finally, in the
tion of the organisation’s vision, objectives, and prototyping phase of the eWARE process, the sys-
measurements aims to translate the future vision tem/application stakeholders consider the unclear
of the organisation into detailed and prioritised set of requirements, and agree on prototyping the
Web application requirements (this will be fully ambiguous requirements to verify them. Moreover,
covered in the coming subsections). the user interface (UI) and Web site structure are
The requirements elicitation phase of the presented in a throw away prototype.
eWARE process is used to produce a WRS docu- The mentioned WRE phases and activities are
ment based on requirements collected during the perceived to be iterative and incremental in nature
strategy articulation phase. Requirements elici- where unmet targets are questioned. This cyclic
tation relies on the identification of the relevant view of the WRE process will trigger the strategy

346
Focused Requirements Engineering Method for Web Application Development

Figure 2. eBSC perspectives

articulation phase to enter in a feedback loop in Strategic Objectives Perspective


order to refine services and change requirements
of the Web application in order to enhance the Objectives are statements that clarify what the
organisation’s chance to achieve its set vision strategy aims to achieve. Hence, organisations
and strategy. pursue strategies in the belief that, when imple-
mented, they will enable the organisation to better
Strategy Articulation Phase achieve its strategic objectives.

As mentioned earlier, the strategy articulation Web Application Services Perspective


phase of eWARE process aims to align the Web
application’ requirements to the organisation’s The Web application services are the collection
business strategy through a process of strategy of functionality, quality, content, and all what the
analysis. eWARE delivers such alignment via Web application must provide in order to achieve
eBSC (extended Balanced Score Card). Ac- the strategic objectives of the organisation and
cording to eBSC, aligning the Web application’ to meet the expectations of its stakeholders. This
requirements to the business objectives yields perspective is considered as a precondition to
four perspectives to focus upon (stakeholders, process improvement, stakeholder satisfaction,
strategic objectives, internal processes, and Web and business objectives’ realisation. Hence, the
application services). Figure 2 provides a diagram- development team, comprising of requirement
matic representation of the perceived relationship engineers and end user representatives, must be
between the four perspectives. empowered to select which services (function-
alities) of the Web application will be included,

347
Focused Requirements Engineering Method for Web Application Development

in order to deliver the optimal business solution. acquisition, customer retention, and customer
Decisions and trade-offs need to be made between profitability. The goal of using eBSC is to classify
new services elicited through the eBSC process, the organisation’s strategy by stakeholder, which
and less important original functionalities that will lead to introducing Web application services
may have to be excluded, being less strategically for each group to meet their strategy or objective.
important. The organisation must determine whom it serves
and how their requirements can be met.
Stakeholders1 Perspective
Internal Process Perspective
The stakeholder perspective is arguably the most
important perspective in the Web application Organisations’ activities can be grouped into
development process. As mentioned earlier, a “business processes” that describe the way work
stakeholder is defined by Potts et al. (1994) as is to be implemented. Some of the organisation’s
“anyone who can share information about the sys- activities will be affected by the introduction of
tem, its implementation constraints or the problem a Web application, since such applications have
domain.” According to Potts et al. (1994), the list of the potential to significantly change an organisa-
stakeholders will include end users, indirect users, tion’s work practices and procedures (Ginige &
other customer representatives, and developers. Murugesan, 2001; Pressman, 2004). The internal
Since a stakeholder is, in essence, a requirements process perspective focuses on key processes at
source; it can be an application user, a competitor, which the organisation must excel in order to
or even a third party. Hence, if the stakeholder add value to its stakeholders through the Web
represents a customer, then the requirements application.
elicitation effort will focus on new customer

Figure 3. Web application cause and effect relationships

348
Focused Requirements Engineering Method for Web Application Development

When designing a score card, the starting point views of the Web application. These perspec-
will be asking “what strategies do we have to put tives are partial or incomplete descriptions of the
in place to satisfy the wants and needs of the key system/application, and reflect the environment
stakeholders?” The inclusion of BSC within the in which the system/application will operate in.
WRE process aims to help clarify, consolidate, The integration of multiple views can contribute
and gain consensus around the business–Web to augment the overall understanding of the Web
application strategy of the organisation, and to application. The authors recommend the use of
translate the business strategy into Web applica- a new and amended list of VPs for developing
tion services (requirements) to ensure that the Web applications. These WebVPs’ abstracts
elicited requirements are strategy focused. Hence, have been identified as a starting point that acts
a strategy can be best described as a series of cause as a template for WebVPs classes and hierarchy.
and effect relationships that provide a translation Figure (4) shows a high-level abstract of WebVPs
from future vision to Web requirements, as shown structure.
in Figure (3). WebVPs can be classified into direct VPs that
interact directly with the Web application and fall
eVORD for Eliciting Web into two subclasses (receivers and providers of
Requirements services), and indirect VPs that have “interest”
in some or all of the services that are delivered
The second phase of the proposed WRE process by the system but do not interact directly with
presents the requirements elicitation phase of the it; hence, they provide high-level organisational
Web application development process. eWARE requirements and constraints. Indirect VPs fall
delivers such elicitation via eVORD (extended into a number of subclasses: environmental
viewpoint-oriented requirements definition). Ac- WebVPs, which reflect the requirements of the
cording to eVORD, templates are created to de- business domain, that is, legalisation, localisation,
scribe each viewpoint (VP), service, nonfunctional taxation, and competitors; engineering WebVPs,
requirement (NFR), and content. As mentioned which reflect the requirements of the development
earlier, Kotonya and Sommerville (1996) define a team, that is, software engineers, team leaders, and
VP as anyone who may have some direct or indirect creative designers; and system WebVPs, which
influence on the system/application requirements. include all existing information systems that the
The VORD requirements engineering process in- application being analysed needs to interface to,
cludes activities concerned with VP identification, that is, payment systems and supplier systems.
VP service description, cross-viewpoint analysis
to discover inconsistencies, omissions, and con- Eliciting Requirements Details
flicts, and developing an object-oriented model
of the system/application from the VP analysis. The second step of this phase is concerned with
In addition, a VP diagram is used to show the re- documenting the details of each WebVP (identified
lationships among VPs, while sequence diagrams in the previous step) and its associated services.
illustrate the interactions among VPs. For every identified WebVP, a number of templates
are used to elicit and specify its requirements.
Web Application Viewpoints (WebVPs) The authors have extended and adapted VORD
Identification and Structuring templates to cater to the particularities of Web
applications development (please refer to Al-
The construction of Web applications involves a Salem & Abu-Samaha, 2005a and Al-Salem &
great number of stakeholders who have different Abu-Samaha, 2005b for more details).

349
Focused Requirements Engineering Method for Web Application Development

Figure 4. Abstract of WebVPs structure

Documenting Requirements to bring such artefacts to the changing contexts of


the development projects. The presented eWARE
The objective of the requirements process is to method is an extension of two existent methods
deliver a requirements specification document that used to align requirements of a Web application
defines the system/application to be developed to the organisation’s business strategy. eWARE
(Sawyer, Viller, & Sommerville, 1997). This needs to be tested further in different industrial
document is used for contractual purposes, and settings to validate the applicability of the method.
can be used as a basis for facilitating a competi- As well, eWARE, like many other RE methods,
tive tendering for the system/application design needs to be supported through the development of
and implementation (IEEE, 2004). The authors a computer aided software engineering (CASE)
have enhanced the typical software require- tool to facilitate the generation of the WRS docu-
ments specification (SRS) document to reflect ment. Compared to similar WRE methods (Bleek
the changes introduced to eVORD and eBSC, as et al., 2002; Bleistein et al., 2004; Bolchini &
depicted in Figure (5). Paolini, 2004; Gordijn et al., 2000; Meldrum &
Rose, 2004), eWARE provides its users with a
number of benefits: it extends the BSC approach
Future Work to help with the formulation of Web application
strategy as a stage in the WRE process in advance
The domain of RE, in general, and WRE, in par- to requirements elicitation; it provides a prior-
ticular, is evolving with the ever-changing contexts itisation framework synchronised with business
of software engineering and information systems strategy; it extends VORD to elicit requirements
development projects. Despite the availability of for Web applications; and it creates a WRS
many requirements, engineering methods, pro- document that specifies both the business strategy
cesses, techniques, and tools, such artefacts are and requirements for the Web application. Such
in need of constant extensions and enhancements advantages proved to be valuable in Web applica-

350
Focused Requirements Engineering Method for Web Application Development

Figure 5. Software requirements specification (SRS) document template

tion development projects, as Web applications Conclusions


are different in many aspects when compared to
other applications. The differing characteristics of There are two general opinions on whether current
Web applications can be summarized as diverse software engineering methods and techniques are
and volatile requirements, vast and unknown end applicable to face the challenges of developing
users, multiple stakeholders, adaptable architec- a Web application. One opinion advocates the
ture, short development life cycle, high visibility, need for a new “software engineering” discipline
heavy content, integration with backend databases that handles the Web application particularities
and third party applications, Web applications’ (Ginige & Murugesan, 2001, Murugesan, 1999;
relevance and direct effect on business, and mul- Murugesan, Deshpande, Hansen, & Ginige,
tidisciplinary development team. 1999). In contrast, the other opinion believes
that current software engineering methods, tools,
and techniques are applicable to Web applica-

351
Focused Requirements Engineering Method for Web Application Development

tion development. For example, McDonald and References


Welland (2001) recommended developing new
methods, techniques, and approaches to address Al-Salem, L. S., & Abu-Samaha, A. (2005a). As-
the challenges of developing Web applications in sessing the usability of VORD for Web applications
order to increase the possibility of their success. requirements engineering - An industrial case
This implies the need for a new breed of WRE study -. 1st International Workshop on Require-
methods, tools,and working practices. In contrast, ments Engineering for Business Need and IT
Pressman (2004) argues that Web applications Alignment (REBNITA 2005), Paris.
are a natural evolution of existing applications/ Al-Salem, L. S., & Abu-Samaha, A. (2005b). As-
systems offering a solution to classical problems sessing the usability of VORD method for Web ap-
exhibited by previous information systems (IS). plications requirements elicitation. International
The authors of this chapter hold a middle posi- Conference on Internet Technologies and Applica-
tion in between these two opinions. The authors tions (ITA 05), Wrexham, North Wales, UK.
perceive Web applications as a natural evolution
of “traditional” information systems, yet they Bleek, W.-G., Jeenicke, M., & Klischewski, R.
believe that such applications possess special (2002). Developing Web-based applications
characteristics that need to be provided for by the through e-prototyping. 26th Annual International
“traditional” RE method(s). Hence, existing RE Computer Software and Applications Conference
methods are considered valid for WRE, though (COMPSAC 2002), Oxford England.
they need to be enhanced and extended to cater to
Bleistein, S. J. (2005). Requirements analysis
the distinguishing features of Web applications.
framework for alignment of IT with competitive
The chapter has presented a WRE approach
strategy of business organizations. IEEE Interna-
that enables businesses to develop/procure Web
tional Conference on Requirements Engineering
application(s) capable to achieve the organisations’
- Doctoral Consortium, Paris.
business strategic objectives, and to effectively
harness their business processes. The combina- Bleistein, S. J., Aurum, A., Cox, K., & Ray, P.
tion of BSC (Kaplan & Norton 1993, 1996a & (2004). Strategy-oriented alignment in require-
b) and VORD (Kotonya & Sommerville 1996) ments engineering: Linking business strategy
within eWARE process provides the development to tequirements of e-business systems using the
team with the ability to translate strategy into SOARE approach. Journal of Research and Prac-
Web application requirements and to incorporate tice in Information Technology, 36, 259–276.
views from different complementary perspectives.
Bleistein, S. J., Cox, K., & Verner, J. (2005).
Hence, eWARE is a Web-specific RE process to
Validating strategic alignment of organizational IT
cope with the complex aspects of requirements
requirements using goal modelling and problem
elicitation, alignment, and specification for Web
diagrams. Journal of Software and Systems, 79,
applications development/procurement. eWARE
362–378. doi:10.1016/j.jss.2005.04.033
views Web applications as organisational initia-
tives and as such, it takes into account the need Bolchini, D., & Paolini, P. (2004). Goal-driven
to address strategic objectives, business processes requirements analysis for hypermedia-intensive
issues, requirements details of services, NFR, and Web applications. Requirements Engineering
integration with other systems. Journal Special Issue.

352
Focused Requirements Engineering Method for Web Application Development

Ceri, S. Fraternali, P., Bongio, A., Brambilla M., Kautz, K., & Madsen, S. (2003). Web development
Comai S., & Matera M. (2003). Designing data- - The differences, similarities and in-betweens.
intensive Web applications. Morgan Kaufman. Melbourne, Australia: Information Systems De-
velopment Conference.
Ginige, A. (2002). Web engineering: Managing the
complexity of Web systems development. In 14th Kotonya, G., & Sommerville, I. (1996). Require-
International Conference on Software Engineering ments engineering with viewpoints. BCS/IEE .
and Knowledge Engineering (SEKE ‘02), Ischia, Software Engineering Journal, 11(1), 5–18.
Italy: ACM Press.
Lowe, D. (2003). Web system requirements: An
Ginige, A., & Murugesan, S. (2001). The essence overview. Requirements Engineering Journal, 8,
of Web engineering: Managing the diversity and 102–113. doi:10.1007/s00766-002-0153-x
complexity of Web application development.
McDonald, A., & Welland, R. (2001). Agile Web
IEEE MultiMedia, 8(2), 22–25. doi:10.1109/
engineering (AWE) process. Scotland: Department
MMUL.2001.917968
of Computing Science, University of Glasgow.
Gordijn, J., Akkermans, H., & van Vliet, H. (2000).
Meldrum, M., & Rose, J. (2004). Activity based
Value based requirements creation for electronic
generation of requirements For Web-based in-
commerce applications. In 33rd Hawaii Inter-
formation systems: The SSM/ICDT approach.
national Conference on System Sciences, IEEE,
The 12th European Conference on Information
Hawaii, USA.
Systems (ECIS 2004), Turku Finland.
Haire, B., Henderson-Sellers, B., & Lowe, D.
Murugesan, S. (1999). Web engineering.
(2001). Supporting Web development in the open
ACM SIGWEB Newsletter, 8(3), 28–32.
process: Additional tasks. COMPSAC’2001:
doi:10.1145/951440.951445
International Computer Software and Applica-
tions Conference, Chicago, Illinois, USA, IEEE Murugesan, S., Deshpande, Y., Hansen, S., &
Computer Society. Ginige, A. (1999). Web engineering: A new dis-
cipline for development of Web-based systems. In
IEEE. (2004). Guide to the software engineering
Proceeding of the Int’l Conf. Software Engineering
body of knowledge. IEEE.
(ICSE) Workshop on Web Engineering (pp: 1-9).
Kaplan, R., & Norton, D. (1993). Putting the Sydney, Australia: IEEE Multimedia.
balanced scorecard to work. Harvard Business
Nuseibeh, B., & Easterbrook, S. (2000). Require-
Review, (September/October): 134–147.
ments engineering: A roadmap. In International
Kaplan, R., & Norton, D. (1996a). The balanced Conference on Software Engineering (ICSE-
scorecard: Translating strategy into action. Bos- 2000), Limerick, Ireland, ACM Press.
ton: Harvard Business School Press.
Pohl, K. (1994). The three dimensions of require-
Kaplan, R., & Norton, D. (1996b). Using the ments engineering. Information Systems, 19(3),
balanced scorecard as a strategic management 243–258. doi:10.1016/0306-4379(94)90044-2
system. Harvard Business Review, (January/
Potts, C., Takahashi, K., & Anton, A. (1994).
February): 75–85.
Inquiry-based requirements analysis. IEEE Soft-
ware, 11(2), 21–40. doi:10.1109/52.268952

353
Focused Requirements Engineering Method for Web Application Development

Pressman, R. S. (2004). Software engineering: A A Requirement: “A condition or capability


practitioner’s approach. McGraw-Hill. that must be met or fulfilled by a system to sat-
isfy a contract, standard, specification, or other
Sawyer, P., Sommerville, I., & Viller, S. (1996).
formally imposed documents.” (IEEE Standard,
PREview: Tackling the real concerns of require-
610,12-1990)
ments engineering. Lancaster: Computing Depart-
Requirements Engineering (RE): “The pro-
ment, Lancaster University.
cess of discovering that ‘purpose’ by identifying
Sawyer, P., Viller, S., & Sommerville, I. stakeholders and their needs, and documenting
(1997). Requirements process improvement them in a form that is amenable to analysis, com-
through the phased introduction of good munication, and subsequent implementation”
practice. Software Process Improvement and (Nuseibeh & Easterbrook, 2000)
Practice, 1, 19–34. doi:10.1002/(SICI)1099- Stakeholder: “Anyone who can share in-
1670(199703)3:1<19::AID-SPIP66>3.0.CO;2- formation about the system, its implementation
X constraints or the problem domain, including end
users, indirect users, other customer representa-
Sommerville, I. (1995). Software engineering.
tives and developers” (Potts et al., 1994)
Addison Wesley.
Viewpoint (VP): “Any one who may have
Sommerville, I., & Sawyer, P. (1997). Require- some direct or indirect influence on the system
ments engineering: A good practice guide. John requirements” (Kotonya & Sommerville, 1996)
Wiley & Sons Lts. Viewpoint Oriented Requirements Defini-
tion (VORD): “A software requirements engineer-
Vidgen, R., Avison, D., Wood, J. R. G., & Wood-
ing approach used to organise both the elicitation
Harper, A. T. (2002). Developing Web information
process and the requirements themselves into
systems. Butterworth Heinemann.
viewpoints” (Sommerville, 1995)
Web Business Application (WebApp): “An
application that tends to be used to integrate and
key Terms streamline an organisation’s business processes
beyond organisational (customers, agents, suppli-
Balanced Score Card (BSC): “A multidimen- ers, others) and geographical borders, to provide
sional framework for describing, implementing an organisation with competitive products and
and managing strategy at all levels of an enterprise services that give it a strategic advantage over its
by linking objectives, initiatives and measures to competitors in the marketplace
an organisation’s strategy” (Kaplan & Norton, to promote business innovation and/or to
1993, 1996 a & b) improve operational efficiency” (Al-Salem &
eWARE (extended Web application re- Abu-Samaha: 2005a)
quirements engineering): “A strategy-focused
requirements engineering method used to align
Web application requirements to business strat- endnote
egy and to elicit legal, technological, business,
1
marketing and content requirements” Throughout eWARE, a stakeholder /view-
point is used interchangeably.

This work was previously published in Encyclopedia of Information Science and Technology, Second Edition, edited by M.
Khosrow-Pour, pp. 1537-1545, copyright 2009 by Information Science Reference (an imprint of IGI Global).

354
355

Chapter 2.6
A Logic Programming Based
Framework for Intelligent
Web Service Composition
Enrico Pontelli
New Mexico State University, USA

Tran Cao Son


New Mexico State University, USA

Chitta Baral
Arizona State University, USA

Abstract ing the practical use of different service description


formalisms and ontologies.
This chapter presents a comprehensive logic pro-
gramming framework designed to support intelligent
composition of Web services. The underlying model Introduction
relies on the modeling of Web services as actions,
each described by a logic programming theory. One of the main goals of the Semantic Web initia-
This view allows the use of logic-based planning tive (Berners-Lee, Hendler, & Lassila, 2001) is
to address the Web service composition problem, to extend the existing Web technology to support
taking advantage of the fact that logic-based plan- the development of intelligent agents, which can
ning enables the elegant introduction of a number automatically and unambiguously process the
of extensions and generalizations (e.g., dealing with information available in millions of Web pages.
incomplete knowledge and preferences). The theory This led to numerous works on Web services and
describing each Web service is encoded as a logic Web service composition. The primary goal of Web
programming module, and different semantics are service composition is to determine an appropriate
allowed within different modules, thus better reflect- sequence of Web services to accomplish a user goal.
The majority of the existing proposals dealing with
DOI: 10.4018/978-1-60566-042-4.ch009 Web service composition build on the principle of

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
A Logic Programming

viewing Web services as actions, thus, represent- scalable planning systems have been devel-
ing the Web service composition problem as a oped and are available for use, such as FF
planning problem that can be addressed using (Hoffmann & Nebel, 2001), SHOP (Nau,
existing planning techniques. A second popular Cao, Lotem, & Muñoz-Avila, 1999), SAT-
approach to Web service composition relies on based planners (Kautz & Selman, 1996),
techniques developed in the area of workflow and logic programming based planners
development. The survey by Rao and Su (2004) (Lifschitz, 1999). All these planners can be
provides a good overview of various proposals used as the backbone in the development of
for Web service compositions. systems for Web service composition with
McIlraith and Son (2002) propose to use an architecture similar to the one described
GOLOG (Levesque, Reiter, Lesperance, Lin, by McIlraith and Son (2002).
& Scherl, 1997), a logic programming based • AI planning allows the Semantic Web re-
language, for Web service composition. In such search community to focus on the devel-
a proposal, each Web service is translated to opment of Web service representation and
a primitive action. GOLOG provides control- reasoning languages and tools for translat-
flow constructs, such as if-then-else, while-do, ing Web service representation into a plan-
sequence (denoted by ‘;’), procedure, and test ning language.
(denoted by ‘?’), which can be used to combine
the primitive actions into programs. The resulting Indeed, this view of Web service composition
programs can be provided to a GOLOG inter- has been embraced by many researchers, and a
preter for finding the sequence of Web services number of tools have been proposed, for example,
that need to be executed to achieve the goal of translators to map Web services encoded using
the user. Alternatively, the program can be given DAML-S or OWL-S to PDDL (PDDL Techni-
to an execution monitoring module, for direct cal Committee, 1998), a well-known planning
execution. Sufficient conditions for a successful language used by many planning systems.
execution of a program are also provided. This While the use of planning in Web service com-
direction of work has been adopted by Au, Kuter, position is advantageous, there are a number of is-
and Nau (2005), Kuter, Sirin, Nau, Parsia, and sues that need to be addressed before this approach
Hendler (2005), and Wu, Parsia, Sirin, Hendler, can be widely applied. The first problem, which
and Nau (2003), where SHOP2, a hierarchical can be termed as the service selection problem,
planning system, is used as the underlying system derives from the huge number of available services
for automatic Web service composition. The lat- that can be used to achieve the same goal. The
ter work also addressed an important aspect of second problem lies in the lack of information in
Web service composition, namely, the problem service composition, which imposes additional
of incompleteness of information in Web service requirements on the planning system, such as the
composition, by adding to the planning algorithm ability to plan with incomplete knowledge and to
a module for gathering information during the handle knowledge producing actions. The third
planning process. problem centers on the quality of the composed
Viewing Web service composition as high-level service, which can be translated into the problem
planning is not only natural, but also advantageous of planning with preferences, an area of research
for different reasons: that has only recently attracted interest from the
planning community. The fourth problem derives
• AI planning has made remarkable progress from the fact that each Web service is often encoded
in the last 10 years, and several robust and within an ontology or a knowledge base, whose

356
A Logic Programming

semantics is specified by the service provider, and problem faced by the Web service composition
hence, might be different from service to service. problem. Our specific objectives are:
A final problem, that has been recognized very
early in the development of the Semantic Web 1. The design of a theoretical framework for
(Berners-Lee & Fischetti, 1999; Berners-Lee et reasoning with heterogeneous knowledge
al., 2001), is the need for mechanisms to encode bases, which can be combined with logic pro-
rules, for example, rules for the description of gramming-based planners for Web service
Semantic Web services and business rules inter- composition. This framework supports:
change in e-commerce applications. a. The interoperation between knowledge
Answer set programming (ASP) is a declarative bases encoded using different rule
programming framework, originally proposed markup languages, and
by Marek and Truszczynski (1999) and Niemelä b. The development and integration of
(1999). To solve a problem in ASP, we translate different components that reason about
it into a logic program, whose answer sets cor- knowledge bases.
respond one-to-one to solutions of the original 2. The development of a prototype of the pro-
problems. ASP has found its way in several posed framework.
real-world applications, for example, diagnosis
of components of the space shuttle (Balduccini, The chapter is organized as follows. We start
Gelfond, & Nogueira, 2006). ASP has been widely by briefly reviewing the representation of Web
adopted in the area of reasoning about actions services and rules in the Semantic Web, the
and planning; it has been applied to solve vari- foundation of logic programming and answer set
ous forms of planning, such as classical planning, planning, the use of answer set planning in Web
conformant planning, and conditional planning service composition, and discuss the challenges
(e.g., Son, Tu, Gelfond, & Morales, 2005; Tu, faced by the current architecture for Web service
Son, & Baral, 2006). ASP has been used to in- composition. In the successive sections, we
corporate various forms of domain knowledge in describe the main contribution of this chapter, a
planning (Son, Baral, Tran, & McIlraith, 2006a) framework for reasoning with distributed hetero-
and to construct plans satisfying multidimensional geneous knowledge bases, and its implementation.
preferences (Son & Pontelli, 2006). This existing Finally, we relate our work to other proposals and
body of knowledge provides the foundations for provide conclusions and indications for future
the development of a logic programming-based developments.
framework for Web service composition that ad-
dresses all but the problem of dealing with Web
services encoded in heterogeneous knowledge Web Services and Logic
bases. This last problem is becoming more and Programming
more critical with the development of distinct
standards and ontologies for Web service encoding We begin with a review of the common elements
and the wider availability of Web services in the of Web services and the RuleML language, and the
Internet. This problem will be our main concern basics of logic programming. We then discuss how
in this chapter. logic programming can be used in Web service
In this chapter, we propose a logic program- composition and discuss the issues that need to
ming framework for reasoning with distributed be addressed.
heterogeneous knowledge bases that contain rules
and facts, that is, we concentrate on the fourth

357
A Logic Programming

Figure 1. CheckAvailability function for hotel reservation Web service

Web Services and RuleML If a service requesting agent would like to use
a service, it needs to place a request with the
Many Web service description languages have proper parameters. For example, Figure 1 shows
been proposed (e.g., www.w3c.org). The most an excerpt from Booth and Liu (2005), used to
recent proposal, WSDL 2.0 (Chinnici, Moreau, define the ‘checkAvailability’ function for a hotel
Ryman, & Weerawarana, 2007), describes a Web reservation service, which takes three parameters
service in terms of (i.e., checkInDate, checkOutDate, and roomType)
and returns either an error message indicating
• The kinds of messages that the service will that the date is improper (invalidDataError), or a
send/ receive (using the types element). number indicating the number of available rooms
• The functionalities of the service (inter- (checkAvailabilityResponse).1
face element). Many of the existing Web services are de-
• The way to access the service (binding scribed within some ontology (e.g., an OWL
element). description). More and more are associated with
• The location of the service (service a set of rules (also called a knowledge base). Such
element). a knowledge base is often written in one of the
variants of the RuleML markup language.
In essence, a service can be viewed as a The RuleML initiative2 is a response to the
collection of subroutines (or functions), and it need of a common XML-based rule markup lan-
is described by a valid WSDL 2.0 document. guage, which has precisely defined semantics and

358
A Logic Programming

Figure 2. RuleML language modularization


enables efficient implementations. In recent years,
a significant amount of work has been devoted
to develop knowledge representation languages
suitable for the task, and a variety of languages
for rule markup have been proposed. The initial
design (Boley, Grosof, Sintek, Tabe, & Wagner,
2002) included a distinction (in terms of distinct
DTDs) between reaction rules and derivation
rules. The first type of rules is used for the encoding
of event-condition-action (ECA) rules, while the
second is meant for the encoding of implicational
and inference rules.
The derivation rules component of the RuleML
initiative has originated a family of languages.
Figure 2, from Hirtle and Boley (2005), shows
the most commonly referred languages; observe
that Datalog plays the role of a core language,
with simplified versions (e.g., unary and binary
Datalog) developed for combining RuleML with
OWL (as in SWRL) (Horrocks, Patel-Schneider,
Boley, Tabet, Grosof, & Dean, 2004). Various
sublanguages have been created to include features
like explicit equality (e.g., fologeq), negation as
failure (e.g., naffolog), and Hilog layers (e.g.,
hohornlog). Specific instances of RuleML for mented either through translators, for example,
handling different aspects of Web services have GEDCOM (Dean, 2001), which translates to XSB
been proposed (The Policy RuleML Technical Prolog and JESS, or using independent inference
Group, 2004). engines, for example,
Kifer, de Bruijn, Boley, and Fensel (2005) argue
that any realistic architecture for the Semantic • j-DREW (Spencer, 2002), a top-down en-
Web must be based on various independent but gine for RuleML,
interoperable languages, one of them being the • DR-Device (Bassiliades, Antoniou, &
logic programming language with negation-as- Vlahavas, 2006), an engine supporting de-
failure. The use of rule-based languages requires feasible logic and both strong and default
the coexistence of different languages with dif- negation, and
ferent semantics and associated reasoning mecha- • CommonRules (Chan & Grosof, 1999),
nisms, and the need to integrate reasoning across a bottom-up engine for the Datalog
these languages. The need for these languages and sublanguage.
their interactions have been extensively discussed
(e.g., Kifer et al., 2005; May, Alferes, & Amador, It should be noted that RuleML, WSDL 2.0, and
2005) and it is at the foundation of the most recent the other markup languages employed when deal-
work of the Rule Interchange Format working ing with Web services are simply representation
group. It is also important to note that many of languages. An agent making use of Web services
the sublanguages of RuleML have been imple-

359
A Logic Programming

can be developed in any programming language appear qualified in a rule. A literal is an atom, a
(e.g., JAVA, C++, Perl, etc.). For this reason, qualified atom, or a formula not A, where A is
previous proposals for Web service composition an atom/qualified atom. not A is also referred to
assume the existence of translators which map as a negative literal. We will denote with BP the
Web service descriptions to representations that Herbrand base for this language (i.e., the set of
are more adequate for manipulation and reasoning. all ground atoms). For an atom (qualified atom,
For example, McIlraith and Son (2002) map Web negative literal) l, we denote with π(l) the predicate
service descriptions to situation calculus, while symbol used in l.
other recent proposals make use of encodings in A general rule is of the form
PDDL; for example, translators of Web service A ← B1,…,Bk (1)
descriptions (in DAML-S) to PDDL (McDermott, where A is an atom and B1,…,Bk are literals.
Dou, & Qi, 2002). Intuitively, a Rule (1) states that if the literals
For the purpose of this chapter, we will assume B1,…,Bk are true, then we can conclude that A is
the existence of translators that map Web service also true. We view interpretations and models as
descriptions into a logic programming representa- subsets of BP. For an atom A, the literal A (resp. not
tion. Many of such tools are already available on A) is true in a model M if A∈M (resp. A∉M).
the Web (e.g., Rainer, 2005). Depending on the type of programs we wish
to represent, different restrictions can be imposed
Logic Programming on the rules:

We will consider a logic programming language a. Datalog: The Bi’s in Rule (1) can be only
〈ℑ,Π,ϑ〉, where ℑ is a denumerable collection of atoms/qualified atoms and the terms used in
function symbols, Π = Πu ∪ Πd is a denumerable the literals can be only variables or constants
collection of predicate symbols, and ϑ is a col- (i.e., of arity 0).
lection of variables. Πu are called user-defined b. Ground datalog: The Bi’s in Rule (1) can
predicates, while ∏d are called built-in predicates. only be atoms/qualified atoms, and the only
We will assume that Πu ∩ Πd = ∅. We will denote terms allowed are constants.
with ar(α) the arity of the symbol α ∈ Π∪ℑ. We c. Ground binary datalog: The rules satisfy
typically assume that Πd contains at least the the conditions of Case (b), and in addition
predicates assert, retract, and model. we require all predicates used to construct
A term is either a variable, an element of ℑ of atoms to have arity at most 2.
arity 0 (i.e., a constant), or an expression of the d. Datalog with negation: The rules have the
form f(t1,…,tn) where f∈ℑ, ar(f)=n and t1,…,tn format as in Case (a) but negative literals
are terms. We will say that a term t is ground if it (not A) are allowed in the body of the rule.
does not contain variables. We will denote with
ΗP the Herbrand universe for this language, that We will refer to a rule as a Ξ-rule (where Ξ is
is, the set of all ground terms. Datalog, ground Datalog, binary Datalog, etc.) to
An atom is a formula of the form p(t1,…,tn) denote a rule that meets the corresponding require-
where p∈Π, ar(p)=n and t1,…,tn are terms. The ments. A Ξ-program is a collection of Ξ-rules.
atom is ground if t1,…,tn are ground. A qualified Given a rule r, we denote with used(r) the set of
atom is a formula of the form t:A where t is a ground terms t such that t is a label of a qualified
ground term (called the label of the qualified atom) atom in r. Given a Ξ-program, we denote with
and A is an atom. In particular, if the predicate p used(P) = { t | ∃ r ∈P, t ∈ used(r) }.
of an atom belongs to Πd, then the atom can only We also introduce

360
A Logic Programming

def(P) = { p | p∈Π, ar(p)=k, ∃r∈P. ∃ t1,…,tk, Answer Set Planning


head(r) = p(t1,…,tk) }.
Intuitively, a Ξ-program encodes a knowledge Let us consider the classical AI problem of going
base, whose semantics is defined by a class of to the airport (McCarthy, 1959). In this problem,
models satisfying certain properties. It should we have an action, named drive_home_airport;
be noted that the semantics of a program might in the initial state, we are at home and have a car.
be defined in different ways and depends on the We know that driving will bring us to the airport
program type. For Datalog programs, the least and our goal is to go be at the airport.
fixpoint semantics can be used (van Emden & In answer set planning, this problem is encoded
Kowalski, 1976). However, for Datalog with ne- by a Datalog with negation program that consists
gation programs, the well-founded (Van Gelder, of the different groups of rules described below. In
Ross, & Schlipf, 1991) or the answer set semantics each of these rules, T denotes a number between
can be used (Gelfond & Lifschitz, 1988). 0 and a predefined constant length, indicating
the maximal length of the plan that we wish to
Planning in Logic Programming compute. The rules
and Web Service Composition action(drive_home_airport) ←
fluent(at_home) ←
A planning problem is specified by a domain, with fluent(at_airport) ←
its properties (called fluents) and actions with their fluent(car_available) ←
preconditions and effects, a specification about the specify the fluents and the actions of the do-
initial state of the world, and a formula describing main. The rules
the final state (goal). The objective is to determine false←not holds(at_home,T), occ(drive_
a sequence of actions that can transform the initial home_airport), T)
state into a state satisfying the goal. false← not holds(car_available,T), occ(drive_
Logic programming under the answer set home_airport), T)
semantics was first used in planning by Subrah- encode the precondition under which the ac-
manian and Zaniolo (1995) and is now known as tion drive_home_airport can be executed. They
answer set planning (Lifschitz, 2002). Answer set state that the action drive_home_airport can only
planning has gained popularity thanks to the devel- occur at the time T if at_home and car_available
opment of fast answer set solvers (e.g., Smodels, are true at the time T. The rules
Simons, Niemelä, & Soininen, 2002 and DLV holds(at_airport,T+1)←occ(drive_home_air-
Eiter, Leone, Mateis, Pfeifer, & Scarcello, 1998). port, T)
Since our intention is to use logic programming holds(neg(at_home),T+1)←occ(drive_home_
for Web service composition, following the AI airport, T)
planning approach, we will give a brief overview say that if the drive_home_airport action oc-
of answer set planning by illustrating its key ideas curs at the time T, at_airport and at_home will
through an example. As noted earlier, existing be true and false, respectively, at the time T+1.
tools can be adapted to translate Web service The two rules
descriptions into the representation discussed in occ(drive_home_airport,T)←not not_
this chapter. occ(drive_home_airport,T)
not_occ(drive_home_airport,T)←not
occ(drive_home_airport,T)
are often referred to as generation rules, and
are used to generate action occurrences. Intui-

361
A Logic Programming

tively, they state that the action either occurs or • available: there are some rooms available
not. The rules • reserved: the reservation has been
holds(at_airport,T+1)←holds(at_airport, T), successful
not holds(neg(at_airport), T+1)
holds(neg(at_airport),T+1)←holds(neg(at_ They can be encoded as follows
airport), T), fluent(available)←
not holds(at_airport, T+1) fluent(reserved)←
known as inertial rules, encode the fact that The effect of the action checkAvailability is
a fluent normally does not change its value. (We different than the effect of the action makeRes-
omit the rules for at_home and car_available for ervation. The execution of checkAvailability is
brevity.) either available or ¬available, while the execu-
Finally, the initial state is encoded by a set of tion of makeReservation is reserved. The first
facts of the form action, called knowledge producing action, does
holds(neg(at_airport),0) ← not change the state of the world (it does not alter
holds(at_home, 0) ← the number of available rooms) while the second
and the goal is expressed by the rule one does (it reduces the number of available
false←not holds(at_airport,length) rooms by 1).
Each answer set of the above program cor- Knowledge producing actions are often needed
responds to a plan achieving the goal at_airport. when a reasoner does not have complete informa-
For example, for length=1, the program yields tion about the state of the world. For example, an
an answer set corresponding to the plan [drive_ agent that uses the above service to place a res-
home_airport]. For more on answer set planning, ervation often does not know whether or not the
the interested reader is referred to Lifschitz’s hotel would have a room. In other words, neither
(2002) work. available nor ¬available is true in the initial state.
The agent knows about the value of available
Answer Set Planning and (or ¬available) after the execution of the action
Web Service Composition checkAvailability. In our previous work (i.e., Son
et al., 2005; Tu, Son, & Baral, 2006), we discuss
Answer set planning can be used for Web service the use of logic programming in reasoning and
composition, but it requires some modifications planning with knowledge producing actions and
and extensions. For example, the functions of the incomplete information. We also extended an-
hotel reservation Web service (Figure 1) can be swer set planning to incorporate different types
encoded by the following actions: of domain knowledge and user preferences (Son,
action(checkAvailability(DateIn,DateOut,Ro Baral, Tran, & McIlraith, 2006b; Son & Pontelli,
omType))←date(DateIn), 2006).
date(DaeOut),
type(RoomType) Current Architecture and Challenges
action(makeReservation(DateIn,DateOut,Ro
omType))←date(DateIn), The techniques developed so far can be used in
date(DaeOut), the implementation of a Web service composi-
type(RoomType) tion framework similar to other AI-planning Web
It is reasonable to consider the two following service composition frameworks (Figure 3). The
fluents: basic components of this approach are:

362
A Logic Programming

Figure 3. Schematic architecture

• A translator which reads Web service de- system is adequate for the mentioned application,
scriptions and translates them into an ad- the new developments in Semantic Web present
equate logic programming theory (e.g., a two technical challenges to this architecture.
Datalog with negation program). The first problem lies in the fact that knowl-
• A reasoning/planning engine which ac- edge (and hence, Web services) often comes with
cepts the output of the translator and cre- reasoning rules, that were not the focus of earlier
ates a plan that achieves the goal of the Web service description languages, and hence
service composition problem. this aspect has not been addressed in previous
• An execution and monitoring module implementations. The second challenge is the need
which is responsible for the execution of to consider different semantics embedded in the
the plan. encoding of a Web service, since the underlying
semantics of various RuleML knowledge bases
Architectures analogous to the one depicted are specified by the knowledge engineers and not
in Figure 3 have been successfully implemented, by the application developers. In dealing with this
for example, in the development of systems for problem, we can identify a number of issues that
Web service composition of bioinformatics Web need to be addressed:
services (e.g., (Pan, Tu, Pontelli, & Son, 2004)).
In this application, the results of the translation 1. Reasoning within one knowledge base: Being
are Datalog, Datalog with negation, or a similar able to reason within a knowledge base im-
variant language. The reasoning/planning engine plies the ability to interoperate with a com-
employs the answer set semantics for this program. putational framework capable of handling the
Furthermore, it is assumed that all necessary infor- type of knowledge present in the knowledge
mation has been incorporated into the encodings of base (e.g., a Prolog system for hohorn rules,
the corresponding services. Although the resulting a Datalog system for Datalog rules).

363
A Logic Programming

2. Reasoning across different knowledge bases: & Honavar, 2007). In particular, it has been
The capability of combining knowledge is observed that some incorrect annotations in the
essential for intelligent agents (e.g., this is Gene Ontology are characterized by a selection
necessary in the context of services composi- code equal to reviewed computational analysis
tion). This requires (RCA). We consider the following services and
a. The ability to exchange inference knowledge bases:
results between different knowledge
bases (e.g., the interoperability problem • NCBI GenBank Genome-to-Protein
between rules and OWL described by Search: This service expects as input the
Kifer et al. [2005]); RefSeq of a genome, and produces as re-
b. The ability to combine reasoning re- sults the NCBI ids of the proteins coded by
sults produced by different reasoning the genes in the considered genome;
engines; and • Protein Class Ontology: Gene Ontology
c. The ability to properly scope the includes a taxonomy of functional classi-
reasoning w.r.t. a specific knowledge fication of proteins (e.g., its OBO encod-
base (e.g., the scoped inference issue ing can be found in “Gene Ontology, Edit
described by Kifer et al. [2005]). 1.101, ” [2007]). The taxonomy can be
3. Utilizing available knowledge: This requires directly encoded in Datalog; sample rules
the ability to use the results produced by extracted are:
different reasoning processes in the construc- id(0016301)
tion/implementation of complex Semantic name(0016301, ‘kinase activity’)
Web applications. namespace(0016301,molecular_function)
is_a(0016301,0016772)
A Motivating Example ...
id(0004672)
Let us illustrate some of these issues within an name(0004672,’protein kinase activity’)
example, drawn from the field of bioinformatics. alt_id(0050222)
This application domain is particularly interesting xref(reactome(4030))
with respect to the issue of Web service composi- is_a(0004672,0016301)
tion; in recent years a large number of Web services is_a(0004672,0016773)
have been developed, and they rely on a large ...
collection of heterogeneous (and often redundant) Ontological reasoning allows us, for example,
ontologies and knowledge bases. to discover subclasses and superclasses of a given
The problem at hand is to obtain reliable one.
functional annotation, according to the Gene
Ontology (Gene Ontology Consortium, 2007), • Functional Annotation Service: The input
of the proteins coded by the genes present in the is an id of a protein. If the protein belongs
genome of an organism. The input is represented to a class whose functional annotation is
by the NCBI RefSeq id of a genome (i.e., an id not RCA, then the annotation is retrieved
of the form NC_xxxx). It has been observed that from the Gene Ontology annotation data-
functional annotations retrieved by AmiGO (the base and returned. Otherwise, the annota-
query system to the Gene Ontology Database) tion is extracted from the UniProt data-
are occasionally unreliable (e.g., Andorf, Dobbs, base (and returned in UniProt format). The

364
A Logic Programming

semantics of this service requires the use of A Framework for


more complex rules (encoded in RuleML Interoperation and
using the nafdatalog language). Some of Composition
the Datalog with negation rules are:
suspect(ProteinClass) ← In this section, we describe our framework for
amigo:evidence_code(ProteinClass,rca) interoperation and composition of distributed
classification(ProteinClass, heterogeneous knowledge bases. We begin with
FuncAnnotation,go)← an overview of the approach. We then present the
amigo:is_a(ProteinClass,FuncAnnotation), architecture of the component needed to address
not suspect(ProteinClass) the challenges faced by the current Web service
classification(ProteinClass,FuncAnnotation, composition methodology. The precise syntax and
uniprot) ← semantics of the framework are discussed next.
suspect(ProteinClass),
interpro2go:entity_xref(ProteinClass,X), 
uni Overview of the Approach
prot:is_a(X,FuncAnnotation)
• Gene Ontology to InterPro Ontology The objective is to provide a formal logic-based
Mapping: This mapping, expressed in a framework that allows the development of the
ground Prolog facts database is available planning/reasoning component of Figure 3. The
at the “OBO download matrix” (2007). approach we follow is intuitively depicted in
Figure 4. The framework is assembled as a col-
Let us assume that the initial state includes lection of modules, each containing a logic pro-
the RefSeq NC_003075 (Arabidopsis thaliana gramming theory, potentially requiring a different
chromosome 4) and we are interested only in semantics (e.g., datalog/pure prolog, constraint
proteins that are enzyme inhibitors. The genome logic programming, answer set programming,
includes 4,817 coding genes. One of the proteins etc.) The modules are the result of a logic pro-
generated is AT4G00080. The Gene Ontology can gramming encoding of the semantics of the Web
be used (through a transitive closure of the is_a services description (left side of Figure 4) and of
relationship) to determine that the protein belongs the underlying associated static knowledge (e.g.,
to the desired class. Accessing the functional an- ontologies – shown on the right in Figure 4). The
notation, we discover that the protein is annotated framework assumes the possibility of dependen-
as a pectinesterase inhibitor (GO:0046910), but cies between the different modules, for example,
it is RCA. This means that the desired annotation ontologies can be hierarchically defined.
will be extracted from UniProt, and this returns The answer set planner is encoded using a
an annotation IPR006501. The mapping between modular view of the action domain. Each service
ontologies contains the fact is represented by a logic programming module,
metadata_db:entity_ which exports the description of the service as
xref(‘GO:0046910’,’IPR006501’) an action; in particular, the module exports the
This will eventually produce the same Gene description of the action in terms of its execut-
Ontology annotation. ability conditions (predicate executable) and its
dynamic effects (predicate causes). For example,
in the case of the functional annotation service
mentioned previously, these predicates could be
defined as:

365
A Logic Programming

Figure 4. Framework architecture

executable([have_protein(X), have_protein_ executability conditions (which are tested in the


class(C)]) state of the world at time T) and the effects of the
causes(go_classification(X,C)) ← action.
have_protein(X), Static causal laws will be employed to allow
have_protein_class(Class), the planner to make use of background knowledge
classification(Class,C,go) used by the Web services (e.g., ontologies, data-
causes([validated_annotation,uniprot_ bases). For example, the protein class ontology,
classification(X,C)]) ← imported as a Prolog module, would allow us to
have_protein(X), derive rules like:
have_protein_class(Class), holds(have_protein_class(C),T) ←
classification(Class,C,uniprot) holds(have_protein_class(C1),T),
The planner will use these predicates to develop gene_ontology:is_a(T1,T)
plans, for example, using rules like:
holds(F,T+1) ← action(A), Architecture
occ(A,T),
A:causes(F), The approach adopted in this work relies on the
A:executable(List), use of a core logic programming framework to
holds(List,T) address the issues of integration and interoperation.
where the execution of the action A requires In particular, the spirit of our approach relies on
recovering from the corresponding module (which the following beliefs:
we assume has the same name as the action) the

366
A Logic Programming

• The natural semantics of various languages and reals), fuzzy Prolog, and a declarative
for Web service description, ontology de- ODBC interface. In addition, CIAO Prolog
scription, and various levels of the RuleML provides a mechanism that allows Prolog
deduction rules hierarchy can be natu- programs to invoke Java methods, offer-
rally captured by different flavors of logic ing a bidirectional communication and a
programming. reflection of Java objects into Prolog. This
• Modern logic programming systems provide provides, for example, a natural way to ex-
foreign interfaces that allow declarative in- ecute Java-based engines (e.g., Jess) and
terfacing to other programming paradigms. communicate between the core framework
and external Java packages. Furthermore,
The idea is to combine the ASP-Prolog CIAO Prolog includes the PiLLoW library,
framework of Elkhatib, Pontelli, and Son, a standardized Prolog library for Web pro-
(2006)—which allows CIAO Prolog (Bueno, gramming, which provides the framework
Cabeza, Carro, Hermenegildo, López-García, with capabilities for Web access (e.g., man-
& Puebla, 1997) programs to access and modify agement of URLs) and parsing of HTML
modules containing answer set programming and XML documents to Prolog terms.
(ASP) code (Niemelä, 1999)—with the notation Thus, we envision the core framework as
for modularization of answer set programming of the bridge between distinct execution mod-
Answar, Baral, and Tari (2005) and Baral, Dzifcak, els for heterogeneous knowledge bases.
and Takahashi (2006). The result is a logic pro- • Issue 2:
gramming framework, where modules responding ◦◦ Issue 2(a): This issue will be ad-
to different logic programming semantics (e.g., dressed through the introduction of
Herbrand minimal model, well-founded seman- a module system, where different
tics, and answer set semantics) can coexist and knowledge bases can be encoded (di-
interoperate. The framework provides a natural rectly or indirectly) as distinct mod-
answer to the problems of use and interoperation ules. The original import/export of
of RuleML and other Web service semantic de- CIAO Prolog can be combined with
scriptions described earlier. The overall structure the languages for answer set modules
is depicted in Figure 4. Most of the emphasis is of Baral et al. (2006) to allow forms
on the use of answer set programming to handle of bidirectional communication be-
some of the RuleML sublanguages (e.g., datalog, tween the core framework and the
ur-datalog, nafdatalog, and negdatalog), even modules representing the knowledge
though the core framework will naturally sup- bases.
port most of the languages (e.g., hornlog and ◦◦ Issue 2(b): The core framework will
hohornlog). provide the full computational power
The problems mentioned in the previous sec- of Prolog, constraint logic program-
tion are addressed by the proposed framework ming, and answer set programming,
as follows: combined through a sophisticated
module and class system. Module
• Issue 1: CIAO Prolog offers direct access to interfaces will allow extraction of se-
a collection of modules that support differ- mantic information from the various
ent forms of logic programming reasoning, knowledge bases (e.g., result of que-
for example, traditional Prolog, constraint ries, models of knowledge bases) and
logic programming (over finite domains reason with them.

367
A Logic Programming

◦◦ Issue 2(c): The scoped inference is A program P={Mt1, …, Mtk} is a collection of


naturally supported by the module modules named t1,…,tk. The graph of P (denoted
system of ASP-Prolog, for example, by graph(P)) is a graph (N,E), where the set of
skeptical and credulous reasoning nodes N is {t1,…,tk}and
w.r.t. answer set modules. A program P is admissible if it satisfies the
• Issue 3: This aspect can be handled thanks following properties:
to the combination, in ASP-Prolog and
CIAO Prolog, of Web access capabilities • for each tj we have that imp(tj) ⊆ {t1,…,tk};
along with the full computational power of and
Prolog. • the graph graph(P) is acyclic.

We will now introduce the basic structure of The module structure can be expanded by al-
our framework for modules and module inter- lowing cyclic dependencies (i.e., two-way com-
operation. munications between modules) as well as OO-style
organization of modules (e.g., as described by
Module Structure Baral et al. [2006]). We omit this discussion due
to lack of space.
As mentioned in the introduction to logic program- Example: Let us consider the modules associ-
ming, we will assume that assert,retract,model,… ated to the bioinformatics services described earlier.
are elements of ∏d. Given a Ξ-program P, we One module, gene_ontology, is derived from the
say that P is pure if neither assert nor retract ap- Gene Ontology. It will have an overall structure
of the form:
pear in the rules of P; otherwise we say that P is
← module: gene_ontology
impure.
← export: is_a/2, evidence_code/2, …
A module is composed of two parts: a module
id(0016301)
interface and a module body. A module interface
is_a(0016301,0016772)
has the form
is_a(0004672,0016301)
← module: t is_a(0004674,0004672)
← import: t1,…,tk …
← export: q1/k1,…,qm/km is_a(X,Y) ← is_a(X,Z), is_a(Z,Y)
where …
• t is a ground term, called the name of the evidence_code(0016772, rca)
module, …
• t1,…,tk are ground terms, representing A second module is obtained from the RuleML
names of other modules, and semantics of the functional annotation service:
• q1,…,qm are predicates, and are k1,…,km ← module: validated_annotation
non-negative integers, such that ar(qi)=ki. ← import: gene_ontology, …
← export: executable/1, causes/1, …
The body of a module is a Ξ-program, for a suspect(ProteinClass) ←
given Ξ. In that case, we will say that the module gene_ontology:evidence_
is a Ξ-module. Given a module named t, we iden- code(ProteinClass,rca)
tify with the export set of t (denoted by exp(t)) the classification(ProteinClass,
predicates q1,…,qm exported by t. We also identify FuncAnnotation,go) ←
with imp(p) the import set of t, that is, the names gene_ontology:is_a(ProteinClass,FuncAnno
of the modules imported by t. tation),

368
A Logic Programming

not suspect(ProteinClass) This suggests a natural way to handle the


... semantics Φτ of a program P. Φτ is a mapping of
The resulting program P will be a set of modules the form Φτ: P → 2BP. Once the topological sort
{ planner, gene_ontology, validated_annotation, η1,…, ηk of the modules is given, we can construct
...}. □ Φτ as follows:

General Semantics • the semantics of Mη1 is given, since it does


not import any other modules, and
In order to understand the framework, it is im- Φτ(Mη1) = NAT(Mη1)
portant to clearly define the semantics of the • the semantics of Mηi can be constructed by
module-based system. We start by considering computing the natural semantics of a “re-
pure programs, and then we extend the discussion duct” of the module,
to impure programs. Φτ(Mηi) = NAT(MR(Mηi, Φτ))
as defined next.
Pure Programs Let us consider a module Mηi of P. Then:

In this section, we propose a model-theoretic • If t: A is a ground qualified atom and t ∈


semantics for programs that do not contain any imp(Mηi), then
impure modules. Given a program, a model nam- Mηi╞═Φτ t: A iff for each model M ∈ Φτ(Mt)
ing function τ is an onto function τ: HP → 2BP. We we have that M╞═ A.
will use this function to assign distinct names to • If t: A is a ground qualified atom and t ∉
the models of the different modules. In the rest imp(Mηi), then Mηi╞═Φτ t: A if there exists
of this work, we will assume that the function τ x ∈ imp(Mηi), M ∈ Φτ(Mx) such that π(A) ∈
is fixed. exp(x), τ(t)=M and M╞═ A.
Given a program P = {Mt1, …, Mtk}, the • Mηi╞═Φτ t: model(s) if t ∈ imp (Mηi) and
acyclic nature of graph(P) guarantees the abil- τ(s) ∈ Φτ(Mt).
ity to construct a topological sort of {t1,…,tk}, • If not t: A is a ground qualified literal and
say η1,…, ηk, such that if (ηi,ηj) is an edge of the t∈imp(Mηi), then
graph, then i < j. Mηi╞═Φτ not t: A if Mηi╞/═Φτ t: A.
Given the program P and a topological sorting The model reduct of Mi w.r.t. Φτ, denoted
of the modules η1,…, ηk, we construct the seman- MR(Mi, Φτ), is defined as follows:
tics module by module, following the topological
sort order. The semantics of each module Mi will • Remove from Mi all rules that contain in
be given by a collection of models Φτ(Mi), where the body a qualified element l such that
Φτ(Mi) ⊆ 2BP. Given a Ξ-program T, not contain- Mi╞/═Φτl.
ing any qualified atoms and not containing any • Remove from the remaining rules all oc-
occurrence of predicates from Πd, we assume that currences of qualified elements.
its semantics NAT(T) is given. For example, if T
is a Datalog with negation program meeting these One can easily see that MR(Mi, Φτ) is a pro-
conditions, then NAT(T) will be the set of answer gram without qualified atoms whose semantics is
sets of T, while if T is a pure Prolog program defined by NAT. This allows us to set
(i.e., definite logic program without extra-logical Φτ(Mηi) = NAT(MR(Mηi, Φτ))
predicates such as assert and retract), then NAT(T) From now on, we will denote with Φτ/P the
contains the least Herbrand model of T. semantics of a program P.

369
A Logic Programming

Example: Let us continue the example we Let P={Mt1, …, Mtk}be a program; for the sake
started earlier. The module gene_ontology is a of simplicity, we assume that t1,…,tk is a topo-
Datalog program. Thus, NAT(gene_ontology) = { logical sort of graph(P). We also consider impure
M0 } where M0 is its least Herbrand model; some programs under the following restrictions:
of the elements in M0 are:
is_a(0016301,0016772), • The planner is itself a module, encoded us-
is_a(0004672,0016301), ing Prolog (pure or impure), and it is repre-
is_a(0004674,0004672), sented by the module Mtk.
is_a(0004672,0016772), • The impure predicates assert and retract are
is_a(0004674,0016772), allowed to appear only in Prolog modules.
is_a(0004674,0016301),...
The module validated_annotation is a Data- Because of the nonlogical nature of the impure
log with negation program, and its reduct will predicates, we rely on an operational semantics to
be performed with respect to the semantics of chaacterize the meaning of programs.
gene_ontology (and the other imported modules). The state of a computation is given by a tuple
For example, the rule 〈G,θ, P〉, where G is a Prolog goal, θ is a substitu-
classification(ProteinClass, FuncAnnotation, tion, and P is a program. The operational seman-
go) ← tics for a goal executed in the module named ti is
gene_ontology:is_a(ProteinClass, FuncAn- defined through a state transition system 〈G, θ,
notation), P〉 →i 〈G’, θ’, P’〉 where
not suspect(ProteinClass)
will be grounded and only the instances for • If G = (A ∧ Rest) and π(A) ∈ def(Mti), and
which the is_a atom is true in M0 will be kept, h ← body is a variant of a rule in Mti such
for example, that σ is the most general unifier of Aθ
classification(0004674, 0016772, go) ← not and h, then we have that G’=body∧Rest,
suspect(0004674) θ’=θσ,3 and P’=P.
Only the answer sets of the resulting program • If G = (t: A ∧ Rest), tj ∈ imp(Mti), π(A) ∈
will be maintained as NAT(validated_annota- def(Mtj), tj is a Prolog module, and 
〈A, θ,
tion). □ P〉 →j 〈□, θ’, P’〉 then we have G’=Rest
• If G = (t: A ∧ Rest), tj ∈ imp(Mti), π(A) ∈
Impure Programs def(Mtj), tj is an ASP/datalog module, and
let σ be a ground substitution fpor Aθ such
We say that a program is impure if it contains that Aθσ is true in each model in Φτ/P(Mtj)
impure modules and/or it contains modules that then we have that G’=Rest, θ’=θσ, and
are not based on logic programming. The use of P’=P.
impure programs allows the planner to provide • If G = t: model(t’) ∧ Rest and t ∈ imp(Mti),
the Web services with knowledge about the state and σ is a ground substitution such that
of the world the services will be executed in. It t’θ is ground and τ(t’θ) ∈ Φτ/P(Mt), then
also allows the reasoner to temporarily modify G’=Rest, θ’=θσ, and P’=P.
the content of a module (e.g., a module describ- • If G = t: A ∧ Rest and there exists t’ ∈
ing an action) to support the planning process. imp(Mti), and σ is a substitution such that (t:
This is particularly important when dealing with A)θσ is ground, τ(tσ) ∈ Φτ/P(Mt’), and Aθσ
very complex plans (Son & Pontelli, 2007) or is true in τ(tσ), then we have that G’=Rest,
incomplete knowledge (Tu et al., 2006). θ’=θσ, and P’=P.

370
A Logic Programming

• If G = t: assert(Head, Body) ∧ Rest and σ is to quickly respond to a request. This is


a substitution such that tσ∈imp(Mti), then translated in a fact of the form low_load or
we have that G’=Rest, θ’=θσ, and P’=(P\ high_load. The logic programming mod-
{Mtσ})∪ Ntσ where Ntσ = Mtσ ∪ { (Head ← ule allows the action to successfully return
Body) σ }. only in case of low load, that is, the defini-
• If G = t: retract(Head, Body) ∧ Rest and σ is tion of the action in the service description
a ground substitution such that tσ∈imp(Mti) module will include
and (Head ← Body)σ ∈ Mtσ, then we have executable(…):- low_load.
that G’=Rest, θ’=θσ, and P’ = (P\{Mtσ}) ∪ If the resulting plan is sufficiently long, we
Ntσ where Ntσ= Mtσ\{(Head ← Body)σ}. would like to replace the load information with a
pessimistic value, which may reflect the fact that
Given a goal G and a program P with main the load figure is not reliable any longer. We could
module Mtk, we say that θ is a solution of G if modify the rule used by the planner:
where →* denotes an arbitrarily long sequence holds(F,T+1) ← action(A), occ(A,T),
of transitions. (T > Threshold -> retract(A:low_load);
Example: Let us continue with our example. true),
The ability to perform service composition using A:causes(F), A:executable(List),
logic-based planning provides the added advan- holds(List,T)
tage that the plan and the complete trajectory are □
represented by logic statements, and it is possible Framework Implementation
to write additional modules to reason about the
trajectory. For example: Logical Core Implementation

• Given a list of proteins L we can write a The implementation of the logical core is based
simple query that will provide the func- on the combination of two logic programming
tional class to which the majority of the systems: CIAO Prolog (Bueno et al., 1997) and
proteins belong to: Smodels (Niemelä & Simons, 1997). CIAO Prolog
majority(L, Class) ← is a full-fledged Prolog system, with a sophisti-
findall(C,(member(X,L),annotate(X,C)),Fu cated module system, and designed to handle a
ns), variety of flavors of logic programming, includ-
findall(K-F, (member(A,Funs), ing constraint logic programming (over reals and
findall(A,member(A,Funs),S), finite domains), fuzzy logic programming, and
length(S,K) concurrent logic programming. Smodels is a logic
), programming engine which supports computation
Res), of the well-founded and answer set semantics for
keysort(Res,Result), NAF-datalog programs. The proposed system is
last(Result,Class) composed of two parts: a preprocessor and the
annotate(P,F) ← actual CIAO Prolog system.
assert(planner:initially(have_protein(P))), The input to the preprocessor is composed
planner:holds(go_classification(P,F),length) of
• Let us assume that the service description
includes an estimate of the reliability of the • The main planner module (Pr), typically a
service, provided by the service provider Prolog module, which extracts information
and measuring the ability of the service from a planner encoded in ASP;

371
A Logic Programming

• A collection of CIAO Prolog modules code, a representation of the parameters to be


(m1,…, mn); used for the computation of the answer sets (e.g.,
• A collection of ASP modules (e1,…, em). values of constants), a list containing the objects
representing the models of the ASP module, a
The main task of the preprocessor is to trans- counter of the number of answer sets currently
form the ASP modules to enable their use in the present, and so forth.
CIAO Prolog system and to manage the interaction The preprocessor creates a graph-module
with the Smodels solver. structure which represents the call hierarchy of
The transformation of each ASP module modules. If a Prolog/ASP module up calling an-
leads to the creation of two entities that will be other prolog/ASP module uc then up is called the
employed during the actual program execution, parent module and uc is one of its children. The
that is, an interface module and a model class. tree-module structure is stored in each interface
These are described in the following subsections. module.
The preprocessor will also automatically invoke Each interface provides also a timestamp
the CIAO Prolog top-level and load all the ap- predicate, which is used to inform of the time
propriate modules for execution. The interaction at which the module’s semantics have been last
with the user is the same as the standard CIAO computed (recorded as a discrete system time);
Prolog top-level. each interface module will recompute the local
semantics whenever the timestamp of one of
Interface Modules the imported modules changes. This allows the
system to propagate the effect of changes (e.g.,
The preprocessor generates one interface module assert/retract) to all modules that depend on the
for each ASP module present in the original input modified one.
program. The interface module is implemented as Finally, the preprocessing will add code to
a standard CIAO Prolog module and it provides handle the different types of goal qualification.
the client Prolog modules with the predicates
used to access and manage the ASP module. The Model Classes
interface module is created for each ASP module
by instantiating a generic module skeleton The preprocessor generates a CIAO class defini-
The overall structure of the interface module tion for each module. The objects obtained from the
is illustrated in Figure 5. The module has an instantiation of such class will be used to represent
export list which includes all the predicates used the individual models of the module. In particular,
to manipulate ASP modules (e.g., assert, retract, for an ASP module we will have one instance for
model) as well as all the predicates that are defined each answer set, while for a Prolog module we
within the ASP module. will have a single instance, corresponding to the
Each module has an initialization part, which least Herbrand model.
is in charge of setting up the internal data struc- Prolog and ASP modules can obtain reference
tures and invoke the ASP solver (Smodels) for the to these objects (e.g., using the model predicate
first time on the ASP module. The result of the supplied by the interface module) and use them
computation of the models will be encoded as a to directly query the content of one model. The
collection of model objects (see the description definition of the class is obtained through a
of the model classes in the next subsection). The straightforward parsing of the export declaration of
module will maintain a number of internal data each module, to collect the names of the predicates
structures, including a representation of the ASP defined in it; the class will provide a public method

372
A Logic Programming

Figure 5. Structure of an interface module

for each of the predicates present in the module’s fication of the form (α: t) have to be computed
export list. The class also defines a public method and stored as new facts with a newly created
add/1 which is used by the interface module to ASP-variable. Therefore, if we have a positive
initialize the content of the model. qualification literal (α: t), then the partial grounder
To facilitate computation, the preprocessor will evaluate the predicate t in module α. If α is a
generates an additional module for each ASP Prolog module, then the built-in predicate findall
module, called skeptical module. The module is is used to compute all possible answers to the goal
instantiated to a collection of facts representing ?-t. If α is an ASP-module, then the predicate t
the elements that have the same truth value in all is evaluated against the skeptical answers of the
answer sets of the corresponding ASP module. ASP-module α. If α is an ASP-model name, then
predicate t is computed against the corresponding
Further Implementation Details answer set named α. After that, all ground values
of t are added as facts into the ASP-module.
Further details about the implementation of this The same rewriting rules are needed for nega-
framework can be found by Pontelli, Son, and tive qualification literals (not α: t), domain quali-
Baral (2006). Let us discuss only some significant fication literals, cardinality qualification literals,
aspects here. and weight qualification literals.
ASP-State: Every ASP-module has a set of RuleML Specific Issues: RuleML knowledge
models, atoms, and skeptical model called ASP- bases are retrieved and encoded as modules to
states. A stack is used to store the ASP-states which support the reasoning activities. The translation
will be used for backtracking. Any change in the process relies on the PiLLoW library (which
ASP-state of an ASP module m, requires that all supports HTTP protocol and basic XML ma-
the ASP ancestor modules of m change their own nipulation) and the sophisticated XML library
ASP-states. This can be done with the help of the provided by CIAO Prolog (which allows XML
tree-module structure. parsing, URIs management, and even XPath
Partial Grounder: In order for the Smodels queries). The translation process is performed
system to ground the ASP-modules, all quali- in two steps. During the first step, the RuleML

373
A Logic Programming

document is parsed and converted into a Prolog ontologies are mapped to Prolog (specifically, the
XML representation (as a compound Prolog term). Java-Internet Prolog engine) for interrogation and
In the second phase, the Prolog XML representa- reasoning. This project has a greater emphasis on
tion is parsed and translated into logical rules and integration between a fixed-rule language (the
collected into a module. fragment of RuleML captured by OWLRuleML)
The import component of the module is au- and OWL ontologies, providing an orthogonal
tomatically derived by retrieving those atoms contribution to the integration problem.
used in the program and linking (through URIs) A remarkable direction has been explored in
to external components (e.g., used in the rel ele- the context of the DLV project; dlvhex (Eiter,
ments). By default, the export list will contain all Ianni, Schindlauer, & Tompits, 2006) is an exten-
the rel that appear as heads of rules/facts in the sion of DLV:
knowledge base.
Another specific issue related to RuleML is the • It allows answer set programs to invoke
inclusion of ECA rules. They can be effectively external source of computation, that is,
included in the proposed framework, as discussed truth value of selected predicates can be
by Pontelli, Son, and Baral (2006). achieved through external deduction sys-
tems, for example, accessing RDF state-
ments; and
Related Work • It allows answer set programs to contain
higher-order atoms.
The importance of developing languages and
frameworks to integrate different aspects of se- This language has been motivated by similar
mantic Web reasoning has been highlighted in the reasons as those in this chapter, to support semantic
literature. Most of the existing focus has been on Web reasoning (Eiter et al., 2006; Eiter, Lukasie-
integrating rule-based reasoning with ontology wicz, Schindlauer, & Tompits, 2004), with par-
reasoning. Two relevant contributions in this field ticular emphasis on integration of ontologies.
are represented by the work of Golbreich (2004) Relatively few proposals have appeared in the
and Laera, Tamma, Bench-Capon, and Semeraro literature dealing with the high-level integration
(2004). Golbreich (2004) describes a combination of different forms of logic programming reason-
of reasoning about ontologies (encoded in OWL) ing, specifically top-down goal-oriented Prolog
with rule-based reasoning (with rules encoded in and bottom-up answer set semantics. ASP-Prolog
SWRL and processed by the Jess system). Differ- (Elkhatib et al., 2006), on which the work described
ent from our framework, the system is not based in this chapter builds, is a system that provides
on logic programming (relying on Java’s Jess Prolog programs with the ability to seamlessly
platform) and limited to SWRL’s rule format. access modules processed under answer set seman-
A wider-breadth approach, and closer to the tics. A simplified interface, between the Smodels
spirit of ASP-Prolog, is SweetProlog (Laera, system and XSB Prolog, has been described by
Tamma, Bench-Capon, & Semeraro, 2004). Sweet- Castro, Swift, and Warren (2002).
Prolog offers a platform for integrating rules and Lower level interfaces between answer set
OWL ontologies build using logic programming. systems (DLV and Smodels) and traditional im-
Rules are encoded using OWLRuleML, which perative languages have been developed (Calimeri
represents an OWL ontology for the RuleML & Ianni, 2005; Ricca, 2003; Syrjanen, 1998).
dialect captured by the courteous logic program-
ming scheme (Grosof, 1999). Both rules and OWL

374
A Logic Programming

Conclusion and Future Work References

In this chapter, we addressed the issue of dealing Andorf, C., Dobbs, D., & Honavar, V. (2007).
with heterogeneous Web services in the context Exploring inconsistencies in genome-wide protein
of the Web service composition problem. We function annotations. BMC Bioinformatics.
presented a logic programming framework for Answar, S., Baral, C., & Tari, L. (2005). A lan-
intelligent Web service composition and a pre- guage for modular ASP: Application to ACC
liminary prototype of parts of the framework tournament scheduling. Paper presented at the
within the ASP-Prolog system. The key idea in our ASP Workshops.
framework is to view Web services as actions and
Web service composition as a planning problem. Au, T. C., Kuter, U., & Nau, D. (2005). Web
Under this view, Web service documents (e.g., service composition with volatile information.
WSDL or RuleML documents) are encoded as Paper presented at the International Semantic
logic programs which can be reasoned about and Web Conference (ISWC).
combined using logic programming techniques.
Balduccini, M., Gelfond, M., & Nogueira, M.
The variety of semantics of logic programs sup-
(2006). Answer set based design of knowledge
ports a natural way to deal with heterogeneous Web
systems. Annals of Mathematics and Artificial
service description and ontologies. Furthermore,
Intelligence.
the use of answer set planning in Web service
composition allows us to easily incorporate users’ Baral, C., Dzifcak, J., & Takahashi, H. (2006).
preferences and effectively handle incomplete Macros, macro calls, and use of ensembles in
knowledge. modular answer set programming. Paper pre-
As future work, we propose to demonstrate sented at the International Conference on Logic
the framework on real-world applications, with Programming.
particular focus on applications on description and
Bassiliades, N., Antoniou, G., & Vlahavas, I.
manipulation of bioinformatics Web services. The
(2006). DR-DEVICE, a defeasible logic reasoner
proposed framework is also expected to have a
for the Semantic Web. International Journal on
significant role in facilitating the deployment of
Semantic Web and Information Systems, 2(1),
and reasoning about rule-bases constructed ac-
1–41.
cording to the guidelines of the rule interchange
format (RIF) initiative (Ginsberg, Hirtle, McCabe, Berners-Lee, T., & Fischetti, M. (1999). Weaving
& Patranjan, 2006). Recent developments in RIF the Web: The original design and ultimate destiny
have highlighted the importance of being able to of the World Wide Web by its inventor. Harper.
merge rule sets, and the requirement of supporting
Berners-Lee, T., Hendler, J., & Lassila, O. (2001).
standard ways to characterize rule sets dialects.
The Semantics Web. Scientific American, 284(5),
The proposed framework could easily avail of
34–43.
such standardized identification of dialects to
guide the automated translation of RIF rule sets Boley, H., Grosof, B., Sintek, M., Tabe, S., &
to ASP-Prolog modules. Wagner, G. (2002). RuleML design, version 0.8.
Retrieved May 30, 2008, from www.ruleml.org/
indesign.html
Booth, D., & Liu, C. K. (2005). Web service de-
scription language primer.

375
A Logic Programming

Bueno, F., Cabeza, D., Carro, M., Hermenegildo, Gelfond, M., & Lifschitz, V. (1988). The stable
M., López-García, P., & Puebla, G. (1997). The model semantics for logic programming. Paper
Ciao Prolog system. Unpublished manuscript. presented at the Logic Programming: Proceed-
ings of the Fifth International Conf1 and Symp.
Calimeri, F., & Ianni, G. (2005). External sources
(pp. 1070-1080).
of computation for answer set solvers. Paper
presented at the Logic Programming and Non- Gene Ontology (Ed.). 1.101. (2007). Retrieved
Monotonic Reasoning. May 30, 2008, from http://www.geneontology.
org/ontology/gene_ontology_edit.obo
Castro, L., Swift, T., & Warren, D. (2002). XASP:
Answer set programming with XSB and smodels. Gene Ontology Consortium. (2007). The gene
Retrieved May 30, 2008, from xsb.sourceforge. ontology. Retrieved May 30, 2008, from www.
net/manual2/node149.html geneontology.org
Chan, H., & Grosof, B. (1999). CommonRules. Ginsberg, A., Hirtle, D., McCabe, F., & Patranjan,
Retrieved May 30, 2008, from www.alphaworks. P.-L. (2006). RIF use cases and requirements.
ibm.com/tech/commonrules W3C.
Chinnici, R., Moreau, J.-J., Ryman, A., & Weer- Golbreich, C. (2004). Combining rule and on-
awarana, S. (2007). Web service description tology reasoners for the Semantic Web. Paper
language, version 2.0. presented at the RuleML.
Dean, M. (2001). RuleML experiments with GED- Grosof, B. (1999). A courteous compiler from
COM. Retrieved May 30, 2008, from www.daml. generalized courteous logic programs to ordi-
org/2001/02/gedcom-ruleml nary logic programs. IBM T.J. Watson Research
Center.
Eiter, T., Ianni, G., Schindlauer, R., & Tompits,
H. (2006). Effective integration of declarative Hirtle, D., & Boley, H. (2005). The modularization
rules with external evaluations for Semantic- of RuleML. Retrieved May 30, 2008, from www.
Web reasoning. Paper presented at the European ruleml.org/modularization
Semantic Web Conference.
Hoffmann, J., & Nebel, B. (2001). The FF planning
Eiter, T., Leone, N., Mateis, C., Pfeifer, G., & system: Fast plan generation through heuristic
Scarcello, F. (1998). The KR system dlv: Progress search. Journal of Artificial Intelligence Research,
report, comparisons, and benchmarks. Paper 14, 253–302.
presented at the International Conference on
Horrocks, I., Patel-Schneider, P., Boley, H., Tabet,
Principles of Knowledge Representation and
S., Grosof, B., & Dean, M. (2004). SWRL: A Se-
Reasoning (pp. 406-417).
mantic Web rule language combining OWL and
Eiter, T., Lukasiewicz, T., Schindlauer, R., & RuleML (No. SUBM-SWRL-20040521). W3C.
Tompits, H. (2004). Combining ASP with descrip-
Kautz, H., & Selman, B. (1996). Pushing the
tion logics for the Semantic Web. Paper presented
envelope: Planning, propositional logic, and sto-
at the Principles of Knowledge Representation
chastic search. In Proceedings of the 13th National
and Reasoning
Conference on Artificial Intelligence (AAAI-96)
Elkhatib, O., Pontelli, E., & Son, T. C. (2006). (pp. 1194-1199).
A tool for knowledge base integration and
querying. Paper presented at the AAAI Spring
Symposium.

376
A Logic Programming

Kifer, M., de Bruijn, J., Boley, H., & Fensel, D. McIlraith, S., & Son, T. C. (2002). Adapting
(2005). A realistic architecture for the Semantic GOLOG for composition of Semantic Web
Web. Paper presented at the RuleML. services. In Proceedings of the Eighth Interna-
tional Conference on Principles of Knowledge
Kuter, U., Sirin, E., Nau, D., Parsia, B., & Hendler,
Representation and Reasoning (KR’2002) (pp.
J. (2005). Information gathering during planning
482-493).
for Web service composition. [JWS]. Journal of
Web Semantics, 3(2-3), 183–205. doi:10.1016/j. Nau, D., Cao, Y., Lotem, A., & Muñoz-Avila,
websem.2005.07.001 H. (1999). SHOP: Simple hierarchical ordered
planner. In Proceedings of the 16th International
Laera, L., Tamma, V., Bench-Capon, T., & Se-
Conference on Artificial Intelligence (pp. 968-
meraro, G. (2004). SweetProlog: A system to
973).
integrate ontologies and rules. Paper presented
at the RuleML. Niemelä, I. (1999). Logic programming with stable
model semantics as a constraint programming
Levesque, H., Reiter, R., Lesperance, Y., Lin, F., &
paradigm. Annals of Mathematics and Artificial
Scherl, R. (1997). GOLOG: A logic programming
Intelligence, 25(3,4), 241-273.
language for dynamic domains. The Journal of
Logic Programming, 31(1-3), 59–84. doi:10.1016/ Niemelä, I., & Simons, P. (1997). Smodels: An
S0743-1066(96)00121-5 implementation of the stable model and well-
founded semantics for normal logic programs. In
Lifschitz, V. (1999). Answer set planning. Paper
Proceedings ICLP & LPNMR (pp. 420-429).
presented at the International Conference on Logic
Programming (pp. 23-37). OBO download matrix. (2007). Retrieved May
30, 2008, from http://www.berkeleybop.org/
Lifschitz, V. (2002). Answer set programming and
ontologies/
plan generation. Artificial Intelligence, 138(1-2),
39–54. doi:10.1016/S0004-3702(02)00186-8 Pan, Y., Tu, P. H., Pontelli, E., & Son, T. C. (2004).
Construction of an agent-based framework for
Marek, V., & Truszczynski, M. (1999). Stable
evolutionary biology: A progress report. Paper
models and an alternative logic programming
presented at the Declarative Agent Languages
paradigm. The logic programming paradigm: A
and Technologies (DALT).
25-year perspective (pp. 375-398).
PDDL Technical Committee. (1998). Planning
May, W., Alferes, J., & Amador, R. (2005). Active
domain definition language. Author.
rules in the Semantic Web: Dealing with language
heterogeneity. Paper presented at the RuleML. Pontelli, E., Son, T. C., & Baral, C. (2006). A
framework for composition and inter-operation
McCarthy, J. (1959). Programs with common
of rules in the Semantic Web. Paper presented at
sense. In Proceedings of the Teddington Confer-
the RuleML.
ence on the Mechanization of Thought Processes
(pp. 75-91). Rainer, W. (2005). Web service composition us-
ing answer set programming. Paper presented
McDermott, D., Dou, D., & Qi, P. (2002). PD-
at the Workshop on Planning, Scheduling, and
DAML: An automatic translator between PDDL
Configuration.
and DAML.
Rao, J., & Su, X. (2004). A survey of automated
Web service composition methods. Paper presented
at the SWSWPC.

377
A Logic Programming

Ricca, F. (2003). The DLV Java wrapper. Paper Syrjanen, T. (1998). Lparse 1.0: User’s manual.
presented at the Italian Congress on Computational Helsinki University of Technology.
Logic (CILC).
The Policy RuleML Technical Group. (2004).
Simons, P., Niemelä, N., & Soininen, T. (2002). RuleML-powered policy specification and inter-
Extending and implementing the stable model change. Author.
semantics. Artificial Intelligence, 138(1-2),
Tu, P. H., Son, T. C., & Baral, C. (2006). Reason-
181–234. doi:10.1016/S0004-3702(02)00187-X
ing and planning with sensing actions, incomplete
Son, T., & Pontelli, E. (2007). Planning for bio- information, and static causal laws using logic
chemical pathways: A case study of answer set programming. Theory and Practice of Logic Pro-
planning in large planning problem instances. gramming, 7, 1–74.
Paper presented at the Software Engineering for
van Emden, M., & Kowalski, R. (1976). The
Answer Set Programming.
semantics of predicate logic as a programming
Son, T. C., Baral, C., Tran, N., & McIlraith, S. language. Journal of the ACM, 23(4), 733–742.
(2006a). Domain-dependent knowledge in answer doi:10.1145/321978.321991
set planning. ACM Transactions on Computational
Van Gelder, A., Ross, K., & Schlipf, J. (1991).
Logic, 7(4).
The well-founded semantics for general logic
Son, T. C., Baral, C., Tran, N., & McIlraith, S. programs. Journal of the ACM, 38(3), 620–650.
(2006b). Domain-dependent knowledge in answer doi:10.1145/116825.116838
set planning. ACM Transactions on Computational
Wu, D., Parsia, B., Sirin, E., Hendler, J., & Nau,
Logic, 7(4), 613–657.
D. (2003). Automating DAML-S Web services
Son, T. C., & Pontelli, E. (2006). Planning with composition using SHOP2. Paper presented at
preferences using logic programming. Theory the The SemanticWeb – ISWC.
and Practice of Logic Programming, 6, 559–607.
doi:10.1017/S1471068406002717
Son, T. C., Tu, P. H., Gelfond, M., & Morales, R. Endnotes
(2005). Conformant planning for domains with
1
constraints: A new approach. In Proceedings of the For simplicity, we omit the binding and
the Twentieth National Conference on Artificial service elements and the XML namespace
Intelligence (pp. 1211-1216). definitions.
2
www.ruleml.org
Spencer, B. (2002). The design of j-DREW, a de- 3
We use the notation θσ to denote the com-
ductive reasoning engine for the Semantic Web. position of the two substitutions; we also
Uni. Pol. Madrid. denote with Aθ to denote the application of
Subrahmanian, V. S., & Zaniolo, C. (1995). Re- the substitution θ to the entity A.
lating stable models and AI planning domains. In
Proceedings of the International Conference on
Logic Programming (pp. 233-247).

This work was previously published in Managing Web Services Quality: Measuring Outcomes and Effectiveness, edited by K.
Khan, pp. 193-221, copyright 2009 by Information Science Reference (an imprint of IGI Global).

378
379

Chapter 2.7
The Effectiveness of Scaffolding
in a Web-Based, Adaptive
Learning System
Mei-Yu Chang
National Hsinchu University of Education, Taiwan

Wernhuar Tarng
National Hsinchu University of Education, Taiwan

Fu-Yu Shin
Chien-Kuo Elementary School, Taiwan

ABSTRACT had studied using the web-based, adaptive learn-


ing system. Specifically, scaffolds in the form of
This study combined ideas from learning hier- face-to-face conversations greatly enhanced the
archy and scaffolding theory to design a web- learning of high-achievement students. However,
based, adaptive learning system to investigate there were no significant differences between the
the effectiveness of scaffolding for elementary low-achievement students with or without the pro-
school students having different levels of learning vision of scaffolds. It was also discovered that the
achievement. The topic chosen for learning was web-based, adaptive learning system could help
the Three States of Water. A quasi-experiment students develop their learning responsibility.
was conducted. In this experiment, students were
divided into three groups: control group (without
scaffolds), experimental group A (scaffolds pro- INTRODUCTION
viding by on-line conversation) and experimental
group B (scaffolds providing by face-to-face The purpose of science education is to provide
conversation). The experimental results showed students with scientific knowledge, concepts, at-
significant improvement for students after they titudes and methods for application in their daily

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

lives. The role of a teacher is to assist students Two sources of knowledge are suggested
in learning and solving problems. At the begin- by Vygotsky. The first is everyday knowledge,
ning, students are usually interested in learning i.e., gut knowledge, instinctive knowledge and
science, but to some extent, they lose it or become spontaneous knowledge. This type of knowledge
confused, especially when learning abstract con- is influenced by peer interaction, language and
cepts. The Three States of Water is a learning unit experience of the individual who tries to under-
containing abstract and complicated concepts for stand his or her environment. The second is from
elementary students. formal education in the classroom, which is called
Use of the Internet and instructional technol- formal knowledge and it possesses strict logic and
ogy can help teachers and students in many ways. clear definitions. Therefore, learners construct
It is easier for students to understand abstract meaningful knowledge through both their daily
concepts if their learning process is assisted by lives and school experiences.
instructional technology. This study combined However, some scientific concepts are very
with ideas of learning hierarchy and scaffolding abstract and difficult for students to grasp. In ad-
theory to the design of a web-based, adaptive dition, they may often be influenced by different
learning system to improve the quality of web- cultural and social environments. For example,
based learning. the concept of “The sun rises in the east and
The major purpose of this study was to investi- falls in the west” generally exists in textbooks
gate students’ learning via a web-based, adaptive and daily communication, and students have
learning system where scaffolds were provided this misconception before they are educated with
to help the students to study the concept of the formal knowledge. Students may think the experi-
Three States of Water. This study also tried to find ences they have in daily life constitute their full
out whether students’ learning responsibility was knowledge base. Therefore, it is important for
being developed during their learning process. them to know if the knowledge they have received
The concept of scaffolding is based on Vy- is formally right or not.
gotsky’s social constructivist view of learning According to scaffolding theory, teachers
(1978). Vygotsky proposed that there were two should hold continuous and active conversations
major factors, i.e., culture and social context, with students to find out the possible levels of
which influence learning. He claimed that every their potential development (where they are) and
mental function in a child’s development first to control their learning environments (where
came from the social interaction with an adult. they should be) by providing proper support to
This kind of interaction provides a supportive make the concepts they acquire consistent with
environment in which children can extend their scientists’ current definitions.
current knowledge and skills. The supported situ-
ation occurs in what Vygotsky referred to as the Empirical Studies on Related Topics
zone of proximal development. That is the area
between what children can do independently and Osbome and Cosgrove (1983) discovered in their
what they can do with assistance, such as they study that students lacked the support of sub-
get from teachers and other students. The assis- stantial scientific concepts when they explained
tance that other people provide is a scaffold for a changes in the states of water. The finding was
child. Given repeated experiences, children can that they had only superficial knowledge of
internalize the supported situation of the mental the terms and expressions. For example, some
processes, and can engage in them in new contexts students would think that a solid changing into
(Clark & Graves, 2005). a liquid will lose weight or condensation will

380
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

make particles more compacted. Bar and Travis Current Status of Adaptive
(1991) investigated the concepts of liquid and gas Learning
as possessed by Israeli children and they found
that most children had misconceptions about Electronic Learning (or e-Learning), sometimes
evaporation and condensation, suggesting that it provided as online learning or web-based learn-
was more difficult for the children to understand ing, means learning through digital content to
these abstract concepts. reach the goal of learning anytime and anywhere.
The curriculum for science and technology In general, e-Learning includes any means of
in elementary schools in Taiwan includes the delivering teaching materials through networks
learning about substances and energy, natural (Whittington, 2000). Since 1990, the Internet has
environments, ecological conservation and infor- developed rapidly. After the establishment of the
mation technology. The learning unit chosen in World Wide Web (WWW), many educational
this study was focused on the important concepts websites emerged (Berge & Collins, 1995; Ca-
of the Three States of Water and changes such as hoon, 1998; Collis, 1996; Porter, 1997).
solidification, melting, condensation and evapora- Digital content may consist of multimedia as
tion, because it is related directly or indirectly to well as online interactions, allowing learners to
most of the subjects mentioned above. Although play a more active role in the process of learning
most of the phenomena can widely be seen in our (Aivazidis, Lazaridou & Hellden, 2006). Learners’
daily lives, students may still have misconcep- locations are no longer limited to classrooms, and
tions due to their own mis-understanding and they can learn anytime they want. In addition to
mis-reasoning. a variety of teaching materials, e-Learning has
After reviewing related research conducted gradually become the most convenient way for
in this area, the authors found that most children students to access new information and knowl-
do have misconceptions about the states of water. edge.
Only a few of them think that vapor can also Since traditional ways of learning can not meet-
become water at a low temperature. Very few ing the needs of individual learners, an increasing
of them discover that wind blowing can make number of researches in adaptive learning systems
water evaporate more easily. Senior students in have been conducted in recent years. An adaptive
elementary schools know much about evapora- learning system can provide suitable contents for
tion, but they know little about condensation as different learners according to their backgrounds,
a formal scientific concept. Many children think prior knowledge, individual demands and learning
condensation is solidification. statuses (Papanikolaou et al. 2002).
In general, condensation is a more difficult con- Atkinson (1976) suggested that an efficient
cept than evaporation for students to understand. instructional strategy must be adaptive. Therefore,
Most students do not know the white smoke above an adaptive learning system must change instruc-
boiling water is a gas or vapor; they think of it as tional strategy according to learners’ situations
smoke. Again, some students believe the weight during a learning process under teacher’s control
of water changes after it becomes ice. This study to enhance learning and achieve the expected
therefore designed diagnostic tools and adaptive instructional objectives. Besides, the system can
learning mechanisms to correct the students’ collect some features of learners and store them
misconceptions that often occur. in certain learning modules, which can be used
to provide suitable contents for different learners
(Brusilovsky,1996).

381
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

Web-based learning has become more and However, there are still some disadvantages,
more popular today. A web-based learning sys- such as complicated definitions for the rules of
tem is accessible by a large number of students learning activity and huge frameworks for learn-
at the same time according to individual neces- ing contents, making it difficult to manage, reuse
sity and learning pace, and it is not limited by and combine these with traditional educational
time or space. Therefore, it is suitable for the theories. E-Learning does need a good system
development of adaptive learning environments to support online services, such as course de-
(Chang, 2005). According to Lin’s study (1998), sign, learning resource delivery, integration of
an adaptive learning system can achieve the goal courses and teaching methods (Suthers, Johnson
of teaching students according to their aptitudes & Tillinghast, 2002), etc. For this reason, the
in a normal class grouping environment, and it Graduate School of Online Learning Applica-
makes learning more active and efficient. tions, National Chiao Tung University, Hsinchu,
A personalized web-based learning system was Taiwan developed their Object Oriented Learning
proposed by Chang et al. (2006) based on item Activity System (OOLAS) based on the standards
response theory (IRT). The system was aimed of SCORM, Learning Design Specification,
at providing a suitable learning environment by Knowledge Tree and Multibook. In the SCORM
considering the learning portfolios of content dif- standards, most metadata models were used to
ficulty, learner’s ability and conceptual coherence. define learning materials (Su, Tseng, Lin & Lu,
Lin and Kuo (2005) developed a virtual learning 2005). Chang et al. proposed in their study (2004)
environment based on the theory of learning ob- a metadata model to define online assessment.
jects and constructivism. The learning contents OOLAS is composed of an authoring system and
and statuses of learners on the system can be in- a learning system. In addition to a simple process
dividualized to achieve the goal of adaptive learn- for editing materials, it presents the contents of
ing. Also, learners can cooperate in the inhabited curricula easily and allows teachers to plan related
virtual world to increase learning effectiveness programs to assist students with learning effec-
during their learning processes. tively. OOLAS also allows teachers to design the
Because e-Learning systems are usually devel- related tests of concepts.
oped in accordance with learners’ requirement in In this study, the authors used OOLAS to
terms of platforms, materials, presentation styles develop a web-based, adaptive learning system
and virtual communities, it often occurs that some that was integrated with scaffolding to improve
systems are only suitable for certain users. If a the learning efficiency of students. They also
system is transferred to another environment, for developed teaching materials for the concepts
some other people, tremendous manpower and of the Three States of Water. The system mod-
costs must be spent in making changes (Dodds, ules were: diagnostic tools, a test-item database,
2001a; 2001b). To solve this problem, international adaptive learning modules and learning contents
standard organizations have specified e-Learning for the scientific concepts in the Three States of
standards. E-Learning systems such as SCORM, Water (Figure 1).
AICC, IEEE, IMS, IEEELOM, ARIADNE, and The diagnostic tools were designed using
Microsoft promulgate these standards (Hodgins two-tiered test items to identify the prior mis-
& Conner, 2000). Among them, SCORM has now conceptions of students. After detecting their
become an important standard for e-Learning misconceptions, an adaptive learning module
content (Chang, Hsu, Smith & Wang, 2004). was initiated by a rule-based inference engine
Learning contents that conform to SCORM to generate individualized learning content for
standards have reusable and sharable features. students according to the particular misconcep-
tions of the particular student.

382
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

RESEARCH DESIGN Based on the learning hierarchy, the authors used


the diagnostic tools of the web-based, adaptive
This study was conducted using a quasi-experi- learning system to find out the misconceptions
mental design. A total of 200 students belonging of students. The authors developed the adaptive
to two elementary schools in Taipei County were learning modules by using a rule-based inference
chosen for the experiment. The authors divided engine to provide remedial instructions for stu-
the students into a control group and experimental dents according to a categorization of the types
groups to investigate the learning of students with of their misconceptions. After that, the authors
or without scaffolds. The authors also measured conducted a quasi-experiment to find out the
the changes in their learning responsibility after influence of scaffolding on students when they
the e-Learning. For the control group, the students were learning using the system.
studied using the web-based, adaptive learning
system by themselves without the provision of Research Samples
any scaffolds. For the experimental groups, the
students could ask teachers for help if they had The samples in this study were 5th and 6th grade
questions during the learning processes and after students, randomly selected from two elementary
the diagnostic tests. There were two experimental schools in Taipei County. Because these students
groups, provided with scaffolds in different ways. had already taken computer courses during their
For experimental group A, the scaffolds were 3rd grade year, by the time of this experiment,
provided in the form of online conversation be- they had two to three years of experience in us-
tween the teachers and students. For experimental ing computers and the Internet. This reduced the
group B, the scaffolds were provided in the form influence of computer skills on the effectiveness
of face-to-face conversations between the teachers of e-Learning. As stated, the students were di-
and students. For both media of communication, vided into the control group (without scaffolds)
teachers provided scaffolds by discussing with and the experimental groups (with scaffolds).
the students to find out their problems and correct The experimental groups consisted of group A
their misconceptions. (providing scaffolds via online conversation) and
group B (providing scaffolds via face-to-face
Research Procedure conversation).
According to a pretest, students with scores
This study selected the Three States of Water as above the first 50% were classified as high achieve-
the topic of learning. Therefore, the authors first ment students while the remaining 50% were
analyzed the related concepts covered by text- classified as low achievement students. In the
books of elementary schools, and then developed control group, there were 33 high-achievement
a learning hierarchy according to these concepts. students and 33 low-achievement students, making

Figure 1. The operation of the web-based, adaptive learning system


Misconceptions
Diagnostic Tools Adaptive Learning
Test-Item Learning Content
Adaptive Database Modules Database
Learning
Contents

383
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

Table 1. The number of students in each group

Content Groups Scaffolds Achievement Number of Students


High 33
Control Group None
Low 33

Three States of Experimental Online High 33


Water Group A Conversation Low 32
Experimental Face-to-Face High 35
Group B Conversation Low 34
Total Number of Students 200

a total of 66 students. In the experimental group Pretest and Posttest


A, there were 33 high-achievement students and
32 low-achievement students, making a total Limited by the sampling method and number
of 65 students. In the experimental group B, of samples, we adopted a non-equivalent group
there were 35 high-achievement students and 34 pretest-posttest design in our experiment, and
low-achievement students, making a total of 69 adjusted the difference in students’ background
students (as listed in Table 1). using an analysis of covariance (ANCOVA) to
investigate the influence of scaffolding on the
Diagnostic Tools students with different levels of achievement.
The questions in the pretest and posttest were
The questions to diagnose misconceptions about designed based on the learning contents and re-
the Three States of Water in this study were modi- lated concepts of the Three States of Water. The
fied from the paper-based test questions designed reliability of the achievement test was verified
by Chang (2003), who analyzed the misconcep- by an internal-consistency reliability test with
tions that students in elementary schools often Cronbach α=0.71.
have. Since two-tier test items have been used After removing the inappropriate questions,
widely and efficiently in discovering miscon- there were 113 multiple-choice questions left.
ceptions and conceptual changes (Tsai & Chou, Among them, 37 questions were selected to
2002), the authors thus used them to develop the form the pretest and posttest, and the remaining
diagnostic tools based on the empirical study of questions were used as diagnostic test items. The
Lin (1995). The authors consulted three science scores of the pretest were used to measure the
teachers and two experts in science education students’ levels of achievement. The questions
about the validity of the test items and diagnostic in the posttest were similar to that of the pretest,
tools. Finally, the two-tier test items were uploaded enabling the authors to compare the experimental
and stored in the test-item bank by using authoring and control group results. An example of a ques-
tools of the system. An example of the two-tiered tion follows:
question follows: Which of the following is the correct descrip-
There are two identical wet towels. One is tion for the speed of evaporation? (A) The more
spread and the other is folded. Which one do you humid the environment is, the faster the evapo-
think will dry first? ration is. (B) The hotter the environment is, the
Answer: (A) Folded one. (B) Both at the faster the evaporation is. (C) The more airtight the
same time. (C) Spread one. (D) I don’t know. environment is, the faster the evaporation is.
Explain your reason:.

384
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

Measurement of Learning the total internal consistency was calculated as


Responsibility Cronbach α=0.89.

In this study, the authors used a questionnaire


to measure the development of self-learning INSTRUCTIONAL DESIGN
responsibility after the web-based learning. The
questionnaire was developed by the researchers The principle of instructional design in this study
and verified by three elementary school teachers was based on Gagne’s learning hierarchy (1985).
and two experts in the research area of science The major concepts of the Three States of Water
education. It contained 18 questions. Each ques- were analyzed to form a hierarchical diagram.
tion had five scales in its answer: 5 points for The learning hierarchy was constructed according
very likely, 4 points for likely, 3 points for no to competence indicators for the curriculum of
opinion, 2 points for not likely, and 1 point for science and technology in elementary schools in
not likely at all (as shown in Appendix I). This Taiwan. For example, the concepts of the Three
is an indicator for the degree of development of States of Water are based on solidification, melt-
the students’ learning responsibility. All ques- ing, evaporation and condensation. Deeper, the
tions in this measurement were positive-response concept of solidification contains the lower-level
designed. Thus, the more points a student gained, concepts of water changing from liquid to solid,
the more responsible and active he or she was in increasing volume, and mass conservation. Be-
learning. The reliability of this measurement was cause condensation, solidification, melting and
verified by testing on 32 6th grade students and evaporation are the major concepts for the Three

Figure 2. Hierarchy of scientific concepts in the three states of water


1

2 3 4 5

6 7 8 9 10 11 12 13

14 15 16 17 18 19 20 21 22 23 24 25

26 27 28 29 30 31 32 33

34 35 36

1. three states of water 12. evaporation speed 21. temperature 30. tasteless
2. solidification 13. gas to liquid 22. convection 31. without fixed shape
3. melting 14. liquid water 23. relative temperature 32. evaporation absorbing
4. evaporation 15. solid water 24. Gaseous water is heat
5. condensation 16. decreasing invisible. 33. steam in the air
6. liquid to solid temperature 25. decreasing 34. evaporation at room
7. mass conservation 17. freezing point temperature temperature
8. increasing volume 18. liquid water 26. achromatic 35. Heating speeds up
9. mass conservation 19. solid water 27. tasteless evaporation.
10. solid to liquid 20. increasing 28. without fixed shape 36. Heating causes water
11. liquid to gas temperature 29. achromatic to evaporate.

385
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

Figure 3. The online experiments incorporated in the learning content

States of Water in elementary science courses, a did a diagnostic test to see if they had completely
number of teaching activities were designed to understood the scientific concepts. The design of
introduce these concepts. diagnostic tests followed the sequence of solidi-
The authors encoded all concepts in the learn- fication, melting, evaporation and condensation.
ing hierarchy by number to simplify the diagnostic After students had completed a learning activity,
sequences. A student with a misconception at a they were directed to the diagnostic tests to see
high level implies that he or she may have miscon- if their understanding about the related concepts
ceptions at lower levels and thus requires further was correct. If the concepts of students were valid
diagnostic processes. Generally, the higher the according to the diagnostic results, they could keep
hierarchical level is, the smaller the number is. For on learning the remaining concepts. Otherwise,
example, the highest-level concept is “three states the students had to proceed with remedial instruc-
of water”, which was encoded as 1. The second- tion to correct their misconceptions.
level concepts are “solidification”, “melting”, For example, if students passed the tests on
“evaporation”, “condensation”, and which were the concept of solidification, they could move to
encoded as 2, 3, 4, and 5. The concept of “solidi- the next stage to start the learning activities on
fication” (2) contains the lower-level concepts of the concept of condensation. Otherwise, they
water changing from “liquid to solid” (6), “mass were directed to the remedial instructions for
conservation” (7), and “increasing volume” (8) the correction of previous misconceptions, fol-
as shown in Figure 2. lowed by formative evaluations to guarantee the
Because experiments and observation are conceptual changes had been internalized. When
very important for the establishment of scientific all misconceptions were corrected after remedial
concepts, the authors designed several online learning activities, they could start learning the
experiments to simulate the phenomena of the contents for the next stage (Figure 4).
Three States of Water to help students develop
their scientific concepts (Figure 3). The experi-
ments were designed using Flash to provide an RESULTS AND DISCUSSIONS
interactive user interface, animations for the
simulation of the phenomena and concepts in the The major purpose of this experiment was to
Three States of Water. investigate students’ learning enhancement on
In this study, there were four lessons for all the web-based, adaptive learning system and the
students to complete. After each one, the students

386
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

Figure 4. Learning diagnosis design of the system

Correct Diagnosis of
Concepts other Concepts

Correct Concepts
Learning Diagnosis
Teaching Activity

Next Stage
Incorrect
Concepts

Remedial Formative
Instructions Evaluations

development of learning responsibility after the tion than that of condensation, which agreed with
e-Learning. In addition, the learning statuses of the results of the studies by Lin (1995), Lai (1994)
students, such as misconceptions and conceptual and Chang (1997).
changes, were also studied by analyzing learn- Because the teaching materials were designed
ing records. Students’ scores in the pretest and based on Gagne’s notion of learning hierarchy
posttest were analyzed using statistical software (1985), it was effective for learning the concepts
SPSS to calculate the mean and standard devia- in the Three States of Water. Also, the web-based,
tion for each group. adaptive learning system was useful for teachers
in reducing the workload, while increasing the
Learning Effectiveness learning effectiveness of the students. Accord-
ing to the diagnostic results, students could go
Based on the analysis of the T-test, all students immediately through remedial instructions to
performed very well on the web-based, adaptive correct their misconceptions. It could also solve
learning system, no matter whether they were the problem of insufficient teachers to provide
high-achievement (t=7.27, p<0.01) or low-achieve- feedback for individual students. Besides, through
ment (t=13.63, p<0.01) students. This revealed that the records of learning processes, it was easy to
the system was efficient for students in learning know the status of each student and provide ap-
the concepts in the Three States of Water. propriate support.
Based on the analysis of distribution in stu-
dents’ misconceptions, over 70% of the students High-Achievement Students
had valid concepts about melting before learning
activities. However, there were also 65% of the The effective samples of the high-achievement
students with misconceptions about condensation students contained 29 students in the control
so that they spent more time in the related remedial group, 31 students in the experimental group A,
instruction. According to the results of the T-test, and 32 students in the experimental group B. Us-
we found the students made significant progress ing ANCOVA, the homogeneity of the regression
in the areas of solidification (t=2.37, p<0.05), slope was tested ( f=0.32, p>0.05), indicating that
evaporation (t=2.98, p<0.05), and condensation the homogeneity assumption was met. After the
(t=2.57, p<0.05). Besides, the results showed that ANCOVA analysis, the results showed that there
more students understood the concept of evapora- were significant differences among these three

387
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

Table 2. ANCOVA summary data of the high-achievement students


Source S.S. DF MS F Sig.

Pretest 76.73 1 76.73 12.04** 0.01

Deviation 105.14 2 52.57 8.25** 0.01

Error 560.68 88 6.37

Sum 75114.00 92

**p<0.01

Table 3. Post-hoc analysis on the posttest of high-achievement students


Comparison Mean Difference Standard Error

Experimental Group B vs. Experimental Group A 2.59 0.65

Experimental Group B vs. Control Group 1.46 0.64

Experimental Group A vs. Control Group -1.12 0.67

groups. Table 2 shows the provision of scaffolds conversation, the teachers could understand the
had a significant impact on the learning effec- students’ misconceptions and then decide what
tiveness of high-achievement students ( f=12.04, kind of support to provide. That is why the learning
p<0.01). It also reveals that the high-achievement effectiveness was significantly increased.
students had different learning effectiveness in
the three groups with deviation in the provision Low-Achievement Students
of scaffolds ( f=8.25, p<0.01).
The post-hoc analysis revealed that the high- The effective samples of the low-achievement
achievement students in the experimental group students consisted of 28 students in the control
B had better learning effectiveness than those in group, 31 students in the experimental group A,
the control group and experimental group A. In and 31 students in the experimental group B. Us-
addition, Table 3 shows no significant difference ing ANCOVA, the homogeneity of the regression
between the high-achievement students in the slope was tested ( f=0.27, p>0.05), indicating that
control group and experimental group A. the homogeneity assumption was met. Through
According to the above results, it can be the analysis of ANCOVA, the authors were able to
discovered that providing the scaffold of face-to- investigate the influences of different approaches
face interactions between teachers and students on low-achievement students. The deviation
was better than that of online interactions for the F value caused by the independent variables
high-achievement students (scaffold-provision difference) was 0.51 (p>0.05),
For the students in experimental group A, the which did not reach the degree of significance.
support of teachers did not assist them too much The results in Table 4 show that the learning of
because the scaffold provided was in the form of scientific concepts by low-achievement students
online conversation. It was not so effective because was not affected by the provision of scaffolds.
the teachers might not understand the student s’ From the above results, it can be implied that
difficulties during the processes of web-based the learning effect of low-achievement students
learning, especially when the students had some was not influenced by the provision of scaffolds
misconceptions. However, in the face-to-face in e-Learning processes, meaning that no mat-

388
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

ter whether scaffolds were provided or not and E-Learning is different from traditional ways
regardless of the forms of scaffolds, there was no of learning. Given well-organized teaching mate-
significant difference in the learning effectiveness rials and learning activities, students were more
of low-achievement students. aware of their learning conditions and therefore
According to the above results, it is better for could do better by themselves to obtain knowledge.
teachers to provide the high-achievement students Therefore, the e-Learning process was helpful for
with face-to-face conversations instead of online the development of students’ learning responsibil-
discussions. However, providing either scaffold ity. They could follow the directions to study by
could not help low-achievement students improve themselves and make their schedules to complete
further because they all made a remarkable prog- the learning activities. For both types of students,
ress in the posttest, meaning that the web-based, the difference was that the low-achievement stu-
adaptive learning system was very effective for dents preferred to find the answers by themselves,
low-achievement students and scaffolding was while the high-achievement students wanted to
not a factor in influencing their learning. ask teachers for help when they had problems. In
both ways, they performed well on the web-based,
Development of Learning adaptive learning system.
Responsibility After the experiment, the authors conducted
some semi-structural interviews with a few
Another objective of this study was to find out if students to understand their ideas about the
the web-based, adaptive learning system could web-based, adaptive learning system and its
help students develop their responsibility for self- performance. The interview results are sum-
learning. There were 18 questions to measure the marized as:
development of learning responsibility, and the
score for each question was from 1 to 5 according I like the way of web-based learning, because I
to the agreement with the answer. A higher score can also do it by myself at home.
meant more responsibility was established by the
student, and vice versa. I can stop somewhere and repeat studying for
The results showed that both levels of stu- several times until I can fully understand the
dents (t=2.82, p<0.01), no matter whether they concepts. However, I cannot do it that way in
were high-achievement (t=2.93, p<0.01) or low- classrooms because I am afraid to ask teachers
achievement (t=2.71, p<0.01), reached statistical to slow down their teaching.
significance in the comparison of the pretest
and posttest. More precisely, the students raised Web-based learning is more efficient in raising my
their sense of self-responsibility as a result of learning motivation than traditional classroom
e-Learning. learning does.

Table 4. ANCOVA summary data of the low-achievement students


Source S.S. DF MS F Sig.

Pretest 299.26 1 299.26 24.84 0.01

Deviation 12.33 2 6.17 0.51 0.60

Error 1036.33 86 12.05

Sum 65606.00 90

389
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

The web-based, adaptive learning system can difference for the low-achievement students with
point out my misconceptions and help me correct or without the provision of scaffolds, since they
them. I can operate the system easily, so I am not all performed well on the web-based, adaptive
afraid of it. learning system. For both types of students, their
learning responsibility was significantly improved
I can learn the concepts of the three states of water in the web-based learning environment.
efficiently on the system. The design of the web-based, adaptive learn-
ing system in this study was based on the notion
Learning on the system is very interesting. I wish of scaffolding theory and Gagne’s learning hier-
teachers in other classes (could) also use the same archy. Because the system provided diagnostic
way in their teaching. mechanisms in the entire learning processes,
it was assumed that the students passing the
The authors also asked the teachers who were diagnostic tests would not have any misconcep-
involved in this experiment for their opinions, tions. However, the results in posttest showed
and they were satisfied with the system because that some students still had misconceptions
it provided a convenient interface for teachers to about the material. Therefore, teachers should
design the teaching activities and edit the learning not rely on the diagnostic tools completely. They
contents. Besides, the system could help teachers should use some other assessment tools to see if
identify the misconceptions of students and direct the students’ concepts were correct or not, and
them to suitable learning content to enhance their then provide appropriate feedback to correct the
learning effectiveness. Therefore, it could save misconceptions.
teachers’ time and effort in providing remedial If the use of teaching materials and diagnostic
instructions for individual students. tools on the system can be easily transferred to
other systems, it will increase the sharing among
schools to help more students learn effectively.
CONCLUSION Moreover, advances in computer and network
technologies make it easier for students to learn
In this study, a web-based, adaptive learning sys- through networks, and this reduces the workload
tem was designed to investigate the effectiveness of teachers in helping students solve their prob-
of scaffolding on elementary school students with lems. This study was quantitatively oriented so
different levels of learning achievement. This that it might not be enough to discover the details
study was conducted using a quasi-experimental of students’ learning statuses and the process of
design and the contents for learning were the conceptual changes. Future studies can be focused
scientific concepts of the Three States of Water. on qualitative methods to understand students’
According to the experimental results, the find- behavior while learning.
ings are:
The web-based, adaptive learning system is
effective in helping students acquire the concepts ACKNOWLEDGMENT
of the Three States of Water. For the high-achieve-
ment students, providing scaffolds in the form The authors would like to thank for the financial
of face-to-face conversation could significantly support of the National Science Council of the
enhance the learning effectiveness, which is better Republic of China under the contract number
than using online interactions or without provision NSC 92-2524-S-134-006.
of scaffolds. However, there was no significant

390
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

REFERENCES sessment authoring in e-Learning. Journal of


Computer Assisted Learning, 20, 305-316.

Aivazidis, C., Lazaridou, M. & Hellden, G. F. Clark, K. F. & Graves, M. F. (2005). Scaffolding
(2006). A comparison between a traditional and students’ comprehension of text. The Reading
an online environmental educational program. Teacher, 58(6), 570-580.
Journal of Environmental Education, 37, 45-54.
Chen, C. M, Liu, C. Y. & Chang, M. H. (2006).
Atkinson, R. C. (1976). Adaptive instructional Personalized curriculum sequencing utilizing
systems: some attempts to optimize the learning modified item response theory for web-based
process. In D. Klahr (Ed.), Cognition and instruc- instruction. Expert Systems with Applications,
tion. New York: Wiley. 30(2), 378-396.

Bar, V. & Travis, A. S. (1991). Children’s views Collis, B. (1996). Tele-learning in digital world:
concerning phase changes. Journal of Research The future of distance learning. New York: In-
in Science Education, 18, 363-382. ternational Thomson Computer Press.

Berge, Z. & Collins, M. P. (Eds.) (1995). Com- Dodds, P. (2001a). Sharable Content Object
puter-mediated communications and the online Reference Model (SCORM) Version 1.2 Techni-
classroom, Vol. III: Distance Learning. Cresskill, cal Report. The SCORM content aggregation
New Jersey: Hampton Press. Model, Advanced Distributed Learning Ini-
tiative, URL: http://www.adlnet.org/index.
Brusilovsky, P. (1996). Methods and techniques
cfm?fuseaction=SCOR-Down [last accessed
of adaptive hypermedia. User Modeling and User
20 November 2003]
Adapted Interaction, 6(2-3), 87-129.
Dodds, P. (2001b). Sharable Content Ob-
Cahoon, B. (1998). New directions for adult
ject Reference Model (SCORM) Version 1.2
and continuing education, Summer, No. 78.
Technical Report. The SCORM Run-Time
Adult Learning and the Internet. San Francisco:
Environment, Advanced Distributed Learning
Jossey-Bass.
Initiative, URL: http://www.adlnet.org/index.
Chang, C. (1997). A study on the concepts of cfm?fuseaction=SCOR-Down [last accessed
evaporation, condensation and boiling by senior 20 November 2003]
students in elementary schools. Chinese Journal
Gagne, R. M. (1985). The conditions of learn-
of Science Education, 5(3), 321-346.
ing and theory of instruction. New York: Holt,
Chang, K. (2003). A research on the concepts of Rinehart, & Winston.
states of water. Master thesis, National Pingtung
Hodgins, W. & Conner, M. (2000). Everything you
Teachers College, Pingtung, Taiwan.
ever wanted to know about learning standards
Chang, S. L.(2005). Applying characteristics de- but were afraid to ask. LiNE Zine, URL:
scribed system to diagnosis of e-learning. Master
http://www.adlnet.org/index.cfm [last accessed
thesis, Department of Information and Computer
on 20 November 2003],
Engineering, Chung Yuan Christian University,
Chung-Li, Taiwan. http://www.adlnet.org/index.cfm [last accessed
on 20 November 2003], and http://www.linezine.
Chang, W., Hsu, H., Smith, T. K. & Wang, C.
com/2.1/features/wheyewtkls.htm [last ac-
(2004). Enhancing SCORM metadata for as-
cessed on 9 Feb. 2004].

391
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

Lai, H. C. (1994). Children’s concepts of water dimension of a web-based learning environment.


cycle and the forms of substances. Master thesis, Computers and Education, 39, 333-360.
National Taiwan Normal University, Taipei,
Porter, L. R. (1997). Creating the virtual class-
Taiwan.
room: Distance learning with the Internet. New
Lin, C. H. (1998). The study of individual differ- York: John Wiley.
ences and automated learning strategies in hyper-
Su, J. M., Tseng, S. S., Lin, H. Y. & Lu, C. Y. (2005,
text learning environment. Journal of National
May 6-7). The development of an object-oriented
Hsinchu Teachers College, 11, 1-14.
learning activities system. Taiwan e-Learning
Lin, C. S. & Kuo, M. S. (2005, July 5-8). Adaptive Forum 2005, Taipei, Taiwan.
networked learning environments using learning
Suthers, D., Johnson, S. & Tillinghast, B. (2002).
objects, learner profiles and inhabited virtual
Learning object meta-data for a database of el-
learning worlds. The 5th IEEE International
ementary and secondary school resources. Inter-
Conference on Advanced Learning Technologies,
active Learning Environments. 9, 273-289.
Kaohsiung, Taiwan.
Tsai, C. & Chou, C. (2002). Diagnosing students’
Lin, S. (1995). A study on students’ concepts of
alternative conceptions in science through a
evaporation and condensation in elementary
networked two-tier test system. International
schools. Technical report, National Science Coun-
Journal of Computer Assisted Learning, 18 (2),
cil, Taipei, Taiwan.
157-165.
Osbome, R. J. & Cosgrove, M. M. (1983). Chil-
Vygotsky, L. S. (1978). Mind in society: The
dren's conception of changes of state of water.
development of higher psychological processes.
Journal of Research in Science Teaching, 27,
Cambridge, Massachusetts: Harvard University
825-838.
Press.
Papanikolaou, K., Grigoriadou, M., Magoulas,
Whittington, D. (2000). Evaluating three years’
G. D. & Kornilakis, H. (2002). Towards new
use of virtual university. Quality Assurance in
forms of knowledge communication: the adaptive
Education, 8(1), 48-52.

392
The Effectiveness of Scaffolding in a Web-Based, Adaptive Learning System

Appendix: A questionnaire to measure the development of


learning responsibility

1. Very likely
2. Likely

3. No opinion

4. Not likely
5. Not likely at all
Dear students,
The following is a questionnaire to measure the establishment of learning respon-
sibility after web-based, adaptive learning. It is not a test and there are no standard
answers. Please read the following questions carefully, and put a check mark inside
the space to represent your opinion regarding each question.

Grade: □5th Grade □6th Grade Sex: □Boy □Girl

1. In the process of learning, I often finish all learning activities by myself.


In the process of learning, I know clearly whether or not I have understood all
2.
learning contents taught in class.
In the process of learning, I would look for the answers actively to prove that my
3.
ideas are correct.
In the process of learning, I would join the discussion among my classmates
4.
actively.
5. When discussing with my classmates, I could judge whose ideas are correct.
I consider that learning is my own responsibility no matter if the result is good
6.
or bad.
7. In the process of learning, I usually doubt if my own ideas are correct or not.
If I have difficulty in learning, I will ask for help actively when my teacher does
8.
not assist me.
In the process of learning, I would check on myself to see if my ideas are cor-
9.
rect.
10. I know clearly how to learn in class.
If I have difficulty in learning, I will ask for help actively when my classmates
11.
do not assist me.
In the process of learning, I would often request myself to learn more conscien-
12.
tiously.
During the in-class discussion, I often doubt if the ideas of teachers or classmates
13.
are correct or not.
When I have doubts about the ideas of teachers or classmates, I would search for
14.
the answers to prove if they are correct or not.
I would rather find the answers to my questions than have my teachers or class-
15.
mates help me.
After school, I do a self test to see if I have learned whatever was taught in
16.
class.
I think I have to learn actively and positively in order to understand the materials
17.
taught in class.
18. I make my own learning schedule and follow it accurately.

This work was previously published in International Journal of Web-Based Learning and Teaching Technologies, Vol. 14, Is-
sue 1, edited by E.M.W Ng, N. Karacapilidis, and M.S. Raisinghani, pp. 1-15, copyright 2009 by IGI Publishing (an imprint
of IGI Global).

393
394

Chapter 2.8
WebSphere Portal 6.1:
An Agile Development Approach

Thomas Stober
IBM Germany Research and Development, Germany

Uwe Hansmann
IBM Germany Research and Development, Germany

Abstract Reaching the limits of the


classical waterfall
IBM’s Portal technology continues to evolve approach
as a powerful infrastructure for integrating the
IT landscape, by presenting it as a consolidated It ain’t what you don’t know that gets you into
view to the user community. The new capabili- trouble. It’s what you know for sure that just ain’t
ties of WebSphere Portal 6.1 are the outcome of so. Mark Twain
a world-wide development team, which focused
on this release for the past 2 years. During that Developing market-leading Enterprise Por-
time major architectural enhancements have been tal products, like WebSphere Portal requires a
introduced and a significant amount of code was first-class development team. Far more than 300
written. In this article the authors will describe developers and test engineers are working for
how developers and testers have adopted agile different organizational units and collaborate
principles to collaborate across the globe. In in very far apart time zones. There are 8 major
detail, aspects like an iterative approach, test development sites across the world. The product
driven development, budget based prioritization has dependencies on other IBM products, such
and cross-organization teaming will be discussed. as WebSphere Application Server, and is the base
The authors will also cover how “tiger teams” for other products, like Lotus Quickr. Further
interact with customers by making early code dependencies arise from customer requirements
drops available and responding to feedback. and commitments.

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
WebSphere Portal 6.1

IBM WebSphere Portal leads this wave of pieces together in order to assemble a complex
innovation, combining the latest user-centric use case requires a significant integration effort,
functionality with reliable security and manage- before the overall system reaches a satisfying level
ability features to meet the needs of the business. of stability. Development and testing were done
The software incorporates extensive Web 2.0 by separate organizations. While the developers
capabilities, allowing companies to fuel social owned the responsibility for design, coding and
interaction by delivering high-performing, intui- unit testing, the test organization covered func-
tive applications through a rich Web interface. tional and system verification testing.
This new release adopts the latest industry-driven Typically, unforeseen issues, like a design flaw,
standards. It also introduces flexible ways to cre- or a growing number of bugs beyond the expected,
ate and manage Portal sites and content. Many or redirection of resources to other activities, made
more enhancements emphasize increased utility it necessary to rework the plan. Typically the
and flexibility, such as web site management, problems are getting really pressing at the end in
integration of non-Portal pages as well as step the final test phase. At that time, content removal
up authentication. isn’t really an option, as the code is already done,
Up to Portal 6.0, the Portal team used to work although not stable. Delaying the shipment is not
a classical waterfall approach. Product manage- a good option, as customers do rely on the prom-
ment captured the requirements and work items ised delivery dates. And sacrificing the quality
for a particular release project, and prioritized is not acceptable either. And obviously the costs
them. The project management team assembled a of fixing problems increase significantly the later
complex project plan with a break down of distinct the issue is detected. It took a tremendous, costly
task assignments for individual developers and team effort to solve the situation and ensure that
testers. There were milestones, test phases, and a solid product is still being shipped on time.
fixed target dates to achieve the well defined goals. As a result of these experiences, the limits of
In general there was an analysis and design phase such a pre-planned waterfall approach became
in which content, architecture, and project plan obvious: the classical approach it is too inflexible
were established. In addition, there was distinct to react quickly enough to the highly dynamic
development phase executing the plan, followed constraints of a complex product within a large
by the distinct test phase. In major releases the organization (Figure 1).
project plan covered a period of up to 1.5 years.
However the complexity of the technology
and especially the growing complexity of the Moving to Agile
team and time constraints have made it more and Software Development
more difficult to execute the established plan as
scheduled. Future needs and issues are difficult One of the key reasons, why the Portal team has
to predict. Each of these distinct phases turned moved away from a classical waterfall approach
out to be not that distinct and isolated. Instead used for Portal 6.0, is to gain more flexibility
there were dependencies, loops circulating back to and improve the ability to react to changing
earlier phases in order to adjust. Communication constraints. Within a release project, the content
and interfaces between different organizational needs to be decided as late as possible, while
units are a challenge in large distributed teams. It tested and usable pieces of functionality are to
is extremely difficult to make sure that the right be made available as early as feasible.
information is made available to the right set of Another goal is to optimize the flow of human
people. Bringing the independently developed interaction. Intensive collaboration should be

395
WebSphere Portal 6.1

Figure 1. Waterfall approach: it is getting tougher towards the end

fostered across organizational structure - espe- management and lead architects do come up with
cially between test and development teams and a first cut of release content by aggregating the
the number of dependencies between different customer feedback and high-level requirement into
teams should be minimized. rough focus areas. Each of these focus areas is as-
The agile approach, which the Portal team sociated with a budget reflecting the approximate
has introduced for Portal 6.1, features 4 key number of developers supposed to be working on
concepts: that area throughout the upcoming release project.
In Portal the designated developers pursuing the
• Budget based prioritization and content same focus area are grouped into a “Tiger Team”.
definition Each team can span multiple geographical loca-
• Cross-organization teaming Structure tions and organizational units. The worldwide
• Iterative development approach team acknowledges that the developers of each
• Test driven development tiger team are the subject matter experts, and that
the ”center of gravity” is lowered to allow quicker,
better and more optimized decisions directly by
Budget based the tiger team. This distributed approach scales
prioritization and much better than a central release management
content definition: deciding on all detailed use cases of the entire
Requirements and Use cases release and maintaining a complex overall project
plan for all worldwide developers.
It is not the strongest of the species that survives, For example, it is the responsibility of the de-
nor the most intelligent; it is the one that is most velopers in each tiger team to translate the given
adaptable to change. Charles Darwin high-level requirements into specific use cases.
Each tiger team prioritizes the use cases it intends
Which features should be included into a to deliver in a team charter document. The charter
release? Who decides the release content with lists highly prioritized use cases, which the team
which level of granularity? For Portal, the product is committing to deliver under all circumstances.

396
WebSphere Portal 6.1

Other items have the disclaimer to be run-at, and a solid staffing and the required skills. Usually
will be implemented in the order of priority as the market is demanding certain features at a
time permits. Further items are listed for com- certain timeframe and even an agile plan has to
pleteness, but are marked as out-of-scope for the commit some minimum of enhancements early
release project. The final scope and timeline of the up in the cycle.
release project is not yet fully defined when team And change needs to be anticipated. For in-
is being established. To leave sufficient room for stance, after 5 months of Portal 6.1 development,
agility, it is extremely important that each team a change of the underlying WAS version has been
is committing only use cases to a limited amount decided to meet customer requests: While the
of its team capacity. original plan has been to ship with WAS 6.0.2, the
Within the constraints of staffing budget, over- final product moved to WAS 6.1. This modification
all release timeline and given high-level require- affected design documents and implementation
ments, the team can proceed very autonomously significantly, since WebSphere 6.1 introduced an
with the planning and execution of their activities. entirely new user management component and
Often tiger teams start their work independently had different non-functional requirements. The
of a specific release and drive towards early project duration had to be extended. As the Web
deliverables outside of an official product. One 2.0 tiger team did only have little impact by this
example is the tiger team, which developed the decision, the team added more use cases to their
web 2.0 capabilities for Portal. The team has been charter document and spent the extra time until
inaugurated long before there where any planning shipment on creating more functionality. In par-
activities for a release 6.1. Without knowing yet in allel another tiger team began driving the move
which product release their code will eventually towards WAS 6.1 (Figure 2).
be shipped, they delivered early prototypes and
presented them at the LotusSphere conference,
on IBM’s greenhouse site, and as well as part of Cross-organization
the Portal 6.1. Beta. Most important, the received teaming Structure: Tiger
customer feedback has been used to improve the Teams drive innovation
web 2.0 capability throughout the project.
Each team is continuously updating and adapt- People fail to get along because they fear each
ing their charter document as well as any other other; they fear each other because they don’t
planning document to reflect changing overall know each other; they don’t know each other
constraints, like modified release schedules, de- because they have not communicated with each
pendencies on other teams, or different high-level other. Martin Luther King Jr.
requirements. Each team’s project plan is only
specific and confirmed for the next few weeks For each of the focus areas, a tiger team is
(one “iteration”), while the rest of the project’s established with the right set of skills (especially
duration is tentatively proposed, but subject to test and development) to fulfill its mission. While
change. the Portal overall organization is traditionally
The initial budget defined for a Tiger Team is structured into product components, the new
only the first rough guess to help the Tiger Team tiger teams, which are founded to address certain
with some guidelines. During the initial planning focus areas, are virtual teams. They focus on their
and the composition of the overall Release Plan, deliverable rather than on existing organizational
this budget may very well get adjusted to ensure structures. They span multiple organizations and
that the focus items for the release get in with components, if necessary and possible. Each

397
WebSphere Portal 6.1

Figure 2. Teaming: adjusting goals

tiger team owns the responsibility for planning, Iterative development


designing, coding, documenting and testing of approach: Evolve a
their deliverables. Use cases are implemented by solution and avoid the Big
a single team end-to-end and across component Bang
boundaries, rather than coordinating multiple
parallel changes done by multiple teams. This Life is what happens to you while you’re busy
approach minimizes the cross-team dependen- making other plans. John Lennon
cies, hand-offs and task switching. It supports
continuous communication and focus on the It is hard to integrate a significant amount of
deliverables. code at a certain milestone date without causing
Most importantly the functional verification a major disruption. To avoid painful integration
testing of the team’s deliverables is the team’s struggle, Portal has adopted an iterative develop-
own responsibility. Continuous testing and direct ment model. Code is continuously integrated into a
collaboration between developers and testers common code stream and functionality is brought
improve the process of troubleshooting and bug forward in multiple, small iterations.
fixing tremendously as well as creates a very ef- The purpose of an iterative development is
ficient bridge between the development and test nicely summarized in IBM’s Agile Manifesto:
organization (Figure 3).
Agile Software Development uses continuous
stakeholder feedback to deliver high-quality and
consumable code through use cases and a series
of short, time-boxed iterations.

398
WebSphere Portal 6.1

This implies a few key assumptions: The build environment gives some support,
by running each code change through a pre-
• The duration of an iteration varies between build to surface potential issues prior to the
4 and 6 weeks. regular production build. It is a mandate to
• The content of each iteration is defined at focus on any open issues and bugs first, be-
the beginning of each iteration. A tiger team fore proceeding with the development of new
picks the top use cases from their prioritized functionality (“Stop-the-line” concept).
team charter and starts designing and coding • Part of the iteration is functional verifica-
those items. tion testing within the tiger team. Only
• Throughout the iteration the teams continu- tested and working use cases are accepted
ously integrate their code, documentation as a delivered achievement.
and automated test cases into a common • Performance and documentation are
code stream. There are daily builds of the further aspects to be covered within the
entire product. Continuous integration with iteration.
immediate testing is extremely important • In order to be able to implement a use case
to avoid the destabilization which typically within an iteration it is extremely important
arises, when multiple developers add code to break larger user stories into smaller,
back into the code stream, which they have digestible chunks.
accumulated over quite some time. • Iterations are time-boxed. They have a
• Everyone’s major goal is to ensure the sta- defined start and end date. Usually all tiger
bility of each build. Disruptive changes are teams operate on the same iteration sched-
to be avoided by all means. Thorough unit ule.
testing and automated regression testing is • At the end of each iteration, all tiger teams
the responsibility of every single developer. are jointly demonstrating their deliverables

Figure 3. Teaming: theme and component matrix

399
WebSphere Portal 6.1

to the worldwide team. This demo event is Test driven development:


referred to as “Integration Fest” and is Gain Trust and Quality from
performed using a regular build. The demo Day One
should proof that a stable, tested, and usable
version of the product has been accomplished If anything can go wrong, it will. Murphy‘s
and that this version can be delivered to Law
exploiters of WebSphere Portal.
• At the end of an iteration, the tiger team As mentioned above, the collaboration be-
needs to proof that it has executed the agreed tween development and test is absolutely key for
test cases, and that no defects are open. This achieving a usable level of quality throughout the
will also be validated by Performance and release project.
System Tests that immediately follow each There are several different test consider-
iteration. ations:
• Feedback by the exploiters is incorporated
into the next revision of design and plan. • Build Verification and Smoke Testing is
done for each product build. It is partially
Throughout the release, the teams maintain automated and covers the very basic product
their prioritized list of use cases which they ten- capabilities.
tatively want to address in the foreseeable future, • Functional verification testing is done
and they have elaborated a rough high level design within each Tiger team and is an integrated
outlining all items of their focus area. But only part of ANY development activities. De-
the current iteration is being precisely planned sign documents with use case descriptions
and detailed into work items as well as low level already consider test scenarios and test
use case descriptions (Figure 4). planning. The testing is includes automated
regression testing.

Figure 4. Iterations: executing the release project

400
WebSphere Portal 6.1

• Extended Functional Verification Test- There have been key challenges to resolve,
ing is executed by a dedicated test team in but there is no generic solution, which fits all.
parallel to the ongoing iterations. Main focus The experience made by WebSphere Portal can
is extending platform coverage and adding be summarized as a set of guidelines and con-
more regression testing. siderations:
• Performance testing is going on through- Finding the right balance between flexibility
out the release project. Goal is to identify and planning is difficult. Do not waste time with
performance regression, but also to help planning exercises on a too granular level of detail.
the development teams with analyzing the The project status 6 months from now is not pre-
performance bottle necks in the implementa- dictable. Making early commitments will narrow
tion. down the possible options for your future.
• System Verification Testing is done in par- Flexibility is good, but no for free! Deferring
allel to the ongoing iterations as well. This decisions will require additional investment, as
testing covers long running test scenarios some simplifying assumptions cannot be made yet.
as well as complex environment such as Planning and documentation are needed to docu-
cluster setups. Goal is to identify potential ment agreements and commitments, to describe
memory leaks or instabilities early. the architecture, and to specify interfaces. Don’t
• Once the last iteration is completed, a “Re- use “Agility” as an excuse for lack of planning
lease Closure” phase is appended, in which and design preparation. This is quite similar to
the test teams conduct further regression purchasing plane tickets: If you decide late and
testing. There is especially a focus on system want full flexibility, the airfare will be higher
verification testing. compared to early bookings for a fixed date.
In the context of the Portal product, key archi-
tectural decisions, like choosing the underlying
The Issues and pain points WAS version or selection of code streams in the
source library system, are constraints which are
No man ever steps in the same river twice, for it rather expensive to hold off. Defining the major
is not the same river. Heraklit focus areas of investment within the release proj-
ect and a rough timeline is crucial. It needs to be
In summary, agile development practices done early, whereas finalizing the exact feature
proofed to improve the flexibility of the release scope or the shaping the exact out-of-the-box
project and especially the efficiency within the functionality can be done rather late. The first
development and test organization. Neverthe- iterations will give a well-founded assessment of
less the move from the waterfall model towards progress, status, quality and the remaining capac-
tiger teams and iterations has been a challenging ity of the teams. That kind of information will be
journey. Portal has mixed and matched suitable a much more reliable baseline for an ongoing fine
ideas from the palette of known agile techniques grained planning.
and concepts. Part of this journey has always It is important to break down big pieces of
been to reflect the applied techniques and adapt work in smaller use cases, which can be imple-
the progress. Learn as you go. Each project is mented and tested within a single iteration. This
different. Each team has different needs to per- helps to create frequent and measurable results,
form. Continuous feedback from the overall team instead of piling up a lot of unfinished work. It
triggered the continuous improvement of Portal’s also helps to monitor and understand progress – an
agile approach. essential information for fine tuning the planning
of the upcoming iteration.

401
WebSphere Portal 6.1

You need to plan for change. Therefore any Acknowledgment


planning or commitment needs to ensure, that there
is sufficient buffer to accommodate the unforesee- IBM, the IBM logo, Lotus, WebSphere are trade-
able events. It is very useful to only commit to few, marks of International Business Machines Corpo-
most important use cases initially. The team can add ration in the United States, other countries, or both.
more, once the product is getting shape and there Java and all Java-based trademarks are trademarks
is less ambiguity in the planning constraints. of Sun Microsystems, Inc. in the United States,
Test automation is crucial. If there is no exten- other countries, or both. Microsoft and Windows
sive coverage of automated functional verification are trademarks of Microsoft Corporation in the
tests, it is impossible to avoid regression problems, United States, other countries, or both.
when the next set of use cases is implemented in a
subsequent iteration.
Agile Software development is demanding Summary
openness for change and flexibility from everyone.
Agile needs commitment by the entire organization The adaptation of agile software development
in order to work out: has been a quite challenging and disruptive
Each Tiger team is accountable for its results. exercise. Most important has been the ability
Everyone needs to understand what other team to break down complex development items into
members do. Teams need to keep risks in mind, manageable use cases, which can be designed,
take over responsibility and be prepared to come scheduled, implemented, tested and tracked easily.
up with mitigation plans quickly. Tiger teams span The major challenge has been to get the ongoing
organizational structures and location. They drive support from the key players in the organization.
customer oriented, end-to-end use cases with less As described above, the move towards a more
gaps in between. They think out of the box rather distributed leadership model, as well as letting
than being cramped by a component centric point go of centralized control has been controversial
of view. This is the key strength of Portal’s agile throughout the project. But in the end, agile
approach. development has been a convincing approach,
The leadership team needs to trust and em- despite all the issues and struggles which had
power the teams. They need to give guidance by to be resolved during the project. Its flexibility
communicating a well defined overall Vision, turned out to be especially supportive for such
high level requirements and a rough timeline in a large development organization. Future Portal
which deliverables are expected. And they need releases will continue to evolve agile development
to encourage a common team spirit within each and further improve and adapt the process to meet
tiger team. the team’s requirements.
Agile Software Development is a lot about
philosophy. As mentioned above, there is no gen-
eral rule, no generic process. Mix and Match of References
various techniques is the right way to go, as long
as you accept that change is part of your plan.
Based on the experiences outlined in this article, Beck, K. (2002). Test Driven Development.
the authors have published a comprehensive book Addison-Wesley.
covering the basics of agile software development. Kessler, C., & Sweitzer, J. (2007). Outside-in
In particular, this book emphasizes best practices Software Development. Addison-Wesley.
for large software development projects.

402
WebSphere Portal 6.1

Poppendieck, M., & Poppendieck, T. (2006). Links


Implementing Lean Software Development.
Addison-Wesley. Agile Forum: http://ibmforums.ibm.com/forums/
forum.jspa?forumID=2710
Schwaber, K. (2004). Agile Project Management
with Scrum. Microsoft Press. Poppendieck Lean website: http://www.pop-
pendieck.com/
Stober, T., Hansmann, U. (2009). Agile Software
Development: Best Practices for Large Software ScrumAlliance.org: http://www.scrumalliance.
Development Projects. Springer. org/

This work was previously published in International Journal of Web Portals, Vol. 1, Issue 3, edited by J. Polgar, G. Adamson,
pp. 44-55, copyright 2009 by IGI Publishing (an imprint of IGI Global).

403
404

Chapter 2.9
Adaptation and
Recommendation in Modern
Web 2.0 Portals
Andreas Nauerz
IBM Research and Development, Germany

Rich Thompson
IBM T.J. Watson Research Center, USA

Abstract INTRODUCTION

In this paper, we propose a generic recommender In recent years Enterprise Information Portals
framework that allows transparently integrating have gained importance in many companies. As
different recommender engines into a Portal. The a single point of access they integrate various ap-
framework comes with a number of preinstalled plications and processes into one homogeneous
recommender engines and can be extended by user interface. Today, typical Portals contain
adding further such components. Recommen- thousands of pages. They are no longer exclusively
dations are computed by each engine and then maintained by an IT department, instead, Web 2.0
transparently merged. This ensures that neither techniques are used increasingly, allowing user
the Portal vendor, nor the Portal operator, nor the generated content to be added to Portal pages.
user is burdened with choosing an appropriate This tremendous popularity and success of Por-
engine and still high quality recommendations tals, has its downsides: Their continuous growth
can be made. Furthermore we present means to makes access to relevant information increasingly
automatically adapt the Portal system to better difficult. Users need to find task- and role-specific
suit users needs. information quickly, but face information overload
and feel lost in hyperspace. The huge amount of

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Adaptation and Recommendation in Modern Web 2.0 Portals

content results in complex structures designed to In addition to providing users with recom-
satisfy the majority of users. However, those super- mendations we also adapt the Portal’s structure
imposed structures, defined by Portal authors and automatically to better satisfy users needs.
administrators are not necessarily compliant to Most of our solutions for adapting and recom-
the users’ mental models and therefore result in mending content are based on user and context
long navigation paths and significant effort to models that reflect users’ interest and preferences
find the information needed. The likelihood of a and on annotations of resources provided by users.
mismatch between a user’s mental model and the For instance, we adapt a Portal’s structure (e.g.
administrator’s mental model increases as more navigation) and provide recommendations to be
users access the Portal. This becomes even worse, able to reach content being of interest easier. We
once user generated content is added, where the recommend background information, experts and
structure may not follow the design the adminis- users with similar interests.
trator had in mind. In addition, the more content a In the following we will first give an overview
Portal offers, the more likely it becomes that users of related work. Next, we outline which informa-
are no longer aware of all the resources available tion is needed to achieve our goals, and, more
within it. They might thus miss out on resources importantly, how to obtain the necessary informa-
that are potentially relevant to their tasks, simply tion. Here the focus lies on collecting informa-
because they never come across them. Thus, on tion about users (and their interests, preferences
the one hand, users obtain too much information and thus needs) and the resources they interact
that is not relevant to their current task, on the with. We will show that a mixture of automated
other hand, it becomes cumbersome to find the information extraction and user provided infor-
right information and they do not obtain all the mation is currently the most realistic approach.
information that would be relevant. Users therefore Afterwards we demonstrate how this information
need the Portal to assist them in finding relevant can be exploited to either adapt the Portal or to
information in an efficient manner. issue reasonable recommendations.
Generally this type of problem falls in the do-
main of recommender systems and numerous such
systems have been proposed in recent years. Each Related Work
of these can recommend relevant items for specific
applications or when certain data characteristics The explosive growth of information on the Web
are met, but none meet the breadth needed to ad- has led to the development of recommender sys-
dress assisting a Portal user. In this paper we will tems [Resnick and Varian, 1997]. Recommender
outline a generic recommender framework into systems are a personalized information filtering
which specific recommendation engines can be technology used to either predict whether a par-
installed. The framework decides which engines ticular user will like a particular item (prediction
are likely to produce relevant recommendations problem) or to identify a set of N items that will
for any particular situation and how multiple re- be of interest to a certain user (top-N recommen-
sults sets are combined when multiple engines are dation problem). In recent years, recommender
invoked. The framework comes with a number of systems have been used ina number of different
preinstalled engines and basic configuration for applications such as recommending products
using them. This alleviates the burden on the Portal a customer will most likely buy; movies, TV
administrator relative to the initial configuration programs, or music a user will find enjoyable.
and transparently leverages the best engines for An excellent survey of different recommender
assisting the user in accomplishing their task. systems for various applications can be found in
[Schafer et al., 1999].

405
Adaptation and Recommendation in Modern Web 2.0 Portals

Over the years, various approaches for build- annotators to annotate information pieces. We
ing recommender systems have been developed have described the first approach in [Nauerz and
[Ramezania et al.,2008]: Collaborative Filtering Welsch, 2007] already. The second approach is
(CF) recommenders use social knowledge - typi- based on information extraction from unstruc-
cally ratings of items by a community of users tured machine-readable documents. Although the
to generate recommendations. Content-based approach to perform the extraction often differs,
(CB) recommenders use item features to recom- most papers in this area regard information ex-
mend items similar to those in which the user traction as a proper way to automatically extract
has expressed interest. Knowledge-based (KB) semantic annotations from web content. Most
recommenders use domain knowledge to generate of these systems are based on machine learning
recommendations. Hybrid recommender systems techniques, e.g. [Dill et al., 2003].
combine two or more techniques to gain better With respect to adaptation systems, systems
results with fewer drawbacks. that build and apply user and usage models to
The recent competition to improve the recom- adapt web sites to the user’s context (interests,
mendation system employed by Netflix has shown preferences, needs, goals, etc.), a lot of research
the value of applying multiple recommendation has already been performed in the field of
engines (a combination of an algorithm and a adaptive hypermedia [Brusilovsky, 2001]. One
specific configuration) to be a problem. The possible approach to derive those models and
Progress winner employed many variants [Robert enable adaptation is to analyze user access data,
M. Bell and Volinsky, 2007], each of which was as Perkowitz and Etzioni [Perkowitz and Etzioni,
designed to do best in certain circumstances, such 1997] propose. Projects in this context include
that the overall quality of the generated recom- PageGather [Perkowitz and Etzioni, 2000], Letizia
mendations was improved. This demonstrates the [Lieberman, 1995] and WebWatcher [Joachims et
value of multiple engines within a single domain al., 1997]. Especially with respect to navigation
and the effect is multiplied when multiple appli- adaptation Smyth and Cotter [Smyth and Cotter,
cation domains are accessible through a single 2003] describe an approach to improve navigation
system, as is common for Portals. in mobile Portals significantly.
More specifically, regarding the recommenda-
tion of expertise, systems that help to find experts
are called expertise finders or expertise location Information about Users,
engines [Zhang and Ackerman, 2005]. A general Behavior, and Resources
architecture for recommendation systems that al-
low locating experts is described in [McDonald From a conceptual point of view (cp. fig. 1) Portals
and Ackerman, 2000]. More specifically Streeter are comprised of various resources such as pages
et al.present Who Knows, a system which recom- and portlets (artifacts residing on pages deliver-
mends experts having knowledge in specific top- ing content). These resources are arranged based
ics based on profiles created from observing the on Portal models, often initially created by some
documents they have selected and worked with administrator with the aim to satisfy the largest set
previously [Streeter and Lochbaum, 1988].Newer of users and not the preferences of each single user.
systems that use information about social networks We therefore need information about individual
to find experts are e.g. [Kautz et al., 1997]. users (or groups of users) and their behavior as a
Providing background information or inter- basis for both adaptation and recommendations.
linking information pieces is based on the ability We apply different techniques such as web mining
to either allow users or programmatic, automated, to construct user models reflecting users interests

406
Adaptation and Recommendation in Modern Web 2.0 Portals

and preferences; we use information from their patterns within web data. Web usage mining is the
static profile (native language, home country, extraction of usage patterns from access log data
working location, age, etc.), their interaction be- to model certain aspects of the behavior of users.
havior (pages and portlets they work with; tags Our system has been incorporated into IBM’s
they apply to resources), and their social network WebSphere Portal. Analyzing its logs reveals
to derive knowledge about their needs. We observe information about, among other things, several
the context (date, time, location, ...) in which they interesting events, e.g. when pages (or portlets)
interact to partition the user model in so called are created, read, updated or deleted, when pages
profiles like private or business. Additionally, we (or portlets) are requested,when users are created,
need enriched information about the resources updated, deleted and many more.
available in the system. We illustrate how we Analyzing the log allows to understand which
extract information pieces of certain type in order pages and portlets a user typically works with.
to provide background information by connecting Obviously, the user model must allow the calcula-
to external sources and to interlink them in order tion of the utilization of pages and portlets from the
to issue recommendations. historical data available. We do this by measuring
how often a user interacts with certain pages and
Extracting Information about Users portlets. Of course, we also consider interactions
that occurred recently to be more important than
User Model interactions that occurred in the past and we hence
apply time-weighting factors when calculating
In order to perform reasonable adaptations, or to the utilization of pages and portlets based on the
provide users with recommendations, we need target hits they received.
to understand users’ interests and preferences. More generally, we apply techniques from
Therefore we construct user models reflecting their the area of frequent set mining [Liu, 2006] to
behavior. We use static information from users’ analyze the usage of pages and portlets. We use
profiles (describing their age, native language, the Apriori algorithm [Agrawal and R., 1994], a
etc.), as well as dynamical information which we standard association rule mining algorithm, to
retrieve via web usage mining. determine items, such as pages and portlets that
Web Mining [Liu, 2006] is the application co-occur frequently. We apply the GSP algorithm
of data mining techniques to discover (usage)- [Srikant and Agrawal, 1996], a standard sequential

Figure 1. Conceptual overview


Structured based on...

User Modeling Portal Models Recomendation &


Interest & Static Page Models Adaptation Engine
Preferences Profile
Web Portal
Task based
Web Adaption
Tagging Social Layout Models Adapts
Mining
Behavior Network
Location based
Adaption

Context Modeling Extended Models


Content
Date Tasks
User Models Adaptation
Context Content
Uses
Modeling Time Device Recommendation
Model Context Models
Location Creation Expertise
Recommendation

Recommends...

407
Adaptation and Recommendation in Modern Web 2.0 Portals

pattern mining algorithm, to determine sequences time, location, etc.) which define when it should
of items, such as pages and portlets, that co-occur become active.
frequently. Comparing the itemsets even allows Our adaptation and recommendation compo-
to find users behaving similarly. nents utilize both the information stored in the user
Tagging Behavior Analysis. In our previous and context model, to perform their operations (i.e.
work we have incorporated a tag engine into to adapt structures such as the navigation). Techni-
IBM’s WebSphere Portal allowing users to an- cally, the user model is partitioned in a separate
notate resources such as pages, portlets and users. partition for each context profile available in the
Analyzing (and comparing) the tagging behavior context model. To determine the best matching
of users allows for refining the user model. The profile, the system continuously observes a set
general assumption is that tagging expresses of defined context attributes. Users always have
interest in a resource. the option to outvote the system’s decision and to
Social Network Analysis. Finally, the analysis manually switch to another profile. As only one
of users’ explicit contacts allows to determine context profile can be active at one specific point in
users’ interests and preferences, too. The assump- time, whatever people do only influences the user
tion is that the fact that users directly know each model partition associated to the currently active
other can be an indication for related job roles and profile. For example, if the currently active profile
hence for sharing similar knowledge. is trip planning, all the navigation behavior, does
not influence the user model partition associated
Context Model to the profile project management.

Focusing on user models only neglects the context Extracting Enriched Information about
users are currently acting in. Hence, these could Resources
be regarded suitable models, only, if the role, the
interests and preferences of users will not change To extract enriched information about the re-
too much over time. In reality, a user’s needs usu- sources, we currently allow for the usage of three
ally change if their context changes. For example, different mechanisms:
a user in the process of planning a business trip Automated Tagging. Here the system analyzes
will need resources that provide information about markup generated by the Portal to find occurrences
hotels, rental cars, and flights. When the same of identifiable information pieces of certain types
user returns to their tasks as a project manager, such as persons, locations, etc., and wraps these
a completely different set of resources is needed. into semantic tags. We have integrated the UIMA
Of course, interests and preferences will be differ- framework 1 and written customized analysis
ent in these roles and access to a different set of engines able to identify such information pieces.
resources (pages, portlets, etc.) will be needed. Semi-automated Tagging If the system cannot
Our solution allows single users to have several unambiguously identify the type of an informa-
context profiles between which either the system tion piece it still allows users to mark it and tell
switches automatically, based on context attributes the system of what type it is. We call this process
being observed (date, time, location, etc.), or the semi-automated tagging. For instance, if we find
user can manually switch. New profiles can be a fragment ”Paris H. was sighted leaving a Hotel
defined using a profile management portlet which in Paris” it becomes difficult for the system to
allows to specify the initial settings of a profile determine whether Paris is a name or a loca-
(which theme to use, which skin to use, etc.) and tion. The user can then mark the corresponding
to associate it with a set of context attributes (date, information pieces and tell the system their type.

408
Adaptation and Recommendation in Modern Web 2.0 Portals

The information pieces are then wrapped into a an extractor framework, a management layer, an
semantic tag exactly as outlined before. engine selector and an engine-hosting layer.
Manual Annotating Moreover, our system The extractor component manages getting
allows semantically tagged information pieces incremental changes from the customer’s un-
to be annotated manually again. For example, if derlying databases and making them available
the name of three persons Alice, Bob, and Charly for processing into the recommendation models
often appear somewhere in the Portal system, e.g. used by the deployed engines. While there are
in blog- or wiki portlets, our system automatically some cases of reusing shared (or system-level)
determines these fragments to be of type person, models, most recommendation engines build
wraps them into semantic tags and allows for ad- their own models and optimize these for rapid
vanced interaction with these information pieces. recommendation generation. As a result, each
Our tag engine allows these enriched fragments engine’s configuration will typically register a
to be annotated e.g. with the term project-x which processing module with the extractor in order to
indicates that all three persons are somehow re- receive and deal with each incremental update to
lated to this project. This means that the options the customer’s data. By centralizing the extraction,
for manual annotating allow for a finer-grain impacts on the customer database are minimized
categorization of information pieces. as is the effort need to provide access to a new
type of data source.
Supporting Framework There are two components which are in-
strumented through the page and page-support
Figure 2 shows the architecture of our recom- mechanisms. The first of these manages contex-
mendation framework. The main components are tual information. Some contextual information is
already gathered elsewhere (e.g. from information

Figure 2. Recommendation framework


Web Portal
Users (REST) API calls
Recommender Framework API
(Contextual) information

Real-time Management Layer


events
Context Recommendation Engine
Manager Manager Selector

Customer Context Context


Database Database Store

Engine Layer
Hybrid
Engines

Engine Engine Engine Engine


Contextual
Engines

System level models

Processed
Extractor Customer
Database

409
Adaptation and Recommendation in Modern Web 2.0 Portals

already logged), as discussed in Section 3.1, and supplied along with the base configuration. The
this is accessed from those sources; for example, recommendation manager will then combine the
the pages activated. Other contextual information responses from all invoked engines and provide
is application specific and are sent as contextual a single, unified response to the client.
events into the system; for example, dynamically The recommendation engines are hosted by a
presented information has been accessed. This layer that uses Inversion of Control [Wikipedia
information is timestamped and placed into a Foundation, 2008] concepts to define engine in-
central store which system’s components, usu- stances and the configuration parameters to apply
ally the recommendation engines, can use when to each instance. Often the referenced components
responding to a request. are wrappers that map our internal API and data
The other part instrumented on the page structures to the ones used by the actual engine.
is requesting recommendations be generated. For example, a wrapper for the open source col-
Each page could be displaying multiple differ- laborative filtering engine Taste [Apache Foun-
ent types of recommendations (experts, similar dation, 2008] has been used to incorporate this
people, pages, portlets, etc). These are requested, externally developed engine. Due to the nature
normally using an AJAX request, and displayed of the design, we expect similar wrappers to be
in whatever manner the user experience prefers. easily developed for other external engines. This
The system provides dojo widgets which can be architecture allows very different engines to be
leveraged for quickly adding recommendations deployed (collaborative filtering, content-based,
to a page. These include a data widget which knowledge-based, etc.) and work independently
manages requesting recommendations, parsing of each in a manner that will together generate
the response into javascript objects and provid- improved results for the user. It also makes it easier
ing easy access to these objects. It also includes to add recommendations to a website as many of
both a simple display widget for showing any one the issues of accessing engines are provided by
recommendation type (e.g. experts) as an HTML the framework.
list and a more complex widget that displays each
recommendation type in a separate tab of a tabbed
pane construct. Exploiting the Models for
When the system receives a request for recom- Adaptation and
mendations, it examines the request to determine Recommendation
what recommendation engines should be used
to generate the response. The management layer Now that we have described which information
delegates this decision to an engine selector and about the Portal resources and users are available
then invokes the identified engines, passing them to our system, we can explain how this information
both the request parameters and a reference to the is used to improve the user experience with the
active context store. The engine selector remains Portal. We propose methods to adapt the content,
an area of active research with the simplest variant to recommend content, to offer additional informa-
being a rule engine. It is expected that this research tion and to recommend experts. In the following,
will realize a significant improvement by having more details about the approach are given.
the engine selector specify both the engines to use
and how their results are to be combined. Each Adaptations
engine will use whatever portion of the supplied
information is appropriate to their algorithm Within the context of this project we have come
in order to generate their response. Section 5.2 up with different solutions allowing for adaptation
discusses some of these engines, including those

410
Adaptation and Recommendation in Modern Web 2.0 Portals

and recommendation of the Portal’s structure. Automated Recommendation. Especially users


Most of them focus exemplarily on the adaptation that navigate according to the aimed navigation
of the navigation. paradigm [Robertson, 1997] will not like auto-
Manual Adaptation. First, options to manu- mated adaptations because of its aggressiveness.
ally adapt the navigation have been introduced. Automatic provisioning of recommendations
Therefore we implemented specialized portlets avoids the permanent restructuring of the naviga-
that allow each single user to generate her own tion while still providing users with shortcuts. We
navigation matching her preferences best. The blend-in recommendations into the Portal’s theme
first portlet allows users to generate their own that provide users with reasonable shortcuts to
navigation by hiding irrelevant nodes (pages) and relevant pages. These shortcuts are dynamically
by reordering nodes being part of the navigation generated depending on the current navigational
in order to reach relevant nodes more quickly. position. Our recommendation system applies a
The second portlet allows users to record paths MinPath algorithm [Anderson et al.,2001]. We try
(i.e. sequences of pages) traveled often. These to predict shortcuts to nodes that are far away from
recordings can be recalled later and navigated the current node but have a high probability to be
through by just clicking previous and next links. navigated to. The probability itself is calculated
The recordings can even be exchanged with other based on Markov chains as described in [Anderson
Portal users which allows experts to record com- et al., 2002; Smyth and Cotter, 2003].
mon paths for their colleagues. Context-adaptivity. As mentioned above, users
Automated Adaptation. Automated adapta- may have several different context profiles. By
tion relieves users from generating an optimized switching to a different profile, the Portal will
navigation manually. We leverage our user models be adapted accordingly based on the information
to understand users’ needs.We use a structure contained in that profile.
reordering algorithm to rearrange pages: more
important nodes are promoted to better naviga- Recommendations
tional positions, less important ones demoted or
even hidden. Continuous adaptation, based on In our framework, we have developed a set of rec-
the most current user models available, guar- ommender engines that provide recommendations
antees that the navigation permanently fits the to background information and related content
users needs as best as possible. As soon as users’ either by displaying shortcuts to relevant pages
behaviors change their user model changes, too or by showing widgets/portlets containing links
and hence the navigation provided. We leverage to relevant information. Currently these engines
our user models to understand users’ needs. We come in two flavors (though the framework is
use a structure reordering algorithm to rearrange not restricted to just these flavors): recommender
pages: more important nodes are promoted to engines based on the user model and collaborative
better navigational positions, less important ones filtering recommenders. Further-more, both types
demoted or even hidden. Continuous adaptation, of recommender engines can leverage the context
based on the most current user models available, model in order to provide recommendations with
guarantees that the navigation permanently fits respect to the current activity of the user.
the users needs as best as possible. As soon as
a user’s behavior changes, their user model is Recommenders Based on User Model
updated and hence the navigation provided takes
into account the new behavior. The first category of recommender engines pro-
vide recommendations to related content based on

411
Adaptation and Recommendation in Modern Web 2.0 Portals

the information stored in the user model, which Other Recommenders


reflects static information about the user (like
language preferences, age, location, etc.) as well as While not included in the current set of supplied
navigation behavior of individual users, including recommendations engines, the framework takes
their favorite pages and routes through the entire into account that other recommendation strategies
navigation topology of the Portal. Recommenda- exist (and continue to be invented). At a minimum
tions are generated based on certain properties of these include Content-Based recommenders and
the user model and the resulting recommendations various different approaches to Knowledge-Based
can be used to suggest forward navigation links or recommenders.
to automatically change the navigation topology
in order to reveal the interesting pages and hide Context-Based Recommendations
irrelevant content.
So far we have described only the recommenda-
Collaborative Filtering tions based on the navigation history of individ-
Recommenders ual users and commonalities of interests among
multiple users without respect to the context in
In contrast to the recommenders that provide which the users are acting. However, the initial
recommendations based on the analysis of the recommenders of both categories can also access
history of a particular user, the collaborative the context model in order to provide recommen-
filtering (CF) recommenders provide recom- dations that could be especially relevant in certain
mendations based on the analysis of the behavior situations. Our recommendation framework al-
of multiple users. The collaborative filtering lows single users to have several context profiles.
engines first try to identify the users that have These profiles can be recommended to the user
similar tastes and behaviors to the current user automatically by the system or users can switch
and then retrieve the items that these users liked between them manually.
most and recommend them to the current user. New profiles can be defined using a profile
The CF-based recommendations help to discover management portlet which allows specifying the
new and unknown content items that might be of settings of a profile (which theme to use, which
interest for a particular user. skin to use, which navigation topology to display
The CF-approach can be applied in numerous etc.) and to associate it with a set of context at-
ways; for recommending navigation links, the set tributes which define when it should become
of similar users might best be chosen based on active. Whatever people do only influences the
having a similar set of target hits as the current models associated to the current active profile.
user. The CF-approach would then recommend If the currently active profile is business, all the
those pages for which these users also have high navigation behavior, all the usage behavior of
target hits, but which the current user does not. Portal pages etc. does only affect the models as-
In domains such a catalog systems (e.g. movie sociated to this profile, but never influences the
sales/rentals), preferrably the ratings of the user vs modelsassociated to the private profile. For the
other users is used to generate the recommenda- determination of the best matching profile the
tions. In the absence of explicit ratings, previous system continuously observes a set of defined
history and the items viewed within the current context attributes. Users always have the option
interaction session can be used as implicit forms to outvote the system’s decision and to manually
of ratings. switch to another profile. A learning mode allows
users to let the system learn about their needs and
tastes in a specific new context.

412
Adaptation and Recommendation in Modern Web 2.0 Portals

Recommending Background Recommending Related Content


Information
Analyzing occurrences of semantically tagged
As said, in today’s Web 2.0 world content is cre- information pieces also allows us to recommend
ated by entire user communities. Different users related content. For instance, if the term Web-
use different terms to describe the same things. Sphere Portal is identified in a news portlet and
Some terms might be well-understood by most hence semantically tagged as a product name our
users, some might not. system would provide users with background
Thus looking up terms is needed more fre- information about WebSphere Portal probably by
quently and becomes a tedious task. But when linking to the product site. But, within a Portal
reading web sites, users want background informa- system, the same term might occur at many other
tion at their fingertips. If they do not understand places, e.g. in a wiki portlet where users have
what an abbreviation or a term stands for, who posted some best practices, tips and tricks when
a certain person actually is, or, where a certain working with this product, in a blog where users
city is actually located, they want to be able to have commented on the product and so forth. We
retrieve this information as easily and quickly track all occurrences and recommend (an appro-
as possible. They do not want to fire up a search priate subset) of them as related content as soon as
engine to search for another site from which they the user interacts with one single occurrence.
could probably get the information they want, but This can even be taken one step further.
rather be provided with that information directly, As mentioned above, we allow users to annotate
in-place. We provide an environment which un- already semantically tagged information pieces.
obtrusively enriches the information pieces to This way we can recommend related content not
allow for such look-ups. only by having identified ”exactly matching”
Figure 3 shows our system in action: it il- occurrences of semantically tagged information
lustrates how a fictious person name (John Doe), pieces, but also by having identified similarly
a location (Stuttgart), and a currency have been annotated, but differently semantically tagged,
identified within a text fragment residing in a information pieces. For example, if Alice, Bob,
portlet and are visualized to the user. Pop-ups and Charly have been annotated as persons and
provide the users with background information. a user tagged them with the term project-x to ex-

Figure 3. Recommending background information and related content

413
Adaptation and Recommendation in Modern Web 2.0 Portals

press their relationship to this project, this allow For example, if users A, B, and C often work
us to recommend other users of the community with the pages and portlets underneath the page
as related ”content” as soon as one user is clicked, entitled My News we can, on the one hand as-
just because they all seem to be assigned to the sume that they have knowledge about how to deal
same project. This can be useful, e.g., if a user has with the pages and portlets provided here, and,
a question to something he is reading about and on the other hand assume that they have similar
tries to contact the author whose name is given interests as they do similar things. A user D ac-
but who is currently unreachable. Recommend- cessing the same pages and portlets rarely can
ing related users allows him to easily determine then be presented with A, B, and C as experts
backups that could probably help him, too. when dealing with the information and services
Figure 3 shows how we can recommend related provided. Therefore we have designed a special-
information for the detected information pieces ized portlet (cp. Figure 4) that can be accessed
Stuttgart and John Doe (other people probably from every page (on demand).At the top of the
working in the same team, on the same project portlet contacts are listed which have explicitly
etc.). been added as such by the user. Adding contacts
explicitly demands sending a request to the contact
Recommending Expertise to be added which, in turn, can accept or reject this
request. The second section displays contacts the
As said, user models also tell us about with which system has determined to behave similarly which
pages and portlets a user is typically working can be derived by comparing user models. These
with. are contacts the user might want to get in touch
The first assumption is that users working with, generally to share knowledge. The third
with certain pages and portlets more often have section displays contacts currently performing
more expertise about how to use them than other similar actions within the Portal (e.g. viewing the
users have. The second assumption is that users same page or working with the same portlet). The
working with the same pages and portlets more last section at the bottom of the portlet displays
often have a similar behavior and hence interests contacts the system has determined to be experts
and preferences. with respect to the (content) area currently being
visited by the user.

Figure 4. Recommending experts

414
Adaptation and Recommendation in Modern Web 2.0 Portals

The list of users displayed in the last three All the approaches proposed in this paper
sections dynamically changes as the current user have been implemented and integrated into IBM’s
interacts with and navigates through the Portal, WebSphere Portal.
while the first section always displays a static list For evaluation purposes we have set up a demo
of contacts. The portlet provides several functions system and performed some initial surveys. 100%
to interact with the contacts being listed: of all participants (all computer scientists, male,
People-awareness ensures that users contact 25-50 years old) regarded the system as useful.
other users being available more likely than the Of course, we plan to perform more systematic
ones being unavailable. To ensure that users do not evaluations within the next months.
disturb the latter, e.g. during meetings, an online Future work includes the extension of our
status is displayed for each single user displayed recommendation and adaptation techniques.
as part of the portlet. Tagging and rating func-
tions allow contacts to be tagged, expert users
even to be rated. This allows for a user-driven Acknowledgment
categorization of the contacts. Additionally, rat-
ing allows assessing how helpful an expert was. IBM and WebSphere are trademarks of Inter-
Profiles display information about the contact’s national Business Machines Corporation in the
official job role, his position within the organiza- United States, other countries or both. Other
tion’s hierarchy, his address, and so forth. Social company, product and service names may be
network visualization functions allow the user trademarks or service marks of others.
to see a visual depiction of how his contacts are
related to him and to each other. It presents a References
graph which nodes are the user’s contacts. It al-
lows users to determine which users are part of
R. Agrawal and Sirkant R. Fast Algorithms for
the same team, who personally knows whom and
Mining Association Rules. In Proc. of the 20th
hence allows e.g. to find backups for persons being
Very Large Data Bases Conf., Santiago, Chile,
currently unavailable easily. It is also possible to
1994.
see contacts as marks on a GoogleMap.
Corin R. Anderson, Pedro Domingos, and Daniel
S. Weld. Adaptive web navigation for wireless
Conclusion and Future Work devices. In IJCAI, pages 879–884, 2001.
Corin R. Anderson, Pedro Domingos,and Daniel
In this paper we have presented a solution for
S. Weld. Relational markov models and their ap-
adapting and recommending content and expertise
plication to adaptive web navigation. In KDD,
to satisfy Portal user needs and improve collabo-
pages 143–152, 2002.
ration among them. We have shown means to
collect the necessary information and to adapt the Apache Foundation. Mahout (a.k.a. Taste), 2008.
navigation structure, to recommend background http://lucene.apache.org/mahout/.
information, related content and expertise.
Peter Brusilovsky. Adaptive Hypermedia. User
We have presented an extensible, pluggable,
Modeling and User-Adapted Interaction, 11(1-
recommender framework which allows several
2):87–110, 2001.
recommender engines, that can even apply dif-
ferent recommender technologies, to be used Stephen Dill, Nadav Eiron, David Gibson, Daniel
transparently. Gruhl, Ramanathan V. Guha, Anant Jhingran,

415
Adaptation and Recommendation in Modern Web 2.0 Portals

Tapas Kanungo, Sridhar Rajagopalan, Andrew MaryamRamezania,Lawrence Bergman, Rich


Tomkins, John A. Tomlin, and Jason Y. Zien. Thompson, Robin Burkea, and Bamshad Mobash-
Semtag and seeker: bootstrapping the semantic era. Recommending recommenders. In Intelligent
web via automated semantic annotation. In WWW, User Interfaces, page 439, 2008.
pages 178–186, 2003.
aul Resnick and Hal R. Varian.Recommender
Thorsten Joachims, Dayne Freitag, and Tom M. systems. Commun. ACM, 40(3):56–58, 1997.
Mitchell. Web Watcher: A Tour Guide for the
Yehuda Koren Robert M. Bell and Chris Volinsky.
World Wide Web. In Proc. of the 15th Intl. Joint
The bellkor solution to the netflix prize, 2007.
Conf. on Artificial Intelligence, pages 770–777,
1997. George G. Robertson. Navigation in Information
Spaces, 1997. Submision to CHI 97 Workshop
Henry A. Kautz, Bart Selman, and Mehul A.
on Navigation in Electronic Worlds, March 23-
Shah. Referral web: Combining social networks
24th, 1997.
and collaborative filtering. Commun. ACM,
40(3):63–65, 1997. J. Ben Schafer, Joseph Konstan, and John Riedi.
Recommender systems in e-commerce. In EC
Henry Lieberman. Letizia: An Agent That As-
’99: Proceedings of the 1st ACM conference on
sists Web Browsing. In Chris S. Mellish, editor,
Electronic commerce,pages 158–166, New York,
Proc. of the 14th Intl. Joint Conf. on Artificial
NY, USA, 1999. ACM.
Intelligence, pages 924–929, Montreal, Quebec,
Canada, 1995. Morgan Kaufmann publishers Inc.: B. Smyth and P. Cotter. Intelligent Navigation for
San Mateo, CA, USA. Mobile Internet Portals. In Proc. of the 18th Intl.
Joint Conf. on Artificial Intelligence, Acapulco,
Bing Liu. Web Data Mining: Exploring Hyper-
Mexico, 2003.
links, Contents, and Usage Data (Data-Centric
Systems and Applications). Springer, 2006. R. Srikant and R. Agrawal. Mining Sequential
Patterns: Generalizations and Performance Im-
David W. McDonald and Mark S. Ackerman.
provements. In Advances in Database Technology
Expertise recommender: a flexible recommenda-
- EDBT’96,1996.
tion system and architecture. In CSCW, pages
231–240, 2000. L.A.StreeterandK.E. Lochbaum. Who knows:
A system based on automatic representation of
Andreas Nauerz and Martin Welsch. (Context)
semantic structure. In RIAO, pages 380–388,
Adaptive Navigation in Web Portals. In Proc. of
1988.
the Intl. IADIS WWW/Internet Conference 2007,
Vila Real, Portugal, October 2007. WikipediaFoundation.Inversionofcontrol,2008.
http://en.wikipedia.org/wiki/Inversion of con-
Mike Perkowitz and Oren Etzioni. Adaptive
trol.
Web Sites: an AI Challenge. In Proc. of the 15th
Intl.Joint Conf. on Artificial Intelligence, pages Jun Zhang and Mark S. Ackerman. Searching
16–23, 1997. for expertise in social networks: a simulation of
potential strategies. In GROUP, pages 71–80,
Mike Perkowitz and Oren Etzioni. Towards adap-
2005.
tive Web sites: conceptual framework and case
study. Artif. Intell., 118(1-2):245–275, 2000.

This work was previously published in International Journal of Web Portals, Vol. 1, Issue 2, edited by J. Polgar and G. Adamson,
pp. 1-17, copyright 2009 by IGI Publishing (an imprint of IGI Global).

416
417

Chapter 2.10
Context-Aware
Applications 
for the Web:
A Model-Driven Development Approach
Florian Daniel
University of Trento, Italy

Abstract context-awareness require augmenting the expres-


sive power of conceptual models in order to be able
Adaptivity (the runtime adaptation to user profile to express adaptive application behaviors.
data) and context-awareness (the runtime adapta-
tion to generic context data) have been gaining
momentum in the field of Web engineering over Introduction
the last years, especially in response to the ever
growing demand for highly personalized services The evolution of the Information Technology in the
and applications coming from end users. Develop- last years has seen the World Wide Web transforming
ing context-aware and adaptive Web applications from a read-only hypertext media into a full-fledged,
requires addressing a few design concerns that are multi-channel and multi-service application delivery
proper of such kind of applications and independent platform. Current advances in communication and
of the chosen modeling paradigm or programming network technologies are changing the way people
language. In this chapter we characterize the design interact with Web applications. They provide users
of context-aware Web applications, the authors with different types of mobile devices for access-
describe a conceptual, model-driven development ing – at any time, from anywhere, and with any
approach, and they show how the peculiarities of media – services and contents customized to the
users’ preferences and usage environments. More
and more users themselves ask for services and
DOI: 10.4018/978-1-60566-290-9.ch003

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Context-Aware Applications 
for the Web

applications highly tailored to their individual goes one step further in the same direction, aim-
requirements and, especially due to the increasing ing at enhancing the application’s usefulness
affordability of new and powerful mobile com- and efficacy by combining personalization and
munication devices, they also begin to appreciate adaptivity based on an application-specific set
the availability of ubiquitous access. In order to of properties (the context) that may affect the
cope with the growing demand for novel, user- execution of the application.
centric application features, such as adaptivity In this chapter, we focus on the development of
and context-awareness, appropriate development context-aware applications for the Web and, in par-
methods for Web applications are required. ticular, we describe a model-driven development
Adaptivity is increasingly gaining momen- method that allows developers to approach the
tum in the context of modern software systems. problem at a level of abstraction that enables him/
Runtime adaptivity provides highly flexible her to focus on the real design challenges of such
and responsive means for the customization of class of applications, leaving low-level implemen-
contents and services with respect to the user’s tation concerns to supporting CASE (Computer-
identity. Varying device characteristics in mobile Aided Software Engineering) tools. Considering
and multi-channel computing environments can that software systems are continuously getting
be adequately taken into account and leveraged more complex and difficult to maintain – partly
by means of adaptive software designs, whose due to the previously described requirements –,
development is facilitated by the availability we believe that efficient abstraction mechanisms
of standardized communication protocols (e.g. and design processes, such as those provided by
HTTP) and markup languages (e.g. HTML or visual, model-driven design methods, are becom-
WML), supported by most of today’s mobile ing crucial. The focus on essential design issues
devices. Multi-channel deployment does no lon- and the ease of reuse in model-driven design
ger require completely different, parallel design methods may significantly accelerate the overall
approaches and rather represents a presentation design process. As we will show in this chapter,
issue on top of unified engineering solutions. starting from application models, code generation
But adaptivity may also enable an application techniques may then provide for the automatic
to take into account a wider range of properties generation of application code or templates, thus
describing the interaction between the user and assuring the fast production of consistent and high
the application, thus paving the way for context- quality implementations.
awareness. Context-awareness (Dey & Abowd,
2000; Schilit & Theimer, 1994) is often seen as
recently emerged research field in information Motivating Examples
technology and in particular in the domain of the
Web. From the perspective of application front- Active application features, such as context-aware
end development it can however be interpreted or adaptive behaviors, may augment the effective-
as natural evolution of personalization and adap- ness of interactions and the efficiency of resource
tivity, addressing not only the user’s identity and consumption in all those situations where services
preferences, but also his/her usage environment. and contents offered by an application strongly de-
Personalization has already demonstrated its pend on environmental situations, users’ abilities
benefits for both users and content providers and or disabilities, or the state or health of a software
has been commonly recognized as fundamental system. For example, typical applications demand-
factor for augmenting the efficacy of the overall ing for active features and adaptivity are:
communication of contents. Context-awareness

418
Context-Aware Applications 
for the Web

• Adaptive personalization. User profile at- to them, supported by proper localization


tributes for personalization purposes may mechanisms. To such kind of applications,
present different levels of variability in the use of context data represents a func-
time. Profile properties may be static in tional requirement, rather than an optional
nature (e.g. the name of a user), slowly feature.
changing (e.g. profile data derived from • Exception handling. Critical events dur-
a user’s browsing behavior) or even fast ing the execution of a software system
changing (e.g. the pulse frequency of a may raise exceptions and require prompt
patient). Adaptive personalization mecha- reactions being performed. Process-based
nisms that take into account such profile or workflow-driven applications, for ex-
peculiarities could allow systems to go ample, represent a typical class of appli-
beyond the common and static tailoring or cations that constantly have to cope with
services and contents. exceptional situations in order to guaran-
• Interaction-enabling functionalities. tee the consistent termination of a running
Context could as well consider handicaps process. Here, adaptive or context-aware
or physical disabilities of users, such as mechanisms could be leveraged to capture
vision problems, blindness or paralysis, respective application events and to enact
to adapt the application accordingly and the pieces of application logic that are nec-
to provide alternative and better suited essary to handle the exceptional situation.
interaction mechanisms and modalities. • Production and control systems. Critical
Adaptivity could thus provide functional- production or control systems may require,
ities enabling handicapped users to proper- for example, highly specific sensing and
ly interact with applications, thus fostering alerting mechanisms to prevent produc-
the accessibility of applications. tion losses or product quality degrada-
• Effective content delivery. In general, tions. Context-awareness could facilitate
whatever context data may be leveraged the timeliness of reactions and the efficient
to provide appropriate contents and pro- handling of dangerous situations, but also
gram features at the right time, priority, proactive maintenance approaches, such as
and emphasis. For example, specifically those adopted in a steadily growing num-
targeted special offers could be advertised ber of hardware/software systems, may be
and directed more effectively, presentation achieved.
properties could be adapted to varying lu- • Self-healing software systems. Autonomic
minosity conditions for better readability, or self-healing software systems elevate
etc. Adaptive or context-aware extensions the idea of proactive maintenance from
could thus enhance the overall effective- hardware to software systems and aim at
ness of applications by adapting individual the creation of computing systems that
application elements to varying users or are able to configure, tune, and even re-
usages of the application. pair themselves. Proactive and adaptive
• Delivery of context as content. Applications capabilities in this context are an essential
may depend intrinsically and in a structural feature.
manner from context data. Location-aware
applications, such as city map services or
navigation systems, treat position data as
core contents of the application and adapt

419
Context-Aware Applications 
for the Web

Reference Scenario electromagnetic interferences that exist in


the presence of high-voltage machineries
To exemplify the concepts introduced in this practically prohibited the use of conven-
chapter and to better convey the underlying ideas, tional, unreliable network technologies.
step by step we will show how we developed one • The pervasiveness and availability of Web
of our demonstration prototypes, the PoliTour applications is continuously growing due to
application. The application runs on a PDA with the introduction of novel networking tech-
wireless Internet access and enables visitors to nologies, such as ADSL (Asynchronous
Politecnico di Milano, Italy, to obtain location- Digital Subscriber Line) or fiber optics for
aware campus details (i.e. information about roads home and office users and WiFi and 3rd
and buildings) while walking through the campus. generation mobile telephony technologies
If a user is about to leave the WiFi-covered area (e.g. UMTS, GPRS, EDGE) for mobile
of the campus, an alert message is shown. users.
• Web applications have proved a high scal-
ability (it suffices to think about certain
Context-Awareness and portal applications that serve millions of
Web Applications users every day), facilitated maintainabil-
ity and high cost efficiency.
Due to a lack of appropriate technologies and
concepts, for a long time context-awareness has Provided that technological advances enable
not been considered suited to the domain of the and facilitate the development of adaptive Web
Web. Web technologies (both hardware and soft- applications, it is important to recognize that
ware) are however continuously evolving and the context-awareness, rather than being a mere
attitude toward reactive and context-aware behav- technological concern, represents a true design
iors in Web applications is changing. As a matter issue. In the following, we will thus focus on the
of fact, support for a multitude of non-functional typical design concerns in the development of
requirements, whose inadequate coverage pre- context-aware Web applications.
vented the adoption of Web technologies for the
implementation of reactive applications, has now Enabling Context-
been developed. Just to mention a few: Awareness in the Web

• The reliability of data communications Developing context-aware applications for the


has been considerably enhanced along Web demands some characteristic architectural
both the software and the hardware dimen- components, in order to support adaptations to
sion. The introduction of reliable messag- context. Figure 1 proposes a possible functional
ing techniques (e.g. digital certificates or architecture that extends the traditional architec-
the WS-Reliability specification) provides ture of Web applications with components aimed
for trustworthy communications on top at supporting the acquisition, storage, and use of
of standard Web protocols, such as HTTP context data.
or SOAP. The success of fiber optics – as The typical context-aware application’s data
an example of hardware evolution – has source includes both the application data (i.e.
allowed the Ethernet protocol (typically the business objects that characterize the ap-
used in the Web) even to enter industrial plication domain and the user) and the context
production environments, where the high model, which offers at any moment an up-to-date

420
Context-Aware Applications 
for the Web

Figure 1. Context data in context-aware Web applications. Gray shaded boxes correspond to conventional,
non-adaptive parts, white boxes correspond to extensions required to support context-awareness

representation of the context state. The context client-side parameters are generated by client-side
model captures all the context-characterizing sensing solutions, server-side parameters are filled
properties and enables the system to adapt to by centralized sensing solutions, and database
changes thereof, assuming that such changes may updates may be performed by both.
demand for proper reactions by the application. Context-awareness in Web applications there-
An application typically consists of adaptive (i.e. fore requires addressing the following issues:
context-aware) and non-adaptive parts; we call
the former adaptive hypertext. The pages of the • Context data modeling. Context properties
adaptive hypertext present some form of adaptive that are relevant for the provisioning of the
behavior, i.e. they are able to react to changes context-aware behaviors of the application
in the context, while pages of the non-adaptive must be identified and represented in an
hypertext do not present any adaptive behavior. application-accessible format. The result
To decide which adaptation is required – if any –, of this task is the context model that can be
the adaptive hypertext makes use of context data queried for adaptation purposes.
during the rendering of hypertext pages. Context • Modeling of adaptive application behav-
data needs to be sensed (e.g. by means of suitable iors. Starting from the context model, ad-
instruments, such as GPS positioning systems, aptation operations need to be defined in
thermometers, or similar) and communicated to order to react to situations demanding for
the Web server that hosts the application, in order adaptation. That is, detected changes to
to be processed. the context data are translated into visible
The above architecture allows for three main effects or operations that aim at augment-
communication mechanisms to pass context data ing the effectiveness and usability of the
from the sensing devices to the application: (i) as application.
parameters sensed at the client side and sent to the • Context model management. The context
application (e.g. GPS position data); (ii) as server- model only captures the static aspect of
side parameters (i.e. HTTP session variables) context data, i.e. their structure; in order to
provided by a centralized sensing infrastructure also capture the dynamics of context data,
(e.g. system usage data); and (iii) by means of and hence to be able to trigger adaptive be-
direct updates of the context model. Typically, haviors, we also need to:

421
Context-Aware Applications 
for the Web

◦◦ Acquire context data by means • The navigation: the application may per-
of measures of real-world, physi- form automatic navigation actions on be-
cal properties, corresponding to the half of the user toward pages that better
properties of the context model. The suit the current context conditions.
so acquired data are then fed into the • The whole hypertext structure: the appli-
context model, so as to keep the con- cation may choose to apply coarse-grained
text model up to date. adaptations (e.g. to the layout of the appli-
◦◦ Monitor context data to detect those cation), for example to react to changes of
variations in context data that trigger the user’s device, role, or activity within a
adaptivity. Relevant variations are multi-channel, mobile environment.
used to enact the adaptation opera- • Presentation properties: the application
tions in the adaptive hypertext, thus may apply more fine-grained adjustments
causing an automatic, adaptive be- to the application’s appearance (e.g. to
havior of the Web application. style properties or fonts in use).
• Generic operations: the application may
While the definition of the context model decide to enact generic operations in the
and the monitoring of context data can easily background, e.g. to log specific applica-
be assisted by proper context modeling methods tion events or to interact with external
and a proper runtime framework providing basic applications.
monitoring functions, it is not as easy to assist
designers in the development of suitable context In this chapter, we will describe how these
acquisition (i.e. sensing) infrastructures. In fact, behaviors have been realized in the model-driven
the former two activities can be generalized beyond design method WebML and how the resulting
the needs of individual applications, while the extended version of the method can be leveraged
design of sensing infrastructures remains tightly for the development of context-aware applications.
coupled with individual application requirements Before proceeding with the discussion, it is thus
and technological choices. The exact development appropriate to shortly introduce the WebML de-
of sensing infrastructures is thus out of the scope velopment method, which will serve as reference
of this chapter. throughout this chapter.

Context-Aware Behaviors The Web Modeling


in Web applications Language (WebML)

But what exactly does it mean to adapt a Web WebML is a visual language for specifying the
application or to react to context? Starting from content structure of Web applications and the or-
the work by Brusilovsky (1996) on adaptive ganization and presentation of contents into one
hypermedia systems, in context-aware Web ap- or more hypertexts (Ceri et al., 2002).
plications, adaptive behaviors may affect: WebML application design starts with the
specification of a data schema, expressing the
• Contents and services delivered by the ac- organization of the application contents by
cessed pages: the application may autono- means of well established data models, such as
mously chose contents or services based the Entity-Relationship model or the UML class
on changing context data. diagram. On top of such data schema, WebML

422
Context-Aware Applications 
for the Web

design then proceeds with the specification of a detailed properties, not conveniently expressible in
so-called hypertext model, which describes how the graphic notation. The availability of the XML
contents, previously specified in the data schema, specification enables the automatic generation of
are published into the application hypertext. The the application code (Web Models, 2008), com-
overall structure of the hypertext is defined in prising rendering formats like HTML (which is
terms of site views, areas, pages, and content units. the standard choice for deployment) or WML. For
A site view is a hypertext, designed to address a a detailed description of WebML, the interested
specific set of requirements. Several site views reader is referred to (Ceri et al., 2002).
can be defined on top of the same data schema,
for serving the needs of different user communi-
ties, or for arranging the composition of pages to Modeling Data for Context-
meet the requirements of different access devices Aware Web Applications
like PDAs, smart phones, and similar appliances.
A site view is composed of areas, which are the Context data can derive from several sources
main sections of the hypertext, and comprise re- integrating sensed, user-supplied, and derived
cursively other sub-areas or pages. Pages are the information (Henricksen, 2004; Henricksen,
actual containers of information delivered to the 2002). While user-supplied data are generally
user; they are made of content units, which are the reliable and tend to be static, sensed data are
elementary pieces of information extracted from typically highly dynamic and can be unreliable
the data sources by means of queries, and published due to noise or sensor errors. The problem of
within pages. In particular, content units denote unreliability has been addressed in literature for
alternative ways for displaying one or more entity example by associating context information with
instances. Unit specification requires the defini- quality data (Lei, 2002). Although we recognize
tion of a source and a selector: the source is the the importance of reliable context data, in this
name of the entity from which the unit’s content work we rather concentrate on the exploitation
is extracted; the selector is a condition, used for of context in the design of Web applications.
retrieving the actual objects of the source entity For simplicity, throughout this chapter we thus
that contribute to the unit’s content. Content units consider sensed data as trustworthy.
and pages are interconnected by links to constitute
site views. Besides representing user navigation, Characterizing Context Data
links between units also specify the transportation
of parameters that can be used by the destination The main goal of context modeling is the formal-
unit in its selector condition. Some WebML units ization and abstraction of the context properties
also support the specification of content manage- that affect the application. In this regard, a first
ment operations. Standard operations are creating, characteristic distinguishing context properties
deleting or modifying an instance of an entity or is the distinction between physical and logical
adding or dropping a relationship between two context. We call physical context those proper-
instances; custom units may be defined. Finally, ties that are immediate representations of sensed,
WebML also allows the management of session physical quantities (e.g. the values of an analog/
parameters; parameters can be set and consumed digital converter), and logical context those prop-
through proper units. erties that enrich physical context with semantics
In addition to the visual representation, WebML and additional abstractions of the raw sensed data
also comes with an XML-based, textual repre- (e.g. the city corresponding to physical longitude
sentation, which allows one to specify additional and latitude values).

423
Context-Aware Applications 
for the Web

Figure 2. Persistence of physical and logical context data

A second characteristic affecting the structure history) typically require the persistent
of the context model is the persistence of context storage of data. Persistent physical context
properties in the system, i.e. the property that data are thus included in the context model
expresses whether individual context properties and updated according to their dynamics.
represent persistent data or volatile data. Persistent • Persistent logical context. Logical context
data need to be stored in the application’s data data is stored as data in the context model,
source and therefore require proper data entities so as to enable the data-driven transforma-
being modeled as part of the context model, while tion of physical context into logical context.
volatile data do not need any storage and can thus Logical context data are typically static, as
be omitted from the context model. The context they provide abstractions of physical con-
model therefore only captures persistent context text; dynamic updates and/or extensions
data (indeed, in WebML the context-model is part can, however, be supported as well.
of the database underlying the application).
Starting from these two characteristics and Physical and logical context data therefore
from the reference architecture introduced in coexist in the application’s data source. This co-
Figure 1, Figure 2 summarizes the resulting existence typically requires a transformation or
characterization of context data: mapping between raw data and information that
can directly be used when specifying hypertext
• Volatile physical context. Context data schemas. Consistently with the data-driven ap-
communicated via client-side parameters proach that characterizes WebML, we propose a
or via server-side session parameters rep- formalization of such transformation at the data
resent volatile data. They are immediately level by means of suitable associations between
available during the execution of the ap- data entities representing physical and logical
plication, independently of the underlying context data, respectively. Although technically
context model. Volatile context data are not legal, we do not expect the use of volatile logical
enclosed in the context model; they might context, as volatile context data typically repre-
however be used during page computation sents sensed raw context data.
to adapt the application. It is worth noting that even though there are
• Persistent physical context. Context data several properties commonly regarded as context
sharing (e.g. between members of a same attributes (e.g. position, time, or device charac-
group) or tracking (e.g. to derive differen- teristics), there exists no universal context model
tial context properties or to keep a context that applies to all kinds of applications. For this

424
Context-Aware Applications 
for the Web

reason, also in this chapter we do not prescribe any model of the application is represented
precise, rigid context model for WebML applica- by the entities Area, Building, and Road,
tions; we rather introduce some WebML-specific which all provide logical context data. The
modeling guidelines that enable the designer to actual GPS position data used for deliver-
provide context-aware applications with suitable ing the location-aware guide through the
context meta-data. Politecnico campus (i.e. longitude and lati-
tude) and the signal strength of the WiFi
Example Data Schema for connection are not part of the context
Adaptation in WebML model in the application’s data source; in
developing the PoliTour application, we
Let’s consider the PoliTour application shortly will handle such as volatile context data.
discussed in the introduction. Figure 3 illustrates Starting from the physically sensed data,
a possible Entity-Relationship diagram with basic the entity Area allows the application to
user profile data and context data, grouped in the identify a geographical area inside the
figure into so-called sub-schemas: campus; an area is then associated either
with a Building or a Road, meaning that
• User profile sub-schema. Users, groups, starting from the user’s position we can
and site views are represented as “first- identify whether he/she is located close to
class citizens” in the application data a building or rather walking through one of
source, as required by the WebML design the roads in the campus.
process. The entity User provides a basic • Application data. The remaining entity
profile of the application’s users, the en- Classroom represents application data that
tity Group associates access rights to users are not part of the context model. This
(i.e. a role), and the entity Site View con- means that from a building it is possible to
tains the site views that may be accessed access the list of classrooms of the build-
by the members of a group. The relation- ing, but there are no adaptive behaviors as-
ship Membership expresses that users sociated with the entity Classroom.
may belong to multiple groups, which in
turn cluster multiple users. The relation- Modeling Context-
ship DefaultGroup connects a user to his/ Aware Hypertexts
her default role and, when logging into the
application, the relationship DefaultSV al- While the first step of the WebML design method,
lows the application to forward the user to i.e. data modeling, does not require any exten-
his/her default group’s default site view. sion of the modeling primitives for capturing
The relationship Access expresses which context data (the standard Entity-Relationship
site views a specific group is allowed to ac- primitives suffice), WebML hypertext modeling
cess; this relationship is required as vary- does require a few model extensions to express
ing context conditions may require differ- adaptivity concerns. Next we therefore introduce
ent interaction and navigation structures the new concepts and primitives that have been
for a same group. In this way, depending developed to express adaptive behaviors, and we
on the context state, the application is able clarify how different adaptivity policies can be
to determine the most appropriate site view used to enact adaptations.
and to forward the user accordingly.
• Context model sub-schema. The context

425
Context-Aware Applications 
for the Web

Figure 3. Adaptation-triggering data in WebML applications, partitioned into basic user sub-schema,
personalization sub-schema and context sub-schema

Context-Aware Pages so-called “access pages” (e.g. containing cat-


and Containers egories or lists) typically are not affected by the
context of use.
Our basic assumption in the modeling of context- As can be seen in Figure 4, we tag context-
aware hypertexts is that context-awareness or aware pages with a C-label (standing for context-
adaptivity is a property to be associated only aware) to distinguish them from conventional
to some pages of an application (the adaptive pages. The label indicates that an adaptivity rule
hypertext), not necessarily to the application as a (stylized as a cloud) is associated with the page
whole. Location-aware applications, for example, and that during the execution of the application
adapt core contents to the position of a user, and this logic must be taken into account when com-

Figure 4. WebML hypertext schema with one context-aware site view and two context-aware pages. The
parameter P exemplifies the propagation of reusable context data by hierarchically passing context
parameters from an outer area to an inner page

426
Context-Aware Applications 
for the Web

puting the page. Specifically, Figure 4 states that The notion of context-aware container allows
pages Buildings and Roads are context-aware, us to define sparse adaptivity rules: we talk about
while the page Classrooms does not present any sparse adaptivity rules in those cases, where
adaptive behavior. adaptivity actions are associated to containers
There might also be the need for adaptivity rules that contain multiple pages; the scope of such
with effects that spread over multiple pages. For actions spans a set of pages, more precisely, all
this purpose, we exploit the hierarchical structure context-aware pages in the container. Coming back
of hypertexts; that is, we allow the definition of to the PoliTour application sketched in Figure 4,
context-aware containers (i.e. site views and areas, we can thus associate the logic to interpret the
in terms of WebML). This allows the designer to signal strength of the WiFi connection to the pages
insulate and to specify only once adaptivity rules Buildings and Roads by applying the logic to the
that are common to multiple C-pages inside a site view as a whole.
container and thus to reduce the redundancy of
the schema. Adaptivity rules associated to contain- Parameter Passing
ers and pages are evaluated recursively, starting
from the outermost container and ending with the Adaptivity logic is associated to a page by means
actual pages. The site view PoliTour in Figure 4 of a directed arrow, i.e. a link exiting the C-label.
is context-aware; we will see later on why. This link ensures the communication between the
page logic and the adaptivity logic: it may transport
Localized and Sparse parameters deriving from page contents, which
Adaptivity Rules may be used to compute the specified actions; in
turn, a link from the adaptivity logic to the page
The adaptivity rules attached to the context-aware may transport context parameters or generic val-
pages and containers in Figure 4 represent the ues that might be required to perform the final
actual adaptivity logic (i.e. the set of adaptivity adaptation during page computation.
actions to be performed). The adaptivity logic is But Figure 4 also illustrates the possibility of
external to the page or container, and the chain of hierarchically passing parameters from an outer
adaptivity actions it clusters is kept separate from container to an inner one. More precisely, if the
the page or container specification. The aim is to evaluation of outer adaptivity logic produces
highlight the two different logics deriving from results to be reused at an inner level, as it might
the role played by pages/containers and adaptivity happen in the case of context parameters, it passes
operations: while the former act as providers of such values back to the C-label that activated
contents and services, the latter act as modifiers the computation of the logic. Subsequently, such
of such contents and services. parameters can then be “consumed” by adaptivity
Adaptivity actions attached to a C-page typi- logics of the inner levels. As for context-aware
cally present effects that are visible in the page pages, parameter passing from a container to
they are attached to. The notion of context-aware its adaptivity logic occurs through the logic-
page and adaptation logic therefore defines what activating link. Links exiting the last evaluated
we call a localized adaptivity rule: the scope of a logic, i.e. at the end of the last adaptivity action,
localized adaptivity rule is strictly coupled with a might carry parameter values for the computation
fixed set of hypertext pages, where “scope” refers of units inside a page.
to those (adaptive) pages to which the page’s Typical actions to be specified at the container
adaptivity actions are associated. level are the acquisition of fresh context data and

427
Context-Aware Applications 
for the Web

the updating of the context model, e.g. if the data with the already available operations, provide the
are to be shared among multiple users or if a history necessary primitives for:
of context data is to be tracked. Hence, especially
if persistent context data are adopted, we propose • Specifying the acquisition of fresh context
two levels for adaptivity actions: data through client-side parameters. A
new Get ClientParameter unit (see Figure
• Actions for context model management, ad- 5) has been defined to support the retrieval
dressing operations for context data acqui- of parameters generated at the client side
sition and context model updating, should and communicated back to the application
be associated with outer containers (site via client-side parameters (e.g. parameter-
views or areas) and are inherited by inner value pairs attached to the page request
containers (areas or pages). These adap- query string).
tivity actions need to be executed prior to • Specifying the acquisition of fresh con-
the execution of any other action possibly text data through server-side parameters.
specified in an inner context cloud, as such Context data directly made available as
“internal” actions could depend on con- HTTP session parameters can be accessed
text data acquired and stored in the context by means of conventional WebML Get
model through “external” actions. units (Ceri et al., 2002).
• Actions for hypertext adaptivity, defining • Specifying the acquisition of context data
the rules for page and navigation adapta- from the context model. The execution of
tion (and possibly depending on persistent adaptivity actions may require the retrieval
context data), should be associated with and evaluation of context meta-data, for
C-pages. example, in situations where certain data
are just needed to evaluate condition ex-
Specifying Adaptivity Logics pressions. For this purpose, a so-called
Get Data unit (see Figure 5) has been in-
The main novelties for modeling context-aware troduced, enabling the retrieval of values
pages reside in the specification of adaptivity rules (both scalars and sets) from the data source
by means of WebML constructs. In the following, according to a selector condition. The se-
we introduce the new WebML modeling concepts mantics of the Get Data unit is similar to
that ensure full coverage for the specification of the one of content publishing units (Ceri
context model management and hypertext adapta- et al., 2002), with the only difference that
tion logics. The new primitives allow designers to data retrieved from the data source are not
visually specify actions for acquiring and updating published in hypertexts, but just used as in-
context data and to define adaptivity actions. put for units or operations.
• Updating the context model. Once fresh
Managing Context Data context parameters have been retrieved,
they can be used to update the context
In order to support adaptivity with respect context, model at data level. This action consists
the application must be able to acquire and manage in modifying values previously stored
context data according to the mechanisms illustrated in the data source. In WebML, this is al-
in Figure 1. For this purpose, some new WebML ready facilitated by operation units (Ceri et
operations have been defined, which, together al., 2002) providing support for the most

428
Context-Aware Applications 
for the Web

Figure 5. WebML units that have been defined for the specification of adaptivity actions

common database management operations adaptivity actions can be performed to adapt the
(e.g., modify, insert, delete). page contents, the navigation, the current site view
structure, and/or presentation style properties.
Evaluating Conditions These actions are specified as follows:

The execution of adaptivity actions may be subject • Adapting Page Contents. Page contents are
to the evaluation of some conditions, refining adapted by means of proper data selectors,
the triggering logic for context clouds. The most whose definition is based on context pa-
recurrent pattern consists in evaluating whether rameters retrieved from the context model
context changes demand for adaptation. The or newly computed within the page’s con-
evaluation of conditions is specified by means text logic. The use of parameterized selec-
of two control structures, represented by the If tors allows for both filtering data items
and Switch operation units, which have been with respect to the current context state
introduced for workflow modeling in WebML and conditionally including/excluding (i.e.
(Brambilla et al., 2003). showing/hiding) individual content units.
• Adapting Navigation. In some cases, the
Executing Adaptivity Actions effect of condition evaluation within the
context cloud can be an automatic, i.e.
Once the current context state has been determined, context-triggered, navigation action, caus-
and possible conditions have been evaluated, ing the redirection of the user to a different

429
Context-Aware Applications 
for the Web

page. The specification of context-triggered Change Style unit for dynamically assign-
navigations just requires connecting one of ing presentation style properties (see Figure
the links exiting the adaptivity logic of the 5). Style properties are collected in proper
page to an arbitrary destination page of the .css (Cascaded Style Sheet) files, and the
hypertext. Therefore, links exiting the con- unit enables the application to change its
text cloud and directed to other pages than associated style sheet at runtime.
the adaptivity logic’s source page represent • Enacting generic operations. The context-
automatic navigation actions. triggered invocation of generic operations
• Adapting the Site View. In some cases, a or, for instance, external Web services can
context-triggered switch toward a differ- easily be specified by placing the respec-
ent site view may be required. Changes in tive WebML operation unit into the page’s
the interaction context may in fact ask for adaptivity logic and by providing the unit
a coarse-grained restructuring of the whole with the necessary input parameters.
hypertext, for example because the user
device has changed, or because the user Triggering Adaptivity Rules
shifted to a different activity. To switch be-
tween different site views, we have intro- But when do we enact an adaptivity rule? In this
duced a Change Site View unit (see Figure regard, it is possible to define two different adap-
5), which takes in input the identifiers of tivity policies for context-aware pages, assigning
the target site view and the target page, to different priorities to users and context:
be visualized in case a switch toward the
specified site view is required. In order to • Deferred Adaptivity: the user is granted the
support “contextual” switching, the input highest priority. Therefore, after the user
link also transports parameters character- has entered the page and the page has been
izing the current state of interaction, i.e.: rendered according to the user’s selections,
1. The input parameters of the source the page’s adaptivity logic is evaluated at
page, which represent the last selec- periodic time intervals, enabling the ap-
tions operated by the user; plication to possibly adapt the already
2. Global parameters, representing ses- rendered page. Periodically evaluating the
sion data (e.g. user identifier and adaptivity logic means periodically refresh-
group identifier), as well as past user ing the page visualized in the browser.
selections that have been used for the • Immediate Adaptivity: context is granted
computation of the current page; the highest priority. The page’s adaptiv-
3. Client-side and server-side context ity logic is evaluated each time the page is
parameters retrieved during the lat- accessed, being the access due to the user
est performed data acquisition cycle or to the periodic refresh of the page. This
and characterizing the current context means that the page is subject to adaptation
state. each time it is rendered, even at the first
• Adapting Presentation Style. Sometimes time the page is accessed by the user.
context changes may require only fine-
grained adaptations of presentation prop- Consider for example our PoliTour guide that
erties (e.g. due to varying luminosity con- shows contents about the buildings and roads in
ditions), not a complete restructuring of the Politecnico campus. At a given point, the user
the overall hypertext. We have defined a might want to get information about a specific

430
Context-Aware Applications 
for the Web

building located in a road that is not related to policy, while inner adaptivity actions keep their
his/her current position; such a preference is deferred policy for front-end adaptations. The
typically expressed by selecting a link to that hierarchical definition of context clouds may
building from a list. With a deferred policy, the therefore also be considered a facility to achieve
requested page shows the building information as different “layers” of adaptivity actions.
requested by the user, without taking into account In our approach, we assume deferred adaptivity
the user’s current location. Only after expiration as default policy. This choice aims at minimizing
of the refresh interval, the page becomes subject application behaviors that might be perceived
to adaptivity and the contents are adapted to the as invasive or annoying by users and has been
user’s location. With an immediate policy, context experienced as the most natural for modeling ad-
is granted higher priority with respect to the user aptation. However, the immediate policy could be
and, thus, the user’s request for the building would needed for handling exceptional situations, as in
be overwritten by the context and the application such cases the timely reaction to context changes
would show the building or road associated to could be more important than following the user’s
the user’s current location, discarding the user’s indications. We therefore, in general, recommend
selection. the selection of the adaptivity policy that is ap-
Note that in addition to these adaptivity poli- propriate to the application requirements and that
cies, we recognize that there may be situations that is able to minimize the application behaviors that
demand for an explicit control of the adaptation could be perceived as invasive or annoying by the
dynamics through the user. Therefore, should for users. In order to choose the right adaptivity policy
example a user temporarily not be interested in for an adaptive page, a developer therefore needs
having the contents adapted to his/her location, he/ to predict what kind of adaptive behavior a user
she can simply disable/enable adaptivity at will. In will expect when accessing that page.
WebML, the adaptivity policy for context-aware
pages and containers is declared by means of the Example Hypertext Model
Adaptivity_Policy property of context-aware
pages and containers. Figure 6 shows the adaptive WebML hypertext
Adaptivity policies can also be associated model of the PoliTour application. The figure pro-
to context-aware containers. When a C-page is vides a refinement of the coarse hypertext model
requested, also the possible context clouds of its introduced in Figure 4 and details the internals of
containers are evaluated recursively (from the pages and adaptivity logics.
outermost one to the innermost one), according to The pages Buildings and Roads share the same
the adaptivity policy associated to each container. adaptivity logic providing location-awareness to
In general, a container’s adaptivity policy is in- the displayed contents. The logic starts with two
dependent of the policy of inner containers and Get ClientParameter units accessing the user’s
pages (if not, this must be taken into account by longitude and latitude, which are then used by
designers when associating policies to containers the Get Area unit to associate a logical area to the
and pages). Therefore, it may happen that the ac- user’s position. A further Get Data unit (the Get
tions in a container’s context cloud are evaluated Building unit) then tries to retrieve a building for
immediately, even if the actions associated to inner the identified area. If a building could be retrieved,
containers or pages adopt a deferred evaluation, or the If unit sends the user to the Buildings page,
vice-versa. If, for example, the adaptivity actions providing updated page parameters. If instead
associated to the container serve for tracking a no building could be retrieved (e.g. because the
context history, they could require an immediate user is located in the center of a road or not close

431
Context-Aware Applications 
for the Web

Figure 6. Hypertext model of the PoliTour application leveraging volatile context data

enough to a building), the If unit forwards the Area The Get RSSI unit accesses the volatile RSSI
identifier to the Get Road unit, which retrieves the parameter sensed at the client side, and the If unit
road associated to the current position. compares the retrieved value with a predefined
Therefore, if the user views the page Buildings level (alertLevel), below of which the connectiv-
while walking around the campus, the application ity is considered low. In case of low connectivity,
automatically updates the contents published each the style sheet warning is adopted; otherwise, the
time a new building can be found. If only the default style sheet is adopted. We therefore model
road can be identified, the application performs the alert of low connectivity conditions by means
an automatic navigation action toward the Roads of a Change Style unit: under low connectivity
page, where the described adaptive behavior starts conditions the application is rendered with a red
again, possibly causing the adaptation of contents background, under normal conditions the applica-
or automatic navigation actions. Only if the user tion is rendered with a gray background.
navigates to page Classroom, no adaptations are We recall that actions associated to containers
performed, as this page is not tagged as context- are evaluated before any action at the page level is
aware. started. Hence, in Figure 6 the actions associated
The adaptivity actions associated to the sur- to the site view are executed before the actions
rounding site view specify how to alert users associated to the pages Buildings and Roads.
who are about to leave the WiFi-covered area.

432
Context-Aware Applications 
for the Web

Runtime Context Model • On the other hand, generating page re-


Management quests allows the client to transmit client-
side sensed data, thus enabling the commu-
In order to manifest context-aware behaviors, nication of context data to the application
the application must be equipped with the capa- server.
bility to monitor the context state and to trigger
adaptivity actions, if required. The standard Context-aware pages are therefore also char-
HTTP protocol underlying most of today’s Web acterized by an individual refresh interval, which
applications implements a strict pull paradigm, in can be specified as property (Refresh_Interval) of
which computations can only occur in response to the page in the XML representation of the WebML
client-side generated page requests. Therefore, in model. Differently from C-pages, a container
the classical Web architecture, lacking proper push does not require the specification of any polling
mechanisms, context monitoring can occur only interval, which is instead derived from the interval
when a page is computed, i.e. when a respective associated to the currently viewed C-page of the
page request reaches the Web server. Three main container.
solutions can be adopted to trigger the evaluation
of adaptivity rules: (i) context evaluation on user- Context Monitoring
generated page requests, (ii) periodical, automatic
refreshes of viewed pages to enable context evalu- Context monitoring in the background (i.e. with-
ation, and (iii) active context evaluation to trigger out the user observing any unwanted rendering
adaptivity in real time. The first solution is not activity) enables the application to limit the use
able to cope with the dynamic nature of context. of the refresh to those situations that really ask
The periodic refresh of context-aware pages for adaptation and to perform context monitoring
provides a way to ensure the update of the page without any visual effect for users.
even in absence of explicit user actions enabling Figure 7 shows a functional architecture for
the re-computation of the page. In the following, adaptive Web applications that extends the de-
we will show an active mechanism for triggering scribed architecture of WebML applications (see
adaptivity, which operates independently of the Figure 1) with a new client-server module, called
user in the background and comes close to the Context Monitor (CM), providing the necessary
real-time triggering solution. context monitoring logic. As further depicted by
In absence of dedicated server-side push the figure, in case of client-side context sensing,
mechanisms for delivering updated pages, the the CM module also enables the communication
HTML http-equiv META-option or JavaScript, of client-side sensed context parameters, which
JavaApplets, and Flash scripts, provide valuable could be required at the server side to evaluate
client-side mechanisms to approximate the re- context changes and/or conditions over context
quired active behavior. The approximation is based parameters.
on periodic HTTP requests toward the application The CM consists of two separate modules,
server, which are operated in the background and one on the client side and one on the server side.
may serve a twofold purpose: The CM Client module is a piece of business
logic embedded into the page’s HTML code
• On the one hand, they provide the neces- and executed at the client side (e.g. a JavaScript
sary polling mechanism to query the con- function, a Java applet, or a Flash object), while
text model and trigger the adaptivity rule the CM Server module works in parallel to the
attached to the page. Web application on the same Web server. The CM

433
Context-Aware Applications 
for the Web

Figure 7. Functional architecture for background context monitoring

Client is in charge of periodically monitoring the individual page’s adaptive behavior is typically
context state and deciding whether possibly occur- influenced by only a subset of the overall context
ring context variations demand for the adaptation data or, more specifically, by a function expressed
of the currently viewed page. over context data. The subset of context data
In order to be able to take a decision about corresponds to a page-specific view over the
whether adaptivity actions are to be triggered or application’s context data, narrowing the focus
not, the CM Client is assisted by the CM Server, of the context monitoring activity. This observa-
which has full access to the context model of tion leads to the definition of a new concept, i.e.
the application maintained at the server side. In page context, which can be leveraged to enhance
response to the polling executed by the CM Cli- the efficiency of the context monitoring activ-
ent, the CM Server queries the context model and ity: the page context of a page corresponds to a
provides the CM Client with an updated picture page-specific view over the application’s context
of the effective context state. By comparing the data, capturing all (and only) those context char-
state of the (server-side) context model acquired acteristics that effectively determine the adaptive
by the current polling with the one acquired by behavior of the page.
the last polling (or the state at page computation Instead of monitoring the whole state of the
time), the CM Client knows whether the state has application’s context data, the definition of a
changed. If the state has changed, the CM Client page context for each adaptive page enables the
asks the Web application for a refresh of the cur- context monitoring activity to focus its observa-
rently viewed page, i.e. the adaptation; if the state tion of the context state to the only page context.
has not changed the CM Client proceeds with the This implies, that during hypertext specification
monitoring of the context state. each adaptivity rule can be related to a subset of
context parameters to be controlled, so that rule
Page Context conditions do not need to check the state of the
whole context model.
In general, the state of the context is expressed by
the values of all the persistent parameters stored in
the context model and of the volatile parameters
sensed at the client or server side. However, an

434
Context-Aware Applications 
for the Web

Page Context Parameters to the values of the page context parameters also
results in a change of the numeric digest. We call
In line with the idea of page context, the CM such a numeric digest context digest: the context
focuses its attention only to the subset of context digest corresponding to the page context of a
data in the context model that really determines page is the numeric checksum computed over the
the adaptive behavior of the viewed page. This ordered list of page context parameters.
implies explicit knowledge of the pages’ page The context digest is the basis for the decisions
context, which can be achieved by defining proper to be taken by the CM Client: its values identify
page context parameters for each context-aware variations in the page context, which correspond
page: page context parameters define the view to the need to adapt the page. The decision is
over the context model that captures all the static based on the comparison of the current context
and dynamic properties of a page’s page context digest with the last context digest; the first con-
by means of suitable queries over the context text digest, i.e. when the user accesses the page,
model. is initialized with the context digest valid during
This definition implies that each change to a page computation.
page context parameter effectively corresponds Figure 8 details the resulting flow of activities
to a need to adapt the page. The granularity of the enabling the active behavior of the application
values of page context parameters must thus be and shows how the single modules cooperate in
chosen in a way that each change of a parameter order to determine whether adaptivity is required
value translates into the triggering of the page’s or not. The diagram has one start node (Generate
adaptivity rule. Each C-labeled page in the adap- user request), which corresponds to the user’s
tive hypertext model is thus associated with an navigation to a C-page, and no end node, since
individual page context by means of proper page the cycle in the lower part of the diagram is only
context parameters stored in the textual repre- interrupted by an explicit user navigation leading
sentation of the WebML schema, as they are not the user to another C-page (which corresponds to
conveniently expressible in a visual manner. Page starting again from the start node of the diagram
context parameters are expressed by means of and monitoring the Page context of the new page)
parametric queries over the context data, where or to a conventional page (which does not cause
the parameters correspond to client- or server-side any context monitoring activity).
context parameters. Note that the described mechanism assumes
that connectivity is available during the viewing
Context Digest of a C-page in order for the CM client to be able
to communicate with the CM server. In case of
In order for the CM to be able to decide whether intermittent connectivity, which is a very frequent
adaptivity is required, changes to the page context situation in mobile environments, the CM client
(i.e. the page context parameters) must be com- keeps working by periodically polling the CM
municated from the CM Server to the CM Client. Server, despite the absence of connectivity. The
In order to enhance the efficiency of the overall CM Client is however programmed to manage
context monitoring activity, the state of the page possible lacks of connectivity and therefore does
context is not communicated from the CM Server not generate errors, with the only side effect that
to the CM Client in form of the set of page context adaptivity is suspended until the connectivity is
parameters, but instead it suffices to transmit and restored.
compare a numeric digest computed over the re-
spective page context parameters, as each change

435
Context-Aware Applications 
for the Web

Figure 8. Background context monitoring for active context-awareness (with client-side context sensing):
communicating context data and triggering adaptivity

Application Implementation power for the specification of adaptivity rules


could be preserved.
The extensions that have been introduced into the Figure 9 shows a screenshot of the WebRatio
WebML development method to cope with the tool at work. The figure shows the WebML hyper-
new requirements posed by context-awareness text model of the Buildings page of the PoliTour
and adaptivity in Web applications have been application, along with its adaptivity logic: two Get
implemented as prototype extension of the We- ClientParameter units access the GPS coordinates
bRatio CASE tool, the official WebML modeling and pass them to the C-label, which forwards them
tool, equipped with a powerful automatic code to the outer adaptivity logic (cf. Figure 6). Starting
generator. Due to implementation restrictions from the shown hypertext model, the PoliTour
imposed by the modeling tool, the implementa- application has been automatically generated
tion of the adaptivity logic slightly differs from on top of a J2EE platform. The configuration of
the models described in this paper (e.g. it was not the Context Monitor has been performed manu-
possible to implement context-aware containers ally. To access GPS position data, we leverage a
or to place all the adaptivity operations outside client-side Bluetooth GPS device, interfaced via
pages). Nevertheless, the described expressive the Chaeron GPS Library (http://www.chaeron.

436
Context-Aware Applications 
for the Web

Figure 9. The WebRatio CASE tool showing the hypertext model of the buildings page with respective
adaptivity actions and the generated PoliTour application running on a PDA

com/gps.html) and wrapped by means of Flash expressed by means of OCL-based customization


(to exchange position data between the CM Client rules, referring to UML class or package elements.
and the GPS library). The WiFi RSSI indicator Casteleyn et al. (2003) present an extension of
is acquired in the PDA using Place Lab (http:// WSDM (De Troyer & Leune, 1998) to cover the
www.placelab.org). specification of adaptive behaviors. In particular,
an event-based Adaptive Specification Language
(ASL) is defined, which allows designers to ex-
Related Works press adaptations on the structure and the naviga-
tion of the Web site. Such adaptations consist in
Several other well-established, conceptual design transformations of the navigation model that can
methods have been so far extended to deal with be applied to nodes (deleting/adding nodes), infor-
Web application adaptations. Frasincar & Houben mation chunks (connecting/disconnecting chunks
(2002), for example, extend the Hera methodology to/from a node), and links (adding/deleting links).
with two kinds of adaptation: adaptability with Baumeister et al. (2005) explore Aspect-Oriented
respect to the user device and adaptivity based on Programming techniques to model adaptivity in
user profile data. Adaptation rules (and the Hera the context of the UML-based Web engineering
schemas) are expressed in RDF(S) (Resource method UWE. Recently, WebML (Ceri et al.,
Description Framework/RDF Schema), attached 2002) has been extended to cover adaptivity and
to slices and executed by the AHA engine (De context-awareness (Ceri et al., 2007). New visual
Bra et al., 2003). The UWA Consortium proposes primitives cover the specification of adaptivity
WUML (Kappel et al., 2001) for conceptual rules to evaluate conditions and to trigger some
hypertext design. Adaptation requirements are actions for adapting page contents, navigation,

437
Context-Aware Applications 
for the Web

hypertext structure, and presentation. Also, the operations implementing these rules is provided.
data model has been enriched to represent meta- A notable feature, promoting portability, is that
data supporting adaptivity. GAC can be integrated as a stand-alone module
Recently, active rules, based on the ECA into any Web site architecture.
(Event-Condition-Action) paradigm, have been
proposed as a way to solve the previous problem.
Initially exploited especially in fields such as Conclusion and
content evolution and reactive Web (Alferes et al., Future Trends
2005; Bailey et al., 2002; Bonifati et al., 2002),
ECA rules have been adopted to support adaptivity In this chapter, we have proposed a model-driven
in Web applications. In particular, the specification approach to the development of context-aware
of decoupled adaptivity rules provides a way to Web applications, an increasingly relevant kind
design adaptive behaviors along an orthogonal of applications on the Web. We have shown that
dimension. Among the most recent and notable context-awareness is a first-class design concern
proposals, the work described in (Garrigos et al., that can considerably be aided by model-driven
2005a) enriches the OO-H model with personaliza- development techniques. But we have also shown
tion rules for profile groups: rules are defined in that properly dealing with context-awareness
PRML (Personalization Rule Modeling Language) and adaptivity at the conceptual level requires
and are attached to links in the OO-H Navigation extending the expressive power of the adopted
Access Diagram. The use of a PRML rule engine conceptual application model, so as to provide
is envisioned in (Garrigos et al., 2005b), but its developers with suitable modeling constructs and
real potential for adaptivity management also at implementation abstractions, proper of such new
runtime remains unexplored. class of application features. In this chapter, such
The previous works benefit from the adoption extensions have been introduced into the already
of conceptual models, which provide designers well-established WebML modeling language,
with powerful means to reason at a high-level but in a similar way we could have also opted
of abstraction, independently of implementa- for another modeling language, as the ideas and
tion details. There are however also co-called concepts introduced in this chapter are general
transcoding solutions, which adopt active rules enough in nature to be applied to other conceptual
for adapting Web pages. Most of them focus on models as well.
the presentation layer and provide mechanisms For the future, we believe that a decoupled
to transform HTML pages according to (possibly runtime management of adaptivity features will
limited) device capabilities (Hori et al., 2000) or represent a next step in the area of adaptive Web
users’ visual disabilities (Yesilada et al., 2004). applications. The development of Web applica-
Moreover, they typically support only adaptability tions is more and more based on fast and incre-
and modify Web pages in relation to a static set mental deployments with multiple development
of user or device parameters. Fiala and Houben cycles. The same consideration also holds for
(2005) adopt the transcoding paradigm for the context-aware and adaptive Web applications and
development of the Generic Adaptation Com- their adaptivity requirements. In (Daniel et al.,
ponent (GAC). GAC provides a broad range of 2008) we describe our first results obtained with a
adaptation behaviors, especially supporting run decoupled environment for the execution and the
time adaptivity. An RDF-based rule language administration of adaptivity rules. The described
is used for specifying both content adaptation approach allows us to abstract adaptive behaviors,
and context data update rules. A collection of to extract them from the main application logic,

438
Context-Aware Applications 
for the Web

and to provide a decoupled management support, Brambilla, M., Ceri, S., Comai, S., Fraternali, P.,
finally enhancing the maintainability and evolv- & Manolescu, I. (2003). Specification and Design
ability of the overall application. of Workflow-Driven Hypertexts. Journal of Web
In line with the current hype of so-called Web Engineering, 1(2), 163–182.
2.0 applications, we are also working on the
Brusilovsky, P. (1996). Methods and Techniques of
mash-up of context-aware Web applications, in
Adaptive Hypermedia. User Modeling and User-
the context of our component-based development
Adapted Interaction, 6(2-3), 87–129. doi:10.1007/
method for Web applications called Mixup (Yu et
BF00143964
al., 2007). The final goal of the work is to enable
even end users to mash up their own context-aware Casteleyn, S., De Troyer, O., & Brockmans, S.
applications, starting from a set of so-called con- (2003). Design Time Support for Adaptive Be-
text components and other components equipped havior in Web Sites. In SAC’03 (pp. 1222-1228).
with own user interface (which is used to build Melbourne, Florida: ACM.
up the user interface of the mash-up application).
Ceri, S., Fraternali, P., Bongio, A., Brambilla,
Mash-up development is assisted by an easy-to-use
M., Comai, S., & Matera, M. (2002). Designing
and intuitive graphical development environment
Data-Intensive Web Applications. San Francisco,
that supports a drag-and-drop development and
CA: Morgan Kauffmann.
by a light-weight runtime environment that is
able to interpret and run the mashup, both fully Ceri. S., Daniel, F., Matera, M., & Facca, F. M.
running in the client browser and based on AJAX (2007). Model-driven Development of Context-
technology. aware Web Applications. ACM Transactions on
Internet Technologies, 7(1), article no. 2.
Daniel, F., Matera, M., & Pozzi, G. (2008).
References
Managing Runtime Adaptivity through Active
Alferes, J. J., Amador, R., & May, W. (2005). A Rules: the Bellerofonte Framework. Journal of
General Language for Evolution and Reactivity in Web Engineering, 7(3), 179–199.
the Semantic Web. In PPSWR’05 (pp. 101–115). De Bra, P., Aerts, A. T. M., Berden, B., de Lange,
Dagstuhl, Germany: Springer. B., Rousseau, B., Santic, T., et al. (2003). AHA!
Bailey, J., Poulovassilis, A., & Wood, P. T. (2002). The Adaptive Hypermedia Architecture. In Hyper-
An Event-condition-action Language for XML. text’03 (pp 81-84). Nottingham, UK: ACM.
In WWW’02 (pp. 486–495). Honolulu, Hawaii: De Troyer, O., & Leune, C. J. (1998). WSDM:
ACM. A User Centered Design Method for Web Sites.
Baumeister, H., Knapp, A., Koch, N., & Zhang, Computer Networks, 30(1-7), 85-94.
G. (2005). Modelling Adaptivity with Aspects. Dey, A. K., & Abowd, G. D. (2000). Towards a
In ICWE’05 (pp. 406-416). Sydney, Australia: Better Understanding of Context and Context-
Springer. Awareness. In CHI’00 Workshop Proceedings,
Bonifati, A., Braga, D., Campi, A., & Ceri, S. The Hague, The Netherlands.
(2002). Active XQuery. In ICDE’02 (pp. 403-412).
San Jose, California: IEEE.

439
Context-Aware Applications 
for the Web

Fiala, Z., & Houben, G.-J. (2005). A generic Kappel, G., Pröll, B., Retschitzegger, W., &
transcoding tool for making web applications Schwinger, W. (2001). Modelling Ubiquitous
adaptive. In CAiSE’05 Short Paper Proceedings, Web Applications - TheWUML Approach. In
volume 161 of CEUR Workshop Proceedings. ER’01 Workshops (pp. 183-197). Yokohama,
CEUR-WS.org. Japan: Springer.
Frasincar, F., & Houben, G.-J. (2002). Hypermedia Lei, H., Sow, D. M., Davis, J. S. II, Banavar, G.,
Presentation Adaptation on the Semantic Web. In & Ebling, M. R. (2002). The design and applica-
AH’02 (pp. 133-142). Málaga, Spain: Springer. tions of a context service. SIGMOBILE Mobile
Computing and Communications Review, 6(4),
Garrigós, I., Casteleyn, S., & Gómez, J. (2005a).
45–55. doi:10.1145/643550.643554
A Structured Approach to Personalize Websites
Using the OO-H Personalization Framework. Schilit, B. N., & Theimer, M. M. (1994).
In APWeb’05 (pp. 695-706). Shanghai, China: Disseminating Active Map Information to
Springer. Mobile Hosts. IEEE Network, 8(5), 22–32.
doi:10.1109/65.313011
Garrigós, I., Gómez, J., Barna, P., & Houben, G.-
J. (2005b). A Reusable Personalization Model in Web Models, l. (2008). WebRatio Site Develop-
Web Application Design. In WISM’05 (pp. 40-49). ment Studio. Retrieved January, 2008, from http://
Sydney, Australia. www.webratio.com.
Henricksen, K., & Indulska, J. (2004). Modelling Yesilada, Y., & Harper, S. Goble. C. A., & Stevens,
and Using Imperfect Context Information. In R. (2004). Screen readers cannot see: Ontology
PERCOMW’04 (pp. 33-37). Washington, United based semantic annotation for visually impaired
States: IEEE. web travellers. In ICWE’04 (pp. 445-458). Mu-
nich, Germany: Springer.
Henricksen, K., Indulska, J., & Rakotonirainy,
A. (2002). Modeling Context Information in Yu, J., Benatallah, B., Saint-Paul, R., Casati, F.,
Pervasive Computing Systems. In Pervasive’02 Daniel, F., & Matera, M. (2007). A Framework for
(pp. 167-180). London, UK: Springer. Rapid Integration of Presentation Components. In
WWW’07 (pp. 923-932). Banff, Canada: ACM.
Hori, M., Kondoh, G., Ono, K., Hirose, S., &
Singhal, S. K. (2000). Annotation based Web
Content Transcoding. Computer Networks, 33(1-
6), 197-211.

This work was previously published in Context-Aware Mobile and Ubiquitous Computing for Enhanced Usability: Adaptive
Technologies and Applications, edited by D. Stojanovic, pp. 59-82, copyright 2009 by Information Science Reference (an
imprint of IGI Global).

440
441

Chapter 2.11
Different Web Strategies for
Different E-Marketplaces
L. Geppert
Catholic University of Milan, Italy

Abstract in two well-defined supply chains, that is, the “steel”


chain and the “gold and silver” chain.
This chapter presents two possible models of elec-
tronic marketplaces put in place at the beginning
of this century, which, after their introduction, the Introduction
first incoming wave of connected economy-based
paradigms was ended. Both the two models show After the first incoming wave of the connected
a particular use of Web-based information technol- economy ended, investors, business managers,
ogy in order to exploit their mission and represent consultants, and specialists began to review their
meaningful cases of application of well-defined Web plans, actions, and methodologies. Many e-business
strategies. Even though, at the moment, the initially models, both business-to-business (B2B) and
built Web sites supporting those related business business-to-consumer (B2C), are currently under
have been closed and merged with other Web sites, consideration in order to assess and verify their
they may introduce examples of a Web strategy real effectiveness (Kalakota & Whinston, 1996).
approach having a relevant historical meaning that The models applied so far have generally shown
may be still redefined in practical implementations that, with some exceptions related to e-procurement
once revised and adequately updated. The mentioned implementations of large enterprises (Timmers,
cases described in this chapter are usteel.com and 1999), vertical marketplaces owners must look for
up2gold.com, two examples of Web-based business different business paradigms in order to economi-
cally survive and to be actually able to create value
DOI: 10.4018/978-1-60566-024-0.ch006 to market industries chains they address (McDonald,

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Different Web Strategies for Different E-Marketplaces

1997). In the following pages, two real cases are an aberrant commencement to the development
described: one trading marketplace addressing of the connected economy that will emerge in
the steel industry as an example of “first wave” e- the 21st century. Businesses interact with other
marketplace and one supply chain-like Web portal businesses in hundreds of ways. It is true that
built for the Arezzo’s (Italy) gold manufacturing the interactions are generally intended to result
supply chain. The first case relates to a “traditional in a commercial transaction. However, for many
Web strategy”-based e-marketplace with a stan- business relationships, the crucial performance-
dard approach and contents (Shapiro & Varian, enhancing interactions take place before and after
1998) and the second shows an innovative strategy the transactions, and B2B will expand to occupy
and a new business model (Porter, 1985). Even these broader horizons. (Kyte, 2001)
though, at the moment, both of the abovementioned
e-marketplaces have merged their activities and Business managers’ and investors’ expecta-
brands in other larger e-commerce initiatives, at tions were very high at the end of the last century.
the time they were presented (2001), they repre- They understood the strategic importance of B2B
sented important and innovative examples of an electronic model in particular and invested their
e-marketplace-based business model. money and their commitment, demanding rapid
deployment of innovative business systems, which
the IT community was increasingly hard-pressed
Background to deliver. Many of these systems were new breed
of applications, designed to change the external
As some analyst observes, “the first high-visibility relations in the B2B world.
Internet businesses conceived to attract signifi- These systems work fine on the drawing board.
cant investor attention were, at the beginning, in However, actually developing externally focused
the business-to-consumer (B2C) relationships, applications as robust, reliable, secure systems
providing facilities for consumers to choose and with high availability and very high adaptability
purchase goods online. It seemed logical to think, was and still currently is a challenge that only few
as a related consequence, a development also of IT specialists know how to address.
B2B, based on the “buying and selling” part of Furthermore, models of marketplaces so far
the relationship between businesses. implemented put the stress and focus of their
Thus, starting in 1999, and accelerating rapidly business model on the transactional aspect. This
through 2000, the focus of B2B attention was very model, which can act and succeed—or, better,
firmly on those initiatives that seemed to be capable survive—in all those case where the liquidity is
of exploiting information and communication reached in a sensible spell of time, does not take
technology to alter the balance of power in trading into account a natural evolution of electronic
relationships. These initiatives, which covered a ecosystems where the complexity of relationships
wide spectrum of business and technical services, is becoming more and more vital.
gained the general title of “marketplaces,” and, However, the connected economy and B2B
throughout 2000, the focus of B2B activities was marketplace models with it have just started their
on developing marketplace interaction strategies evolution and growth, which will not be stopped;
and examining procurement processes. they just need some adaptation mechanism to
The focus on the transactional elements of B2B consolidate the new paradigm.
relationships will come to be seen as Thus, after that, a first pioneer period went
by, where marketplace models have attracted
business managers and, consequently, finance

442
Different Web Strategies for Different E-Marketplaces

Figure 1.

investors, showing the possible strategic value • Quality and quantity of services can ag-
of Web technology, that is, the ability to extend gregate other partners, which can bring ad-
the enterprise business model and to improve ditional financial resources and customers,
process efficiencies; now that paybacks have not in order to give more value to the industry
been respected and investors begin to retire their they address.
support, the marketplace model must be readapted • Processes begin before transactions and
and reengineered (Bollier, 1996). continue after it; support must be given to
In order to better respect investors’ and busi- a wider range of customer tasks.
ness managers’ constraints and expectations, that
is, to present a more realistic behaviour of the In the next pages two significant real imple-
promised return of investment, it is now crucial mentation cases have been considered. The first
for marketplace business models to take into ac- concerns the possible evolution of a typical ver-
count the following aspects: tical marketplace positioned in the steel supply
chain. The second shows how a service center
• Membership and brand is more important for a geographical industrial district populated
than transaction fee (Dibb, Simkin, Pride, by many fragmented small enterprises can give
& Ferrel, 1991). value to the related supply chain and presents

Figure 2.

443
Different Web Strategies for Different E-Marketplaces

a valid paradigm to overcome possible “digital In order to be successful, this new initiative
divide”-based approaches. required three elements: high level stockholders,
capable of ensuring neutrality and independence to
the marketplace; managerial staff knowledgeable
Main Thrust of the Chapter in both the steel and the information technology
industries; and finally, the most advanced technol-
Case Study 1: u-steel.com1 ogy available (Ansoff 1987; Hammer, 1996).
The idea was presented to a number of potential
General Aspects partners, all of whom accepted the project with
great enthusiasm, starting from Iperbusiness of the
U-Steel is the first next generation e-marketplace, TC Sistema Group, who became the technology
neutral, independent, and completely online, to partner, taking responsibility for the coordination
be dedicated to the commerce of steel and related of the project and the system integration require-
services. Initially launched in September 2000 by ments.
the strong sponsorship of a group of European U-Steel has chosen a business model mainly
partners such as COFACE (F), Falck Group (I), based on the concept of membership, despite other
Mastr@ (I), Iperbusiness (I), Sopaf (I) and Mittel initiatives based solely on the transaction mecha-
Generale Investimenti (I), U-Steel’s aim is to bring nism. In an industry like steel-working, which only
the most qualified players of the steel industry in Europe claims a turnover of approximately 285
online, increase the consumers purchasing power, million Euros, U-Steel’s further objectives is to
and expand the markets. bring together partners who have a technologically
From an operational perspective, U-Steel is an innovative offering, placing particular emphasis on
e-marketplace integrated with a large number of the technological brand.
support services. Strategically, however, it is not The marketplace starts off as a platform on which
just a virtual channel between buyer and seller but a number of additional services can be included;
also an important tool to allow all supply chain the objective is to increase the offering based on the
participants to become a part, easily and safely, of efficiency of the cycle with other services ranging
the steel’s industry future dynamics. from insurance, credit, and logistics, all the way
All effective members of the marketplace who to integrating the partner’s IT infrastructures with
purchase steel through U-Steel are certified and a technology across platforms.
assigned a credit rating by COFACE, a worldwide U-Steel’s transactional system is globally among
leader in credit certification. Sellers who require the most advanced, with a variety of exchangeable
total guarantee of their credits towards other U-Steel products including plate, coil, rod, wire, pipe, and
members can take out a credit insurance policy special irons and steels. Additionally, there is the
offered by COFACE exclusively to the portal’s possibility of marketing excess lots or scrap lots
subscribers at extremely favourable terms. for the plants.
When in 1999, as a group of friends, the founders When a purchaser wishes to find a potential
came up with the idea of starting a company with seller, they choose the product to buy, select the
the purpose of building an e-marketplace for steel, supplier, and can contact them through the mar-
they immediately understood the huge importance ketplace to establish prices and terms by means
of the project. Shifting to “virtual” steel meant of an interactive chat.
entering a dimension completely new to the steel This way U-Steel becomes the point of contact
industry, traditionally tied to the business rules of between purchaser and seller, the location where
the so-called “old economy.” the negotiation is conducted and finalized, regard-

444
Different Web Strategies for Different E-Marketplaces

Figure 3.

less of where in the supply chain the negotiation to the retailers, from service centers who purchase
takes place. The model allows considerable large amounts and resell smaller lots of semifin-
savings in terms of time, cost, and back-office ished product, to the retailer or storekeeper who
infrastructure. distribute the product to end users.
But the portal is not only about buying and U-Steel does not promote disintermediation.
selling; this is just the initial step of an online On the contrary, it encourages the repositioning
process to which it will soon be possible to add of sales figures who previously benefited from the
additional services such as invoicing, logistics, inefficiencies of the supply chain and for whom it
warehousing, management and handling, account- is now possible to become value added consultants.
ing, and credit management. A trading tool only adds another level of competi-
Overall, a system capable of handling all tion within the procurement process.
processes not strictly connected to the steel’s A seller approaching U-Steel is given the
core business. opportunity of publishing the seller’s catalogue
Who are the players interested in accessing the directly on the marketplace, complete with prices
marketplace? The system potentially involves all and current available stock, or with the option of
components of the cycle, from the manufacturers showing only the products without any further

Figure 4.

445
Different Web Strategies for Different E-Marketplaces

Figure 5.

information. In both cases, the subscriber is given ers, a Cisco Local Director 416 firewall, and a
the possibility of integrating the U-Steel catalogue Cisco PIX 525 in failover. The entire system was
with the seller’s own information system so that designed to provide the highest level of availabil-
each transaction results in a warehouse operation, ity, and with doubled security and load balancing
a production cycle advancement, or a purchase systems, it was built to handle an initial volume
order. of 1.200 users in its first months.
Alternatively, the manufacturer may simply Furthermore, HP specialists handle all the sys-
replicate a portion of their product catalogue, tem resource administration activities ensuring the
put it online, and choose when to synchronize users’ seamless and transparent functionality.
the two systems.
The technological components was chosen, The E-Business Model
developed, and implemented by Iperbusiness
on the BroadVision application. Hewlett Pack- The business model of u-steel.com is based the on
ard was chosen as outsourcer of all the systems typical characteristics of the European steel supply
management services, including the environment chain and on the model of a B2B marketplace.
infrastructure (i.e., premises, power supplies and The European steel supply chain is formed by
surge suppressors/generators, security systems, a very limited number of large steel producers and
fire prevention, local area networks, etc.) and the brokers, which are positioned at the top links of the
required hardware platform housed in the HP Data chain, and by a great deal of other smaller players,
Centre located in Bergamo (I). grouped in dealers, steel service centers (SSC),
The system, designed and commissioned in and final companies which transform and use
just 4 months time, is based on a UNIX HP UX steel components in their own production. Even
operating system, is configured on two HP9000 though this supply chain includes the presence of
A400 clustered Web servers, two HP9000 A500 to large producers, brokers, and even buyers (e.g.,
run the BroadVision applications, and an HP9000 large automotive industry enterprises), the typi-
A500 box acting as DB server with the Oracle cal power play is well balanced, in that, medium
database. The information is stored on two HP and small companies can usually buy steel also
SureStore ESC10 clustered devices with an HP oversee, that is, in America and Asia.
SureStore E 2/20 backup library. Thus, u-steel.com, which includes among its
The network is based on two Cisco 2900 rout- partners ex-business managers of the steel chain

446
Different Web Strategies for Different E-Marketplaces

Figure 6.

who know very well the market and the industry Steel Service Centres and small dealers,
behaviour, is positioned as a vertical, neutral, which can reach a larger audience of final
independent, aggregating, and collaborative elec- transformers.
tronic market maker for steel trading exchanges • Reducing buying cost, in term of final
and other services, in order to satisfy the needs of price, for small and medium transformers
all the players and to support all the competitive that can increase their ability to compete.
forces of the steel supply chain. • Reducing stocking and work-in-progress
Thus, first of all, u-steel.com is a vertical costs of the whole supply chain through a
marketplace, which implements typical processes better synchronization of production, dis-
of a well identified market and industry where tribution, and delivery.
trading activities are accomplished in a particular
context. Furthermore, taking into account the peculiar-
It is neutral, in that it is neither “seller” or ity of the European steel supply chain, U-Steel
“buyer” advocated. It offers to both the groups of will offer in the immediate future a very sophis-
players an electronic trading place where it enables ticated auction-based system, which will give the
secure commercial transactions and other services. opportunity to large producers (i.e., steel mills a
Buyers and sellers can access the site and operate brokers) to launch forward actions on the European
it using the typical trading approaches: catalogue market and which, at the same time, will provide to
searching and sourcing, online and even interactive small dealers and transformers with a mechanism
exchange, and offer-posting processes. of reverse actions to buy at better conditions also
The expected macro-benefits produced through on oversee marketplaces.
these approaches can be grouped in the following U-steel is also an independent electronic market
main classes: maker, in that it lies in a well-balanced power play
position. Its claimed mission is “acting as infor-
• Improving efficiency of intercompany op- mediary on the steel market,” with no particular
erational processes by reducing the typical preference for any side.
time usually needed to manually accom- The historical role of shareholder partners en-
plish the same tasks. sures both the knowledge of the steel supply chain,
• Enhancing commercial visibility and ex- through the Falck Group currently involved in the
ploitation of resources, particularly for utilities industry, and the absolute transparency

447
Different Web Strategies for Different E-Marketplaces

Figure 7.

and independence, being the others well known the steel business community.
players of other industries. There are two types of services that U-Steel
But U-Steel is an aggregating and collabora- delivers: business services and technological
tive market maker as well. These issues represent services.
the innovative aspects of the U-Steel business Business services include
model.
In fact, after the “hype cycle” of the connected • Credit insurance (already delivered at the
economy has shown failures and constraints of an moment)
expected “no limits” growth, U-Steel proposes a • Logistics delivery
further possible direction in a marketplace model Technological services include
evolution. U-steel offers other services by ag- • Process integration
gregating other business partners and delivering • Platform outsourcing
new functionalities which create new value for

Figure 8.

448
Different Web Strategies for Different E-Marketplaces

Credit insurance gives the opportunity to sell- possibility to reduce this payback time and thus
ers to insure their credit, following a credit rating to shrink its ROI, while other players receive the
mechanism classified and tuned by Coface (F), value to have their own environment, classified
one of the worldwide insurance market leader and as a variable cost in their profit and loss account.
one of the U-Steel shareholder. Through these In other words, U-Steel can reduce the initial
functionality, the seller can decide to insuring trade-off between revenues and costs and the other
the seller’s credit and thus electronically activate players using the feature, use the platform as a
the operation with an automatic interaction with general expenditure, and do not have to directly
the Coface credit system, in order to assess and involve themselves in technology, which is not
possibly accept economic conditions, as proposed their core business.
by Coface. The seller pays a further transaction In this sense, the U-Steel model appears as a
fee for that. vertical model for trading and commercial opera-
Logistics delivery, which is not currently tions, and, at the same time, as a horizontal model
available, but it will be in the future, will give for distributing technological skills and services;
the possibility to the buyer to choose a possible a sort of a hybrid, new model, where U-Steel can
logistics and transportation player to be electroni- act as a business service provider as well. A model
cally involved in the process in order to deliver that, taking into account the current evolution of
goods in a specified spell of time. other vertical marketplaces in other industries,
Process integration means the availability of can be considered a valid evolution and a possible
integrating, when needed, the information system successful paradigm.
of both seller and buyer, in order to automatically
accomplish several collaborating processes, such The Revenue Lines
as procurement, fulfilment, billing, and invoicing
through an XML interface. It means also the pos- The U-Steel model is built on the following eco-
sibility to be integrated with other marketplaces nomic revenue lines:
supporting the same feature in order to extend the
catalogue interaction to other “ecosystems.” • Membership fee, which is an annual fee to
The most innovative collaborative aspect of be paid by sellers and buyer in order to op-
U-Steel is the availability of a technological plat- erate on the marketplace.
form, which can be rented by supply chain players. • Transaction fees on goods, as a percentage
The technological platform on which U-Steel is of the total amount.
based is a very flexible architecture provided by • Transaction fees on services delivered by
BroadVision and Oracle “best of breed” compo- other partners (credit and logistics).
nents. It ensures the possibility to build separate • Annual fee for ERP integration, following
and independent environments where different defined technological standards.
players can implement their corners. In this way, • Annual BSP fee, which is a yearly based
a steel mill company or a broker, for example, fee for using the platform hosted by u-steel
can also deliver their own extranet, outsourced and adequately customized on customer
to U-Steel, avoiding the investment of a huge needs.
amount of money in a capital expenditure which • Advertising.
would ask them for large financial resources and
would force them to track a control on the return U-Steel will involve more then 6,000 European
of investment. users after 3 years and will trade more then 5 mil-
In this model, on the contrary, U-Steel has the lions of steel tons in the same period (3% of the

449
Different Web Strategies for Different E-Marketplaces

whole European market). Expected revenues will The final production output is finished goods
be over 40 Millions Euro at the third year. like gold chains, rings, and ear-rings. The product
The www.u-steel.com site, which is currently is considered to be medium quality goods, usually
available for test purposes to a limited community competing with Indian and Far East equivalent
of “family and friends” players of the steel supply productions.
chain, will be launched and delivered worldwide Such a supply chain, despite its pervasive
on June 20, 2001. presence in many other sectors, is obviously inef-
ficient, for the following reasons:
Case Study 2: up2gold.com2
• The main obstacle to a more efficient chain
General Aspects is, as said before, the dimension of the com-
pany involved, the number of companies,
Up2Gold.com is a service center conceived to and the related fragmentation.
deliver consulting and technological services to • Moreover, the efficiency of the chain seems
the supply chain of gold and silver manufacturing to be influenced by the flow of information
located in Arezzo (I). Arezzo is one of the most regarding stock or availability of goods or
important industrial sites for gold and silver manu- raw material. In fact, while the commercial
facturing, a typical Italian industrial district which side of the chain (the end) does not know
geographically groups many SMEs belonging to the availability, prices, quality, and other
the same supply chain. information, the producers do not know the
real needs of their markets in order to be
At the starting point of the gold (and silver) sup-
able to adequate their production volumes,
ply chain there are the raw material suppliers, that
stiles, and so forth.
is, larger companies which are in a limited number,
• The third type of inefficiency is linked to the
while production plants are divided into a plethora
sales forecast or production schedule, which
of small and very small companies linked together
are extremely difficult to govern, because of
by complex multidirectional channels. a very fragmented catalogue of products; is
The distribution layer is, again, a “long chain” very common the case of a catalogue with
where various intermediaries and dealers are in- more than 10,000 articles, and practically
volved before reaching the end user. every company has a catalogue of thou-
These large companies control the supply links sands of articles. Moreover, many articles
and schedule main production flows, while large are subject to season waves, other are linked
dealers mainly govern the sales channel and, in to styles and stylists.
particular, the export market.
As mentioned before, the productive part of the The result of these inefficiencies are, of course,
supply chain is very fragmented and powdered in a poor service at the end user level (e.g., unavailability
myriad of SME, involved in production subphases of products in certain period of the year) or, which
mainly as third parties. As it is usual in these cases, is worst due to the cost of raw gold, obsolescence
they do not participate to the research and develop- of the entire stock with the consequent inefficiency
ment process; they just execute and manage orders at the producer level. This also causes lack of
received from other supply chain links. competitive presence in those markets where the
These enterprises are often familiar companies, newcomers are getting more and more aggressive
involving even less than 10 persons and unable to due to a lower production cost and a more flexible
base their activities on some information system, supply chain.
which would be useless yet. In order to better understand the subject, the

450
Different Web Strategies for Different E-Marketplaces

Table 1.

Site 1-9 workers 10-49 workers 50-99 workers 100-999 workers Total
Arezzo 991 269 7 4 1271
Companies 3366 4547 423 1279 9615
Workers

Arezzo’s gold and silver supply chain is described The shareholders of the Up2Gold are:
in Table 1.
Arezzo’s supply chain shares the 20% of that • Arezzo’s and National artisans associa-
global export, that is, more than 1.500 billion ITL. tions (CAN).
The total Arezzo’s gold and silver production is • TC Sistema SpA and Datamat SpA, two fa-
more than 6.000 billion ITL. mous IT players of the Italian market;
The aim of up2gold.com is to offer a class of • FlyNet, a local ISP.
services to overcome the inefficiencies described • Banca dell’Etruria e del Lazio, which is a
above through the deployment of the a service local bank.
center able to deliver:
General concepts and strategic issues of
• Consultancy services to help managing the center have been designed in the following
personnel, legal, and work trends issues; terms:
• ERP services, with particular stress on an
accounting and payroll application. • The marketplace model appears innovative
• A B2B supply chain marketplace to ex- and suitable to fit chain needs and SMEs’
change raw material, semifinished goods, expectations.
and human resources work availability • It can be considered as a new paradigm for
and to interconnect along the chain goods’ supply chain local systems and exported
availability and stocks. and disseminated to other SMEs districts;
• A B2C marketplace, sell side-oriented, a real B2B best practice.
in order to increase sales of medium size
companies.

Table 2. The total Italian production export of gold and silver goods is shown.

Region Production export 1989 Production export 1999


(billions ITL) (billions ITL)
Europe 1173 (28.15%) 2323 (28.76%)
North America 1684 (40.41%) 2694 (33.36%)
Latin America 175 (4.20%) 634 (7.85%)
Middle East 468 (11.23%) 747 (9.24%)
Far East 379 (9.10%) 860 (10.65%)
Others 288 (6.91%) 818 (10.13%)
Total 4167 (100%) 8076 (100%)

451
Different Web Strategies for Different E-Marketplaces

The E-Business Model • Procuring human resources or other small-


er production companies.
When considering the business model of up2gold. • Procuring semifinished goods, if any, from
com, it is necessary to underline that the company other chain players.
is a service center which delivers a set of services • Producing components as ordered.
supporting a supply chain. It means that the center • Delivering goods.
offers traditional B2B services along with “con-
nected economy”-like functionalities. The value of information in accomplishing
Yet, even though the business model of the these activities is very high; if existing, it would
center is very articulated and all of its components reflect in a larger economy of scale, and in a bet-
should be considered and analyzed in a broader ter stock and delivery efficiency. Sometimes it
perspective, for the purposes of this chapter we happens that sale channels must refuse oversee
would like to draw the attention to the market- orders, in that they are not able to declare on time
place part, which is an innovative B2B model a right delivery date and final prices to customers,
and includes elements of as the supply chain for a lack of updated information!
management approach. These activities are not supported by an inte-
When addressing this marketplace, it is neces- grated information system and this fact, as already
sary to take into account typicality and behaviours mentioned above, generates inefficiency to the
of the supply chain to which it belongs. As al- whole network chain.
ready mentioned before, the Arezzo’s gold chain Up2gold.com is building a marketplace which
is a very fragmented network of small suppliers offers two important services to the companies:
all working for a few larger enterprises, which
act as raw material suppliers and distribution • A typical exchange marketplace in order
channels (Fariselli, Oughton, Picory, & Sugden, to accomplish the first three activities as
1997). These companies control the chain and described above: procuring raw material,
play a strong power role: they make the research semifinished goods, and human resources.
and development, that is, design gold and silver This marketplace gives raw material sup-
final products, schedule large scale production pliers the ability to exchange gold and sil-
(distribution and sale channels, in particular), and ver and give other enterprises the possibili-
control the raw material procurement process (raw ty to buy possible semifinished goods from
material suppliers). They usually involve the other others’ stocks and/or human working re-
links of the chain as third parties, requesting them sources availability. An additional auctions
to produce finished or semifinished goods by a and reverse auctions system is included in
well defined delivery date (make-to-order). order to sell stocks or buy components at a
This links, on their turn, involve other links more convenient price.
of the supply chain, ordering subcomponents and • A collaborative system to support a supply
other semifinished goods, scheduling delivery chain management marketplace. Through
dates, and so on along the whole supply chain. this remote system, accessed via Web, any
All the subproducers links must generate by and company, even small enterprises, can give
large the following tasks: their availability-to-promise, by claiming
their actual stock availability, production
• Procuring raw material, buying it in gen- capacity and, of course, prices. This sys-
eral from large raw material suppliers. tem can give the sale channel an immediate

452
Different Web Strategies for Different E-Marketplaces

Figure 9.

picture of both the order progress status Italian industrial chains. Up2gold.com named this
and the possible make-to-order availability model partner chain management (PCM).
of the whole supply chain.
The Revenue Lines
The model that up2gold.com implements is
not exactly that of a supply chain management The Up2Gold model is built on the following
marketplace, in that production orders scheduling economic revenue lines:
is not taken into account. It is a sort of warehouse
and stock availability integration, useful to catch • Membership fee, which is an annual fee to
up-to-date information before and after the order be paid by any company in order to access
is assigned. “Before the order” means that the the center services.
sale channel can assess and verify possible online • Transaction fees on goods, as a percentage
availability along the whole supply chain and pos- of the total amount.
sible delivery dates to communicate to customers. • Annual fee for consulting and ERP func-
“After the order” means that the sale channel and tion access.
other links of the chain can assess online the real • Advertising.
progress status of orders, controlling the actual
delivery date. Up2Gold will involve more then 1.200 district
This model is very simple in that no informa- companies after 3 years; the value of transaction
tion system is required for SMEs; they are only involved will be more then 1.000 billion ITL.
requested to keep their order profile updated by The current site is only for presentation and
entering simple information at the up2gold.com advertising purposes. The service center will
site. release its services progressively, starting at the
Up2gold.com conceived this marketplace beginning of 2002. The marketplace in particular
model in order to match actual requirements of will be launched and delivered by the first quarter
SMEs-based supply chains, that is, the typical of 2002.

453
Different Web Strategies for Different E-Marketplaces

Figure 10.

Future Trends and order to make their brand and their prod-
Conclusion ucts more globally known as possible.
Through this mechanism, SMEs will also
The B2B e-marketplaces will represent in the fu- implement collaboration tools and instru-
ture the new efficient paradigm that will be used to ments in a sort of “partners’ chain net-
accompany all trading process concerning goods, work,” where different facilities, such as
commodities in particular. They will be the “low marketing services, information systems,
cost” arenas where large companies will invite administrative, and financial services, can
their provider to submit proposals at the best prices. be shared at lower and more competitive
But they will also be the “low cost” arenas where internal prices by creating at the same time
SMEs will present their products and their brands both efficiency and new dimensions of
in order to be known and contacted by a larger and effectiveness.
larger audience. E-marketplaces will more likely
act per industry, instead of being “cross-industry,
general purpose,” and will show the two natures
as previously mentioned, that is: References

• Organized and driven by a large enterprise Ansoff, H. I. (1987). Corporate strategy (2nd ed.).
that will use the e-marketplace in order to Harmondsworth: Penguin.
attract a great number of suppliers, glob- Bollier, D. (1996). The future of electronic com-
ally widespread, and able to enhance their merce. Queenstown, MD: The Aspen Institute.
good value proposition. This will ensure a
strong competition and low supply prices. Dibb, S., Simkin, L., Pride, W., & Ferrel, O. (1991).
• Organized and driven by communities (as- Marketing, concepts and strategies. Boston:
sociations) particularly composed by small Houghton Mifflin.
enterprises that will group their efforts in

454
Different Web Strategies for Different E-Marketplaces

Fariselli, P., Oughton, C., Picory, C., & Sugden, Shapiro, C., & Varian, H. R. (1998). Information
R. (1997). Electronic commerce and the future for rules: A strategic guide to the network economy.
SMEs in a global marketplace: Opportunities and Boston: Harvard Business School Press.
public policy requirements. Brussels: CEC.
Timmers, P. (1999). Electronic commerce: Strat-
Geppert, L. (2001, December). E-Marketplaces: egies and models for B2b trading. New York:
New models. Paper presented at the First Policy Wiley.
Group Meeting of DEEDS Project, Brussels
Hammer, M. (1996). Beyond reengineering.
London: Harper Collins Business. Endnotes
Kalakota, R., & Whinston, A. (1996). Frontiers 1
At the moment, u.steel is not longer directly
of electronic commerce. Reading, MA: Addison- operating.
Wesley. 2
At the moment, Up2gold.com is not longer
Kyte, A. (2001, February). Paper presented at the directly operating. It has been included in
iEB Europe Gartner Conference 2001, London. another larger initiative.

McDonald, M. H. B. (1997). Marketing plans:


How to prepare them, how to use them. Oxford:
Butterwoth-Heinemann.
Porter, M. (1985). Competitive advantage. New
York: Free Press.

This work was previously published in Business Web Strategy: Design, Alignment, and Application, edited by L. Al-Hakim and
M. Memmola, pp. 118-132, copyright 2009 by Information Science Reference (an imprint of IGI Global).

455
456

Chapter 2.12
Developing Rule-Based
Applications for the Web:
Methodologies and Tools
Vassilis Papataxiarhis
National and Kapodistrian University of Athens, Greece

Vassileios Tsetsos
National and Kapodistrian University of Athens, Greece

Isambo Karali
National and Kapodistrian University of Athens, Greece

Panagiotis Stamatopoulos
National and Kapodistrian University of Athens, Greece

Stathes Hadjiefthymiades
National and Kapodistrian University of Athens, Greece

Abstract The chapter deals with both ways that have been
formally defined for modeling a domain of inter-
Embedding rules into Web applications, and dis- est: the first based on standard logics while the
tributed applications in general, seems to constitute second one stemmed from the logic programming
a significant task in order to accommodate desired perspective. Furthermore, a comparative study that
expressivity features in such environments. Various evaluates the reasoning engines and the various
methodologies and reasoning modules have been knowledge representation methodologies, focusing
proposed to manage rules and knowledge on the on rules, is presented.
Web. The main objective of the chapter is to sur-
vey related work in this area and discuss relevant
theories, methodologies and tools that can be used 1. INTRODUCTION AND MOTIVATION
to develop rule-based applications for the Web.
Nowadays, with the evolution of traditional web
of documents to a more complex web of services,
DOI: 10.4018/978-1-60566-402-6.ch016

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Developing Rule-Based Applications for the Web

an increasing demand for embedding intelligence classical logic. As a consequence, they differ
to Web applications arises. In this context, the ef- from recent Semantic Web technologies in many
ficient management of knowledge seems to play aspects, including representational features and
a key role in order to achieve smart behavior of reasoning functionality, as well. Hence, building
Web applications and to overcome several issues a rule-based application for the Web with existing
of such environment (e.g., information integra- rule technologies is not a straightforward task.
tion). Ontologies, mainly written with Semantic In the rest of this chapter, we provide founda-
Web technologies, constitute a well-established tional knowledge on this topic together with imple-
paradigm for representing knowledge on the Web. mentation issues, techniques and design patterns.
Though, current efforts are focused on extending Section 2 briefly describes how the things have
ontologies with more expressive forms of knowl- gone so far in the area of Web knowledge represen-
edge like rules. In fact, given the state-of-the-art in tation formalisms. In Section 3, various knowledge
the realization of the Semantic Web vision, rules representation methodologies and tools are dis-
constitute the next prominent challenge. Since the cussed. Specifically, Section 3.1 demonstrates the
ontology layer of the Semantic Web architecture different languages and formalisms, derived from
stack has reached a sufficient degree of maturity both classical logic and logic programming view,
through Web Ontology Language (OWL) (Dean while Section 3.2 focuses on various engines able
et al., 2004), the next step of progress involves the to reason over such knowledge bases. Section 4.1
integration of rules with ontologies, most of them gives the main requirements for rule-based web
based on subsets of First Order Logic (FOL). applications. The evaluation presented includes
Rules are capable of extending the expres- both a qualitative comparison (Section 4.2) of the
siveness provided by ontology languages through existing approaches and a performance analysis
the definition of more complex relationships (Section 4.3) of current ontology reasoners and
between individuals. Additionally, as a modular rule engines. Finally, several future trends and
form of knowledge, they fit well in domains like open issues are identified in Section 5. Hence,
personalization, policies and business-to-business this chapter aims at becoming a helpful guide for
(B2B) interaction. However, it has been shown applying rules to Web applications.
that extending ontologies even with simple forms
of rules can lead to undecidability of key infer-
ence problems. 2. THE STORY SO FAR
On the other hand, many business-logic ap-
plications have extensively taken advantage of The knowledge representation languages proposed
existing rule management systems or solvers (see Section 3.1) for representing knowledge on
(Jess, 2008; ILOG, 2008; Drools, 2008), aiming the Web are based either on the Classical Logic
at facilitating the knowledge management process. (CL) perspective or on Logic Programming1 (LP).
As a result, the success of rules in non-Web appli- As a result, a debate was started between the
cations moved Web researchers to use traditional Database community and AI researchers, respec-
rule engines on the Web. tively, in order to determine the more suitable of
However, the aforementioned stable rule sys- the two approaches in the formalization of Web
tems have not been originally created for open knowledge. Additionally, different languages of
and heterogeneous environments like the Web. the same perspective, providing various degrees
Such platforms have adopted different knowl- of expressivity, have been proposed. Hence, the
edge representation formalisms, mainly based integration of knowledge with Web applications
on principles of logic programming, instead of was more complicated. Recently, with the evolu-

457
Developing Rule-Based Applications for the Web

Figure 1. a) Tim Berners-Lee Semantic Web layer “cake” and b) the latest form of Semantic Web stack
diagram (adapted from W3C Semantic Web Activity, 2008)

tion of Semantic Web technologies, these modeling Semantic Web stack in order to take advantage
paradigms have been extensively discussed by of important rule languages based on logic pro-
researchers (Motik, Horrocks, Rosati, & Sattler, gramming. However, in (Horrocks, Parsia, Patel-
2006; Patel-Schneider & Horrocks, 2006; Eiter, Schneider, & Hendler, 2005) the authors criticize
Ianni, Polleres, Schindlauer, & Tompits, 2006a; the aforementioned approach and denote that such
Donini, Lenzerini, Nardi, & Schaerf, 1998; An- a distinction would lead to two Semantic Webs
toniou, et al., 2005; Boley, Kifer, Patranjan, & based on different semantics. These discussions
Polleres, 2007). along with the Semantic Web Activity workgroup
Clearly, this confrontation also affected rules. have led to several modifications of the stack.
Although, significant efforts have been devoted Figure 1 presents the initial and the current form
in order to develop appropriate rule languages of the Semantic Web stack diagram specified by
for the Web (see Section 3.1), a debate between the W3C Semantic Web Activity Group (W3C
proponents of the different perspectives has ap- Semantic Web Activity, 2008).
peared. The main argument involves the degree The aforementioned facts gave rise to confu-
that rules would be combined with ontologies. sion inside the Web community. If the experts
Initially, as formulated by Tim Berners-Lee, (i.e., the logicians stemming from both AI and
it had been commonly accepted that Semantic databases) could not decide and recommend a
Web should be structured in a hierarchy of lay- unified framework for formalizing knowledge,
ers that seamlessly interoperate (Antoniou & how Web developers and users should choose the
van Harmelen, 2004). However, some proposals best language according to their needs? The cur-
(mainly stemmed from the database community) rent solutions are either to study the capabilities
preferred to keep the rules and the ontology layer provided by all these languages (i.e., available
separated, in order to preserve expressiveness and expressiveness, support by existing reasoning
decidability of reasoning process, as mentioned engines, tractability etc.) or to exclude knowledge
in (Eiter et al., 2006a). Specifically, in (Kifer, de from their applications.
Bruijn, Boley, & Fensel, 2005) the authors argue
that the rules layer of Semantic Web architecture
should be placed next to the ontology layer of the

458
Developing Rule-Based Applications for the Web

Figure 2. Architecture of a Description Logics knowledge-based system

3. KNOWLEDGE REPRESENTATION A DL-based knowledge base is composed of


METHODOLOGIES AND TOOLS two components: TBox and ABox. TBox contains
the vocabulary of the application domain, called
This section surveys different knowledge repre- terminology, as well as axioms based on that
sentation methodologies along with reasoning vocabulary. Practically, such vocabulary con-
modules that have been proposed for managing sists of concepts and roles. Concepts are generic
knowledge on the Web. After an overview of De- descriptions of sets of individuals, while roles
scription Logics that comprise the main formalism constitute binary predicates for defining proper-
for representing ontological knowledge in Web ties of the individuals. On the other hand, ABox
applications, we focus on rules. includes assertions of individuals that may refer to
either concepts or roles. For example, a statement
3.1 Knowledge Representation declaring that a specific individual is instance
Formalisms of a concept resides in ABox, while a statement
denoting that “every human is mortal” belongs to
Description Logics TBox. Figure 2 shows the generic architecture of
a DL knowledge representation system.
Description Logics (DLs) (Baader, Calvanese, Each description logic language is determined
McGuiness, Nardi, & Patel-Schneider, 2003) by a set of constructs, enabling the use of atomic
are subsets of first-order-logic, originating from concepts and atomic roles in order to define
semantic networks and frame-based systems. They complex ones. These constructs directly affect
constitute a family of knowledge representation the expressive power of the language and, thus,
languages that aim at providing well-understood the complexity of inference tasks. As a result,
mechanisms in order to formalize knowledge that the selection of the appropriate description logic
describes a domain. This way, DLs are equipped language in order to describe a specific domain
with formal, logic-based semantics, emphasizing includes the examination of the imposed require-
on reasoning process. Typical reasoning tasks in- ments for representation expressiveness.
clude consistency checking of the knowledge base, DLs set up the base for the definition of Re-
concept satisfiability, instance checking etc. source Description Framework (RDF) (Klyne &

459
Developing Rule-Based Applications for the Web

Caroll, 2004), RDF Schema (Brickley & Guha, tification restrictions). For example, Description
2004) and Web Ontology Language (OWL) (Dean Logic Programs do not cover negation in class
et al., 2004) that constitute W3C standards for descriptions nor fully support cardinality restric-
representing knowledge on the Web. Specifi- tions.
cally, RDF is a simple data model for represent-
ing information on the Web. RDF statements Semantic Web Rule Language
are expressed in the form of (subject, predicate,
object) triples. A set of such statements can be Semantic Web Rule Language (SWRL) (Horrocks
viewed as a graph where the subjects and the et al., 2004b) is probably the most popular formal-
objects of these statements constitute the graph ism in Web community for expressing knowledge
nodes while the predicates correspond to the graph in the form of rules. Specifically, SWRL is based
edges. RDFS is an extension of RDF for express- on a combination of Web Ontology Language
ing simple taxonomies through the definition of (OWL) (Dean, 2004) and Rule Markup Language
class/property hierarchies and domain/range of (RuleML) (RuleML, 2008) and has been proposed
properties. OWL is totally based on DLs and it as a W3C candidate standard for formalizing the
comes in three species: OWL-Lite, OWL-DL and expression of rules in Web context. Contrary to
OWL-Full. OWL-Full is the most expressive OWL DLP, SWRL extends OWL-DL with a specific
species, since it takes advantage of all the OWL form of Horn-like rules.
language primitives. OWL-DL limits OWL-Full The proposed rules are in the form of an
to a subset of OWL primitives so as to achieve implication between the body and the head of
efficiency of reasoning. Finally, OWL-Lite is a the rule. A typical SWRL rule can be of the fol-
sublanguage of OWL-DL by restricting more its lowing form:
expressiveness.
a1 ∧ a2 ∧ … ∧ an → b1 ∧ b2 ∧ … ∧ bm
Description Logic Programs
(2)
The expressiveness of Description Logic Pro-
grams (DLP) approach (Grosof et al., 2003) where ai and bi are OWL atoms of the follow-
corresponds to a fragment of OWL defined by ing forms:
the expressive intersection of Description Logics
and logic programming. This approach intends to • Concepts, e.g., C(x), where C is an OWL
define a mapping from DL to logic programming description, in general, and x is either a
(specifically, Horn programs in which no function variable, an OWL individual or a data
symbols, negation and disjunction are allowed) value.
and vice versa. An instance of such a mapping in • Object properties, e.g., P(x,y), where P is
the case of conjunction follows: an OWL property and x, y are either vari-
ables, individuals or data values.
C1  C2  D ≡ D(x) ← C1(x)∧ C2(x) (1) • Datatype properties, e.g., P(x,y), where P
is an OWL property, x is variable or indi-
However, for the sake of decidability, DLP vidual, while y is a data value.
offers limited expressiveness, since the aforemen- • B(x1,x2,…), where B is a built-in relation
tioned mapping covers only a few DL-constructs and x1, x2, … are either variables, individ-
(in particular, conjunction, disjunction and quan- uals or data values.

460
Developing Rule-Based Applications for the Web

• sameAs(x,y) or differentFrom(x,y) where uses specific namespaces of OWL-S ontology


x, y are either variables, individuals or data like “profile” and “process”.
values.
Answer Set Programming
The main advantage of SWRL is the simplicity
it offers, while extending the expressiveness of Answer Set Programming (ASP) (Gelfond &
OWL. Another benefit of SWRL is its compatibil- Lifschitz, 1991) is a paradigm for knowledge
ity with OWL syntax and semantics, since they are representation and declarative programming. It
both combined in the same logical language. On has several advantages compared to other logic
the other hand, extending OWL-DL with SWRL programming paradigms (e.g., Prolog) such as:
rules leads to undecidability of simple inference
problems. A possible solution of this problem is • Full declarativity: the order of rules in a
presented in (Motik, Sattler, & Studer, 2005) which program is not important.
introduces the notion of DL-safe rules. Specifi- • ASP programs are in general decidable
cally, that approach restricts the application of • Non-monotonic inference: both negation
SWRL rules only to individuals of the ABox part of as failure (NAF) and strong negation are
the DL knowledge base. Moreover, SWRL does not supported, thus enabling default reason-
support negation (neither classical nor negation as ing and reasoning under the Closed World
failure - NAF) and disjunctions. Finally, there is no Assumption (CWA).
efficient support of first-order provers to execute • Availability of efficient solvers: there are
reasoning over SWRL. Usually, the SWRL rules several ASP solvers that are scalable enough
are translated to existing rule systems (e.g., Jess to deal with large knowledge bases.
(O’Connor, Knublauch, Samson, & Musen, 2005))
that handle the reasoning tasks partially, since they A general ASP rule is of the following form:
are not aimed to manage knowledge expressed in
terms of first-order logic or subsets. a1 ∨ a2 ∨ … ∨ an ← b1 ∧ … ∧ bk ∧ not bk+1 ∧ …
In the case of a Web service composition ∧ not bm
paradigm, some SWRL example rules could be
the following: (4)

profile:hasOutput(?S1,?out) where ai and bj are literals (atoms or strong


∧ profile:hasInput(?S2,?in) ∧ negations of atoms) and not denotes NAF.
process:parameterType(?in, ?cin) ∧ A set of such rules is an ASP program. What
process:parameterType(?out, ?cout) is interesting is the fact that these rules can
∧ (rdfs:subClassOf(?cout, ?cin) ∨ have disjunctions in their head. This is a very
rdfs:subClassOf(?cin, ?cout)) → important feature of ASP, since it introduces
composableWith(?S1, ?S2) non-determinism in the inference process (i.e.,
an ASP program may have several models which
(3) are called answer sets).
In order to be able to use ASP on the Web, the
The abovementioned rule captures the knowl- ASP rules should be combined with Web knowl-
edge that a web service S1 is composable with a edge. Since ontologies is the most common way
service S2, if an input of S1 is either subclass or to represent knowledge on the Web, an interaction
superclass of at least one output of S2. This rule between ASP programs and ontologies is deemed

461
Developing Rule-Based Applications for the Web

necessary. A solution to this integration problem These languages are based on several different
is description-logic programs (or dl-programs). formalisms such as Description Logics, Logic
These consist of ASP rules that may contain Programming and First-Order Logic. Some of the
queries to DL knowledge bases. For example, basic variants of WSML are the following:
the following rule “brings” into the ASP program
all instances of the class MovieTitle in the DL • WSML-Core: A subset of a Description
knowledge base: Logic which falls inside the Horn logic
fragment of FOL. It supports subsumption
movie(X) ← DL[“Movie”](X). (5) reasoning and query answering.
• WSML-Flight: An extension of WSML-
Several extensions to dl-programs were pro- Core which also supports full Datalog
posed (and implemented) so that they become rules, default negation and integrity con-
more “suitable” for open environments like straints. It can provide query answering in
the Web, where information may be expressed the context of Logic Programming.
in many diverse ways (e.g., multiple different • WSML-Rule: An extension of WSML-
ontologies). The most well known extension Flight with support for function symbols
is HEX-programs (Eiter, Ianni, Schindlauer, & and unsafe rules.
Tompits, 2005), which enable handling knowl-
edge expressed in various formalisms, even with Reasoning for these WSML variants can be
potentially different semantics (e.g., RDF(S) and implemented by several Logic Programming en-
OWL). HEX-programs contain several features gines. For some features of these languages, DL
(e.g., higher-order logic features) that enable more reasoners can also be used or First Order theorem
flexible integration with external knowledge bases. provers. These languages have been extensively
These extensions result in new syntax elements. used in several European projects, mainly in the
For example, the atom &rdf[u](s,p,o) evaluates application domain of semantic web services.
to true if <s p o> is an RDF triple asserted at URI Hence, several APIs, tools and other facilities
u. Finally, another advantage of HEX-programs are available for building WSML-enabled ap-
is that they allow using external data processing plications.
services that logic programming cannot handle
(e.g., string processing). An interesting engine Defeasible Rules
for HEX-programs is dlvhex2 which is described
in Section 3.2. Defeasible logic (Nute, 1994) is a rule-based, non-
There have been proposed ASP-based several monotonic approach able to deal with incomplete
applications in the context of (Semantic) Web. One knowledge and inconsistencies. These features
of the most promising is Web service composition have been widely remarked in the context of real-
(Rainer, 2005). The authors apply ASP techniques izing Semantic Web vision, mostly in information
to “build” service compositions from available integration areas (e.g., ontology merging). As a
services that match a certain service request. result, some efforts in research community (An-
toniou, Billington, Governatori, & Maher, 2001)
Web Service Modeling were devoted to carry the advantages of defeasible
Language (WSML) logic in the area of Semantic Web technologies.
The main idea behind defeasible logic rea-
WSML (de Bruijn et al., 2005) is a language of soning systems is the ability to handle a number
representation languages for the Semantic Web. of additional features with regard to classical

462
Developing Rule-Based Applications for the Web

rules like priorities of rules, default inheritance, this strict separation concerns the rule predi-
exceptions, etc. There are three different types of cates and the ontology elements. Hence, the
rules in a defeasible logic reasoning system: a) vocabulary (concepts and properties) offered
classical rules (called strict rules), b) defeasible by the ontologies is used as a conceptualiza-
rules that can be contradicted by other rules and c) tion of the domain and rules cannot directly
defeaters used to specify exceptions of defeasible define ontology classes or properties. Many
rules. This way, an important aspect achieved integration approaches adhere to this cat-
by such reasoning modules is their capability of egory, including Answer Set Programming
resolving the possible conflicts that arise among (Gelfond & Lifschitz, 1991), dl-programs
defeasible rules. (Eiter et al., 2004) and DL+log (Rosati,
2006c).
Classification of Approaches
Integrating Ontologies and Rules The user may find more details on this topic
in (Eiter et al., 2006a).
Although several approaches have been discussed Table 1 summarizes the aforementioned
for combining rules with Semantic Web ontologies knowledge representation languages and their
(Horrocks et al., 2004b; Grosof, Horrocks, Volz, basic features.
& Decker, 2003; Eiter, Lukasiewicz, Schindlauer,
& Tompits, 2004; Bassiliades, Antoniou, & Vlaha- 3.2 Reasoning Engines
vas, 2006; Rosati, 2006a; Rosati, 2006b), there is
no totally accepted solution in the field. The main In this section we intend to provide the reader
topic of argumentation is the degree of integration with a comparative feature analysis of existing
between the ontology layer and the rules layer. reasoning modules, including description logic
In this section we intend to provide a brief clas- reasoners and rule engines, as well.
sification of the proposed approaches. Jena2 (McBribe, 2002) is the second generation
Two main categories of integration approaches of Jena Semantic Web programming toolkit, which
have been distinguished in this context: is a Java framework for developing applications
based on Semantic Web technologies. Specifi-
a. Homogeneous approaches. These ap- cally, Jena provides an Application Programming
proaches suppose a tight semantic integration Interface (API) for creating, storing, managing
of the two layers. Specifically, both ontolo- and querying RDF graphs as well as RDFS,
gies and rules are embedded in a common OWL ontologies in various formats (RDF/XML,
logical language, permitting predicate shar- N3 and N-triples). The RDFS reasoner included
ing in a coherent way. In such approaches, in Jena framework does not support datatypes
ontology concepts and properties may be and blank node entailments. The built-in OWL
defined through rules. The most typical reasoner is very limited, since it is a rule-based
homogeneous paradigm is the combination implementation of OWL-Lite. However, Jena is
of SWRL rules with OWL ontologies. This supplied with an interface which facilitates the
is also the most In addition, Description connection and interoperation of the framework
Logic Programs (DLP) (Grosof et al., 2003) with any external reasoner that supports the DIG
constitutes another similar approach. (DL-Implementation Group) standard (Bechhofer,
b. Hybrid approaches. These approaches Moller, & Crowther, 2003). Hence, the API pro-
correspond to a strict semantic separation vided by Jena could be integrated with most of the
between ontologies and rules. In particular, existing description logics reasoners. Furthermore,

463
Developing Rule-Based Applications for the Web

Table 1. Basic features of various knowledge representation languages

Feature Logical Serialization


Decidability*
Language Foundation formats
OWL-Lite: decidable
Classical Logic XML/
OWL OWL-DL: decidable
(FOL subset) N-triples (textual)
OWL-Full: undecidable
OWL
Classical Logic
+ undecidable XML
(FOL subset)
SWRL
textual serialization (in terms
DLP DL and LP intersection decidable
of rules)
Extension of LP (disjunction in textual serialization (in terms
ASP decidable
rule heads, DL queries) of rules)
DLs + Datalog rules (disjunc- textual serialization (in terms
DL+log decidable
tive, non-monotonic) of rules)
Note: *regarding key inference problems (e.g., consistency of the knowledge base)

Jena provides a query engine in order to execute ontology engineering (Parsia, Sirin, & Kalyanpur,
SPARQL (Prud’hommeaux & Seaborne, 2005) 2005). Although typical reasoners detect the in-
queries over RDF graphs. consistencies between concepts of the knowledge
RacerPro system (RacerPro, 2008) is the com- base (KB), Pellet can explain why a concept
mercial3 extension of Racer (Haarslev & Moller, description led to unsatisfiability. This way, the
2001), which is probably the most popular reason- reasoner provides user with additional knowledge
ing engine for OWL ontologies to practitioners of (e.g., relevant axioms or restrictions) sufficient
Semantic Web technologies. RacerPro can be seen to understand the problem and reform the KB
as a knowledge-based repository that can handle properly. Finally, Pellet allows ontologies to use
and it is a system for managing OWL ontologies, XML-Schema built-in and user-defined datatypes
in particular. It implements the DIG interface that extend numeric and date/time types.
and it offers an optimized tableau calculus for Bossam (Jang & Sohn, 2004) is a RETE-based,
the description logic SHIQ(D). Additionally, it forward-chaining reasoning module for reasoning
supports qualified cardinality restrictions as well and querying over RDF(S) and OWL documents,
as some extensions of OWL (e.g., OWL-E (Pan while it also supports rules execution (SWRL rules
& Horrocks, 2004) except user defined XML are included). It is based on Logic Programming
datatype expressions). Moreover, the latest ver- (LP), augmented with some expressiveness fea-
sion of RacerPro includes a first implementation tures stemming from First-Order-Logic (FOL). For
of an SWRL rule engine. example, a number of additional to LP features are
Pellet (Sirin, Parsia, Grau, Kalyanpur, & Katz, provided by Bossam, including support for both
2007) is a Java-based, open source reasoner ca- classical negation and NAF and disjunctions in the
pable of handling expressive OWL ontologies. body of rules. Hence, it does not support complete
It implements an optimized tableau algorithm, reasoning over OWL Ontologies. Furthermore,
augmented with a number of additional features it facilitates the integration of rules with Java by
(e.g., support for Unique Name Assumption - supporting a procedural attachments mechanism
UNA, closed world reasoning, SPARQL query for SWRL rules. Finally, Bossam provides an API
answering). Pellet also provides an explanation for managing the engine, loading ontologies and
service in order to facilitate the debugging of the rules, querying RDF(S)/OWL documents and giv-

464
Developing Rule-Based Applications for the Web

ing explanations about derived facts. Currently, programs, which are an extension of Answer
Bossam does not support SPARQL query answer- Set Programs towards integration of external
ing, while the serialization of the knowledge base computation sources. dlvhex can communicate
to a persistent store (e.g., file system) is another with OWL and RDF knowledge bases and can
missing feature. also return the results in RuleML syntax. A strong
FACT++ system (Tsarkov & Horrocks, 2006) point of dlvhex is that it enables developers to
is the descendant of FACT (Horrocks, 1998). write and embed plug-ins to the core engine. In
Contrary to the lisp-based FACT system, FACT++ fact, support for RDF and OWL is implemented
is an open source reasoner for SHOIQ(D) imple- through plug-ins too. Recently, another plug-in
mented in C++. It is based on tableaux algorithms for querying HEX models through the SPARQL
in order to provide both TBox and ABox reason- language (Polleres & Schindlauer, 2007). Through
ing tasks and it can be accessed through the DIG this plug-in dlvhex can be used as a query engine
interface. It also supports a number of additional by providing a rewriter from SPARQL to rules.
features like handling enumerated classes (a.k.a. The source code and binaries of dlvhex are pub-
nominals) and it is a very efficient TBox reason- licly available.
ing engine. However, the main disadvantage of DR-DEVICE (Bassiliades, et al., 2006) is a
FACT++ is its inefficiency to support complete defeasible logic reasoner for the Web based on
ABox reasoning. Hence, FACT++ is unsuitable the CLIPS expert system shell (CLIPS, 2008)
to applications that call for instance classification that intends to integrate Semantic Web standards
and retrieval. (RDF metadata, XML-syntax of rules) with non-
KAON2 (KAON2, 2008) is the successor of monotonicity (e.g., strong negation). Specifically,
KAON Project (KAON, 2008) and unlike pure it provides reasoning services over RDF metadata
description logic reasoners, KAON2 does not by taking advantage of rules defined by defeasible
implement a tableaux algorithm. In fact, KAON2 logic (strict rules, defeasible rules and defeaters).
is a hybrid reasoning module able to handle both It also claims for reasoning efficiency compared
ontologies expressed in description logics terms to other systems based on logic programming.
and Disjunctive Datalog programs. It implements Jess (Jess, 2008) is a Java framework for
algorithms that reduce description logic SHIQ(D) editing and applying rules, since it contains a
to Disjunctive Datalog (Hustadt, Motik, & Sattler, scripting environment and a rule engine, as well.
2004), taking advantage and applying well-known It supports a CLIPS-like language suitable for
practices stemming from deductive databases (e.g., developing applications based on declarative rules
magic sets) to DL reasoning. It can also handle (a.k.a. expert systems). Jess also uses an optimized
SWRL, F-Logic ontologies and SPARQL query version of Rete algorithm (Forgy, 1982) tailored
answering, as well. Moreover, it provides a Java for Java, comprising a very efficient rule engine.
API in order to accommodate the management Recently, the evolution of rule technologies on the
and the integration of different knowledge formal- Web has led Jess to rebound its practical value in
isms (e.g., OWL ontologies with SWRL rules). the community of Web developers. Moreover, the
However, KAON2 does not support reasoning fact that Jess is a Java-based system facilitates its
about nominals and cannot handle a large number integration with a number of Web programming
of cardinality restrictions. Hence, it cannot deal paradigms like Java servlets or applets. Finally, it
with the full OWL-DL expressiveness. supports backward-chaining and some additional
dlvhex (Eiter, Ianni, Schindlauer, & Tompits, features like procedural attachments.
2006b) is the name of a prototype application Table 2 presents the types of inference sup-
for computing the models of so-called HEX- port provided by the aforementioned reasoning

465
Developing Rule-Based Applications for the Web

Table 2. Types of inference support by various reasoning engines

Inference
Rules
Support TBox Reasoning Abox Reasoning
Reasoning
Modules
Limited Limited
JenaRules
Jena2 (incomplete RDFS/OWL (incomplete RDFS/OWL
(forward/tabled backward chaining)
reasoning) reasoning)
nRQL rules,
RacerPro √ √ first implementation of SWRL
(forward chaining)
Pellet √ √ DL-safe rules (SWRL subset)
sound, sound, SWRL, Buchingae
Bossam
incomplete incomplete (forward chaining)
sound,
FACT++ √ -
incomplete
√ √
KAON2 DL-safe rules (SWRL subset)
(except nominals) (except nominals)
DL-Rules
Limited, through interface Limited, through interface with
dlvhex (non-monotonic logic program rules
with external reasoner external reasoner
with queries to DL KB)
DR-DEVICE - - Defeasible rules
SWRL*, Jess rules (forward/back-
Jess Limited* Limited*
ward chaining)
* through appropriate transformations (Mei, Bontas & Lin, 2005; O’Connor et al., 2005)

modules. In (Cardoso, 2007), an analysis of the 4.1 Rule-Based Web Applications


current trends and the adoption of available rea- are Still “Web Applications”
soning engines by the Semantic Web community
are presented in detail. Before evaluating the existing approaches to intro-
ducing rules in Web applications, we should iden-
tify the main requirements for such approaches.
4. EVALUATION OF EXISTING These requirements mainly stem from the nature of
APPROACHES the Web itself and its current status. Firstly, WWW
is a ubiquitous and massive multi-user distributed
In this section, a comparative study across the environment. In fact, this massive characteristic
aforementioned methodologies for introducing is its strong point, and is a direct consequence of
rules in Web applications is presented. Firstly, we its architectural and technological simplicity. Web
give some fundamental requirements that have to technologies such as HTTP, HTML and XML are
be satisfied by such formalisms. Afterwards, we very simple to learn and use, even for plain users
compare various aspects from the perspectives who are not IT experts. Such simplicity should be
of classical logic and logic programming, giving taken as granted for any Web-related technology,
indicative examples. Finally, some experimental and this applies to rule systems as well. In this
results that examine the efficiency of current ef- context, rules for the Web should be written and
forts in the combination of ontologies with rules used even by users/developers not familiar with
are presented. advanced knowledge engineering concepts. For
example, negation is a rather advanced topic in

466
Developing Rule-Based Applications for the Web

logic-based systems (with Horn rules). Hence, and discuss these efforts by presenting the main
the semantics of rules should be such that can be incompatibilities that impede the reconciliation
easily understood by naïve users. However, this of the two approaches.
would affect their expressiveness and would limit
the inference power of rule-based systems that Monotonicity vs. Non-Monotonicity
may be required by more demanding applications
(i.e., Web service discovery engines). It is worthy Classical logic is based on standard model theoretic
to mention that most of the existing rule-based semantics and adheres to monotonicity of entail-
applications for the Web have adopted SWRL ap- ment. Informally, monotonicity means that the
proach (see Section 3.1) in order to express rules. addition of new information to the knowledge base
SWRL is neither a highly expressive language cannot invalidate conclusions inferred by current
(e.g., no negation is available) nor a decidable knowledge. As a result, classical logic is able to
one, but it remains simple. deal with incomplete information by nature.
Another requirement is that the rule-based Instead of classical logic, logic programming
systems for the Web are compatible with existing has non-monotonic features. It assumes complete
(Semantic) Web standards, such as XML, RDF, knowledge and there is a unique model describ-
OWL and SAWSDL (Farrell & Lausen, 2007). ing the state of the world. This way, addition of
This means that direct support for URIs or XML knowledge may reduce the inferences.
syntax should be available. On the other hand,
writing rules in XML (e.g., like the SWRL and Unique Name Assumption
RuleML approaches propose) is a cumbersome
task. This is indeed a difficult decision that has Another difference between the two paradigms
to be taken by the Web architects. is that logic programming approaches typically
Furthermore, the addition of rules in Web ap- deploy the Unique Name Assumption (UNA).
plications requires their compatibility with exist- This assumption supposes that different names
ing forms of Web knowledge that have already represent different objects of the world. This fact
attained a certain maturity level. Most of the imposes severe limitations on the Web context,
knowledge bases developed in the context of Web since several distinct URIs4 may refer to the same
are expressed in the form of ontologies (mainly content or data. On the contrary, in classical logic
written in OWL or RDFS). As a result, rules have there is no one-to-one correspondence between
to be integrated with ontologies properly. the names and the objects of the domain. Hence,
equality between individuals represented by dif-
4.2 Qualitative Comparison ferent names can be derived. Although it imposes
of Classical Logic and Logic a huge computational cost, most of the current
Programming Approaches description logic reasoners support reasoning
with equality.
The aforementioned knowledge representation
methodologies (see Section 3.1) are either based Negation
on classical logic (e.g., DLs and SWRL) or
logic programming (e.g., answer-set program- Classical logic and logic programming face the
ming and defeasible rules). Across the literature, aspect of negation from different perspectives.
several differences between classical logic and With respect to its monotonic nature, negation in
logic programming paradigms have already been classical logic allows inferring new information
identified. In this section, we intend to survey only if the truth or the falsehood of a statement

467
Developing Rule-Based Applications for the Web

is explicitly declared. This fact is related with the individual may have for a specific proper-
Open World Assumption (OWA) of classical logic ty. They can possibly infer the existence of
theory where incompleteness of the knowledge new instances or equality between known
base is considered. individuals.
On the other hand, Negation-As-Failure (NAF)
adheres to the Closed World Assumption (CWA). In terms of Description Logics, an example of
Specifically, if a description is not known to be cardinality restriction in the web service paradigm
true, then the truth of the negated description is could be the following:
drawn. In that sense, absence of knowledge draws
to negated knowledge and, thus, NAF has a non- Profile  ≥ 1 hasInput (6)
monotonic behavior. Modeling the world accord-
ing to CWA seems to be somewhat inappropriate Such rule denotes that every service profile
for the Semantic Web. Since knowledge on the should have at least one input.
Web is not always available (e.g., web servers On the other hand, constraints (also called
breakdowns), such an assumption could lead to integrity constraints) are mainly used in deductive
incorrect inferences. However, the usefulness databases and logic programming in order to check
of different types of negation in rule-based Web if the knowledge base is consistent with a number
applications is demonstrated in (Wagner, 2003; of specified conditions. As a result, constraints
Alferes, Damasio, & Pereira, 2003). cannot draw inference of new information, but
they may lead to inconsistencies in the case that
Constraints and Restrictions some conditions specified by them are violated
inside the knowledge base. Usually, constraints are
Another aspect where the classical paradigm rules without head (i.e., they have a “false” value
differs from logic programming is on the treat- in their head), denoting that the conditions stated
ment of constraints and restrictions. Regardless in the body should not be satisfied concurrently
of the apparent similarity between constraints by the knowledge base.
and restrictions, there are important differences Similarly to restrictions, constraints may be
between them. divided to:
Restrictions are used in the classical logic
paradigm and they constitute part of the logical • Value constraints or range constraints.
theory. By adding restrictions to a knowledge Their importance consists in checking the
base, the knowledge statements are also increased type of a property value.
and, thus, the inference of additional knowledge • Cardinality constraints. They are used for
is permitted. checking the number of values that an indi-
Restrictions may be further classified to the vidual may have for a specific property.
following main categories:
Similarly to (6), a cardinality constraint could
• Value restrictions or range restrictions. be written in logic programming as:
They are used to restrict the values of a
property and possibly infer new informa- ← Profile(s) ∧ ¬hasInput(s,i) (7)
tion according to their type (a.k.a. range).
• Cardinality restrictions. They are used for More details about integrity constraints and
restricting the number of values that an restrictions may be found in (de Bruijn, Polleres,

468
Developing Rule-Based Applications for the Web

Lara, & Fensel, 2005), where an alternative on- as-failure and decidability of reasoning. More-
tology language based on the logic programming over, many of the reasoning engines targeting
subset of OWL is presented. at managing knowledge expressed in classical
logic formalisms have been recently extended
Other Differences to support such features (e.g., Pellet, RacerPro,
Bossam). Additionally, both users and develop-
A number of additional differences between the ers are very familiar with the world of databases
described approaches have also been discussed in and logic programming languages like Prolog. It
the literature. (Eiter et al., 2006a) raises the point of is worthy to mention that even SPARQL engines
non-ground entailment in the logic programming return “no” in ASK query patterns with no solu-
approach, declaring that, since the semantics of tions. However, “don’t know” would fit better in
logic programming is defined in terms of sets of an open environment where multiple solutions
ground facts, the inference of non-factual knowl- may exist outside the current knowledge base.
edge is not allowed. Contrary to the logic program- This way, SPARQL takes advantage of the users’
ming, classical logic permits the entailment of familiarity with common databases, deciding not
non-factual knowledge. Furthermore, decidability to complicate the query answering process.
issues are explored in the combination of the two To summarize, last decade has shown to the
perspectives. In addition, (Pattel-Schneider, 2006) developers of Web applications that different types
describes the treatment of datatypes by classical of Semantic Web enabled applications requires
logic approaches (e.g., DLs) and logic program- different styles of modeling. As a result, the
ming. Moreover, it examines the role of tools development of a framework that satisfy most of
that support each methodology and how they can these needs seems to become unavoidable.
facilitate the modeling of knowledge.
The aforementioned modeling paradigms have 4.3 Quantitative Evaluation
been proposed for formalizing rules on the Web. of Reasoning Modules
Concerning the advantages and the disadvantages
of each approach, we believe that both perspectives This section intends to quantify the total time
are useful in the context of Semantic Web. required by popular reasoning modules in order
The open nature of the classical logic seems to perform common reasoning tasks. Instead
to be more suitable for formalizing such a distrib- of (Pan, 2005; Gardiner, Horrocks, & Tsarkov,
uted environment. By this mean, Semantic Web 2006), where the evaluation of description logic
technologies based on classical logic accomplish reasoners focuses on TBox reasoning tasks, here
the integration of information stemming from we examine both TBox and ABox reasoning as
different sources. Knowledge becomes shareable well as rules application. In particular, Figures 3
in the sense that it can be accessed without any to 6 show the results of some indicative perfor-
obligation to adopt a specific model or schema. mance tests. These tests aim at demonstrating
Hence, data are available to everyone through the efficiency of description logic reasoners and
WWW so as to be retrieved and used in applica- mainstream rule engines.
tions appropriately. This quantitative evaluation involved the last
On the other hand, logic programming makes stable versions of RacerPro (v1.9.0), Pellet (v1.5.1)
several assumptions in order to simplify tasks and FACT++ (v1.1.11) description logic reason-
like reasoning or modeling of the knowledge ers that were available for ontology reasoning
base. There are several useful features of logic as well as the Jess rule engine for SWRL rules
programming like high expressiveness, negation- application. Bossam was also tested as both ontol-

469
Developing Rule-Based Applications for the Web

Figure 3. Evaluation of consistency check (TBox)

ogy reasoner and rule engine in this evaluation. reasoners were managed through Protégé OWL
However, Bossam’s results are not displayed in API in Java except Bossam which provides its own
the figures, since such a comparison would be API. Finally, Jess was accessed via SWRL-Jess
improper. This stems from the fact that although bridge (O’Connor et al., 2005).
Bossam constitutes a sound reasoning module it The knowledge base used in these tests stems
is not a complete reasoner for OWL. However, from the Web service description paradigm and
we should mention that Bossam performed well it consists of a domain ontology and OWL-S
(i.e., it accomplished smaller execution times than service descriptions. The domain ontology, which
the other OWL reasoners) handling small-size represents the domain of football (i.e., player/team
ontologies, while it threw out-of-memory errors details and statistics), includes concepts described
when provided with large-scale ontologies (i.e., through necessary and sufficient conditions.
containing 500 and 1000 instances). Moreover, all The elements of this ontology are used for the

Figure 4. Evaluation of classification (TBox)

470
Developing Rule-Based Applications for the Web

Figure 5. ABox reasoning evaluation (computation of inferred individual types)

annotation of some Semantic Web services that knowledge base. The execution times presented
provide search functionality to some underlying in this figure do not involve the time needed to
football statistics databases. Specifically, the load the models. The figure shows that these times
service inputs and outputs are annotated through increase as the number of instances contained in
concepts of this domain ontology. The extract of the knowledge base also increases. FACT++ and
the domain ontology used contains 37 concepts, 8 Pellet attained the best performance by achiev-
object properties and 3 datatype properties. Each ing similar execution times. On the other hand,
service description involves 19 concept instances RacerPro performed worse than the others.
and 11 property instances. In Figure 4, the times required for classifying
Figure 3 depicts the performance of three the hierarchy of the ontology are presented. Again,
description logic reasoners while checking the the times shown in the figure do not include the
consistency of all classes over the aforementioned loading time of each model. The resulting perfor-

Figure 6. Jess performance executing SWRL rules

471
Developing Rule-Based Applications for the Web

mances are very similar to those of the consistency plications have to support multiple individual
checking process. FACT++ has slightly better instances.
performance than Pellet, while RacerPro has the To summarize, the time required to handle a
worst performance again. large KB and execute reasoning in terms of Se-
Figure 5 shows the efficiency of RacerPro mantic Web technologies like DLs constitutes a
and Pellet reasoners in realizing reasoning over limiting factor, even for the most efficient reason-
ABoxes. In this figure we do not display FACT++, ers. This is a fundamental problem on the Web,
since it provides incomplete ABox reasoning. In since applications have to deal with a large number
particular, the experiment concerns the identifica- of instances representing data. Especially, when
tion of the types (i.e., classes) that each individual considering real time applications that call for
belongs to. The figure shows that even the reason- small response times, the aforementioned results
ing over an ontology containing 100 instances become unmanageable. This is the main reason
using Pellet (which seems to be the most efficient why many researchers have recently focused on
reasoner in this reasoning task) requires about developing efficient and scalable reasoning ap-
20 seconds. Moreover, both reasoners failed to plications over large individual sets. Instance Store
compute inferred individual types for ontologies (Horrocks, Li, Turi, & Bechhofer, 2004a) is such
containing 500 and 1000 individuals. an approach that permits working in conjunction
It is worthy to mention that version 1.9.2 with any ontology reasoner that implements the
Beta of RacerPro was also available during the DIG interface.
experiments. This version seemed to remain
slower than FACT++ and Pellet in TBox reason-
ing tasks (consistency check and classification 5. FUTURE TRENDS
of hierarchy) while it performed better in ABox AND OPEN ISSUES
reasoning. However, the results are not displayed
in the figures above, since this is not a stable ver- Several open issues that deserve further research
sion of the reasoner. efforts in the future have been identified through-
Figure 6 shows the time required by Jess rule out the chapter. Firstly, a Web standard for editing
engine to apply SWRL rules over ontologies and embedding rules into Web applications should
containing various numbers of instances. These be specified. In our opinion, such a language should
rules represent service composition constraints focus on its simplicity, instead of providing ma-
that should apply to the inputs and outputs of the jor expressiveness capabilities that will lead to a
services. We applied two different sets of rules clumsy formalism. Rule Interchange Format (RIF)
to the previously described knowledge bases: the Working Group (RIF, 2008) works in this direc-
first set consisted of 10 rules while the second tion in order to produce W3C recommendations
of 50 rules. In both cases, the execution time in- for enabling interchange of rules. Specifically,
creases proportionally to the number of instances RIF does not intend to provide explicit mappings
contained in the ontology. As expected, the delay between various rule languages, but specify a
imposed by the execution of rules was affected mechanism capable of defining the meaning of
by the number of rules. However, the number of the formulas of a rule language. This way, rules
instances seems to affect execution times more could be automatically translated across different
than the number of rules. Hence, the inference formalisms.
process is complicated when real-time demands Additionally, more tractable algorithms for
come up in distributed environments where ap- Web-rules, TBox and ABox inference problems

472
Developing Rule-Based Applications for the Web

should be developed. These algorithms should not efficient enough to manage knowledge stem-
be able to handle complex and heavy knowledge ming from large-scale environments like the Web,
bases that combine rules and ontologies with a especially in the context of real-time applications
large number of concepts and relationships. Such which impose severe constraints in response times.
knowledge bases are substantial, especially in the Finally, more improvements should be made in
context of information integration where multiple the standardization of rules formalism on the Web
domain vocabularies interact with upper-level and in the development of reasoning modules that
ontologies. Furthermore, the Web is a large scale can handle and reason over both ontologies and
environment, where a huge set of resources are rules as integral knowledge.
added every day. As a result, future knowledge
bases will have to capture and describe more and
more individuals. These facts should be taken 7. REFERENCES
into account by the developers that will target at
designing new reasoning algorithms. W3C Semantic Web Activity (2008). Retrieved
Finally, a more practical issue that should be April 22, 2008, from http://www.w3.org/2001/
investigated is the development of a unified rea- sw/.
soning framework capable of managing both on- Alferes, J., Damasio, C., & Pereira, L. (2003).
tologies and rules. Today, there is no efficient and Semantic Web logic programming tools. ( . LNCS,
easy-to-use integrated reasoning module that can 2901, 16–32.
reason over both formalisms. As a consequence,
the developer should use at least two different Antoniou, G., Billington, D., Governatori,
reasoning modules to handle such an integrated G., & Maher, M., J. (2001). Representation
knowledge base. This can result to unexpected results for defeasible logic. ACM Transac-
situations. For example, the restrictions defined tions on Computational Logic, 2(2), 255–287.
by the ontology can be violated by the applica- doi:10.1145/371316.371517
tion of rules, since the rule engines do not take
Antoniou, G., & Damasio, C. V., Grosof, B.,
into account these restrictions (e.g., disjointness
Horrocks, I., Kifer, M., Maluszynski, J., et al.
of concepts). Similarly, the application of rules
(2005). Combining rules and ontologies: A survey.
could produce knowledge that would be useful
Deliverables I3-D3, REWERSE. Retrieved from,
for further description logic inferences.
http://rewerse.net/deliverables/m12/i3-d3.pdf.
Antoniou, G., & van Harmelen, F. (2004). A Se-
6. CONCLUSION mantic Web primer. MIT Press.
Baader, F., Calvanese, D., McGuiness, D., Nardi,
In this chapter we have discussed the application of
D., & Patel-Schneider, P. (2003). The description
rules to Web applications in order to achieve intel-
logic handbook: Theory, implementation, and
ligent application behavior and efficient manage-
applications. Cambridge: Cambridge University
ment of knowledge. We have described the main
Press.
methodologies and technologies for integrating
rules with ontologies, since the latter constitutes Bassiliades, N., Antoniou, G., & Vlahavas, I.
a mature knowledge technology on the Web. The (2006). A defeasible logic reasoner for the Se-
chapter has also examined aspects of the differ- mantic Web . International Journal on Semantic
ent approaches. We also showed through several Web and Information Systems, 2(1), 1–4.
experiments that current reasoning modules are

473
Developing Rule-Based Applications for the Web

Bechhofer, S., Moller, R., & Crowther, P. (2003). Donini, M. F., Lenzerini, M., Nardi, D., &
The DIG description logic interface. In D. Cal- Schaerf, A. (1998). AL-log: Integrating data-
vanese, G. de Giacomo, & F. Franconi (Eds.), log and description logics. Journal of Intel-
Proc. of the 2003 Int. Workshop on Description ligent Information Systems, 10(3), 227–252.
Logics (DL 2003) (Vol. 81 of CEUR Workshop doi:10.1023/A:1008687430626
Proceedings), Rome, Italy.
Drools (2008). Retrieved April 22, 2008, from
Boley, H., Kifer, M., & Patranjan, P. L., & Polleres, http://www.jboss.org/drools/.
A. (2007). Rule interchange on the Web. In
Eiter, T., & Ianni, G. B., Schindlauer, R., &
Reasoning Web 2007 (No. 4636, pp. 269-309).
Tompits, H. (2005). A uniform integration of
Springer.
higher-order reasoning and external evaluations
Brickley, D., & Guha, R. V. (2004). RDF vo- in answer set programming. In Proc. IJCAI 2005.
cabulary description language 1.0: RDF schema. Morgan Kaufmann.
Recommendation 10 February 2004, W3C. Re-
Eiter, T., & Ianni, G. B., Polleres, A., Schindlauer,
trieved April 22, 2008, from http://www.w3.org/
R., & Tompits, H. (2006a). Reasoning with rules
TR/rdf-schema/.
and ontologies. In P. Barahona, F. Bry, E. Fran-
Cardoso, J. (2007). The Semantic Web vision: coni, U. Sattler, & N. Henze (Eds.), Reasoning
Where are we? Intelligent Systems, 22(5), 84–88. Web, Second International Summer School 2005,
doi:10.1109/MIS.2007.4338499 Tutorial Lectures (LNCS, pp. 93-127).
CLIPS. A tool for building expert systems. (2008). Eiter, T., & Ianni, G. B., Schindlauer, R., &
Retrieved April 22, 2008, from http://clipsrules. Tompits, H. (2006b). dlvhex: A system for in-
sourceforge.net/. tegrating multiple semantics in an answer-set
programming framework. In M. Fink, H. Tompits,
de Bruijn, J., Lausen, H., Krummenacher, R.,
& S. Woltran (Eds.), Proceedings 20th Workshop
Polleres, A., Predoiu, L., Kifer, M., et al. (2005).
on Logic Programming and Constraint Systems
The Web service modeling language WSML (Tech.
(WLP 06), (pp. 206-210).
Rep. WSML. WSML Final Draft D16.1v0.21).
Retrieved from http://www.wsmo.org/TR/d16/ Eiter, T., Lukasiewicz, T., Schindlauer, R., &
d16.1/v0.21/. Tompits, H. (2004). Combining answer set pro-
gramming with description logics for the Semantic
de Bruijn, J., Polleres, A., Lara, R., & Fensel, D.
Web. In Proc. of the International Conference
(2005). OWL DL vs. OWL flight: Conceptual
of Knowledge Representation and Reasoning
modeling and reasoning on the Semantic Web.
(KR04).
In Proc. WWW2005, (pp. 623-632).
Farrell, J., & Lausen, H. (2007). Semantic annota-
Dean, M., Schreiber, G., Bechhofer, S., van
tions for WSDL and XML schema. W3C Recom-
Harmelen, F., Hendler, J., Horrocks, I., et al.
mendation, W3C, August 2007. Retrieved April
(2004). OWL Web ontology language reference.
22, 2008, from http://www.w3.org/TR/sawsdl/.
W3C Recommendation 10 February 2004. Re-
trieved April 22, 2008, from http://www.w3.org/ Forgy, C. (1982). Rete: A fast algorithm for
TR/owl-ref/. the many pattern/many object pattern match
problem. Artificial Intelligence, 19, 17–37.
doi:10.1016/0004-3702(82)90020-0

474
Developing Rule-Based Applications for the Web

Gardiner, T., Horrocks, I., & Tsarkov, D. (2006). Hustadt, U., Motik, B., & Sattler, U. (2004).
Automated benchmarking of description logic Reducing SHIQ- description logic to disjunctive
reasoners. In Proc. of DL. datalog programs. In Proc. of the 9th International
Conference on Knowledge Representation and
Gelfond, M., & Lifschitz, V. (1991). Classi-
Reasoning (KR2004), (pp. 152-162).
cal negation in logic programs and disjunctive
databases. New Generation Computing, 9(3-4), ILOG. Business Rule Management Systems,
365–386. doi:10.1007/BF03037169 Optimization Tools and Engines, Visualization
Software Components, Supply Chain Applications
Grosof, B. N., Horrocks, I., Volz, R., & Decker, S.
(2008). Retrieved April 22, 2008, from http://
(2003). Description logic programs: Combining
www.ilog.com/.
logic programs with description logic. In Proc. of
the Twelfth International World Wide Web Confer- Jang, M., & Sohn, J.-C. (2004). Bossam: An ex-
ence (WWW 2003), (pp. 48-57). ACM. tended rule engine for OWL inferencing. In Work-
shop on Rules and Rule Markup Languages for the
Haarslev, V., & Moller, R. (2001). RACER sys-
Semantic Web at the 3rd International Semantic
tem description. In R. Gor’e, A. Leitsch, and T.
Web Conference (LNCS 3323, pp. 128-138).
Nipkow (Eds.), International Joint Conference on
Automated Reasoning, IJCAR’2001, June 18-23, Jess, The Rule Engine For the Java Platform
Siena, Italy, (pp. 701–705). Springer-Verlag. (2008). Retrieved April 22, 2008, from http://
www.jessrules.com/jess/index.shtml
Horrocks, I. (1998). Using an expressive descrip-
tion logic: FaCT or fiction? In Proceedings of KAON2 – Ontology Management for the Semantic
the Sixth International Conference on Principles Web (2008). Retrieved April 22, 2008, from http://
of Knowledge Representation and Reasoning kaon2.semanticweb.org.
(KR’98), Trento, Italy, June 2–5, 1998, (pp.
KAON – The KArlsruhe ONtology and Semantic
636–647).
Web tool suite (2008). Retrieved April 22, 2008,
Horrocks, I., Li, L., Turi, D., & Bechhofer, S. from http://kaon.semanticweb.org.
(2004a). The instance store: DL reasoning with
Kifer, M., de Bruijn, J., Boley, H., & Fensel, D.
large numbers of individuals. In Proc. of the
(2005). A realistic architecture for the Semantic
2004 Description Logic Workshop (DL~2004),
Web. In Proc. RuleML 2005 (pp. 17-29), Galway,
(pp. 31-40).
Ireland.
Horrocks, I., Parsia, B., & Patel-Schneider, P. F.,
Klyne, G., & Carroll, J. J. (2004). Resource
& Hendler, J. (2005). Semantic Web architecture:
description framework (RDF): Concepts and
Stack or two towers? In Proc. PPSWR 2005, (pp.
abstract syntax. Recommendation 10 February
37-41), Dagstuhl Castle, Germany.
2004, W3C. Retrieved April 22, 2008, from http://
Horrocks, I., Patel-Schneider, P. F., Boley, H., www.w3.org/TR/rdf-concepts/.
Tabet, S., Grosof, B., & Dean, M. (2004b). SWRL:
McBride, B. (2002). Jena: A Semantic Web
A Semantic Web rule language combining owl
toolkit . IEEE Internet Computing, 6(6), 55–59.
and RuleML. W3C Member Submission, 21 May
doi:10.1109/MIC.2002.1067737
2004. Retrieved April 22, 2008, from http://www.
w3.org/Submission/SWRL/.

475
Developing Rule-Based Applications for the Web

Mei, J., Bontas, E. P., & Lin, Z. (2005). OWL2Jess: Patel-Schneider, P. F., & Horrocks, I. (2006).
A transformational implementation of the OWL Position paper: a comparison of two modelling
semantics. In Proceedings of International Work- paradigms in the Semantic Web. In Proceedings
shops on ISPA (LNCS 3759, pp. 599-608). of the 15th International Conference on World
Wide Web (Edinburgh, Scotland, May 23 - 26,
Motik, B., Horrocks, I., Rosati, R., & Sattler, U.
2006). WWW ‘06, New York, NY (pp. 3-12).
(2006). Can OWL and logic programming live
ACM Press.
together happily ever after? In Proceedings ISWC-
2006 (LNCS 4273, pp. 501–514). Polleres, A., & Schindlauer, R. (2007). DLVHEX-
SPARQL: A SPARQL Compliant Query engine
Motik, B., Sattler, U., & Studer, R. (2005). Query
based on DLVHEX. In Proceedings of 2nd In-
answering for OWL-dl with rules. Journal of
ternational Workshop on Applications of Logic
Web Semantics: Science . Services and Agents
Programming to the Web, Semantic Web and
on the World Wide Web, 3, 41–60. doi:10.1016/j.
Semantic Web Services (ALPSWS2007), Porto,
websem.2005.05.001
Portugal.
Nute, D. (1994). Defeasible logic. In Handbook of
Prud’hommeaux, E., & Seaborne, A. (2005).
Logic in Artificial Intelligence and Logic Program-
SPARQL query language for RDF.http://
ming, Nonmonotonic Reasoning and Uncertain
www.w3.org/TR/2005/WD-rdf-sparql-que-
Reasoning (Vol. 3). Oxford University Press.
ry-20050217/, 2005.
O’Connor, M., Knublauch, H., Samson, T., &
RacerPro (2008). Retrieved April 22, 2008, from
Musen, M. (2005). Writing rules for the Semantic
http://www.racer-systems.com/products/racerpro/
Web using SWRL and Jess. Protégé with Rules
index.phtml.
WS. Madrid.
Rainer. A. (2005). Web service composition under
Pan, J. (2005). Benchmarking DL reasoners using
answer set programming. KI-Workshop “Planen,
realistic ontologies. In Proc. of the OWL: Experi-
Scheduling und Konfigurieren, Entwerfenl”
ences and Directions Workshop.
(PuK).
Pan, J. Z., & Horrocks, I. (2004). OWL-E: Ex-
RIF - Rule Interchange Format Working Group
tending owl with expressive datatype expressions
(2008). Retrieved April 22, 2008, from http://www.
(Technical report, IMG/2004/KR-SW-01/v1.0).
w3.org/2005/rules/wiki/RIF_Working_Group
Victoria University of Manchester.
Rosati, R. (2006a). DL+log: Tight integration of
Parsia, B., Sirin, E., & Kalyanpur, A. (2005).
description logics and disjunctive datalog. In P.
Debugging OWL Ontologies. In Proceedings of
Doherty, J. Mylopoulos, & C.A. Welty (Eds.),
the 14th International World Wide Web Confer-
Proceedings of the 10th International Conference
ence (WWW2005), Chiba, Japan, May 2005.
on Principles of Knowledge Representation and
Retrieved from http://www.mindswap.org/papers/
Reasoning (KR 2006), Lake District, UK (pp.
debuggingOWL.pdf.
68-78). AAAI Press.

476
Developing Rule-Based Applications for the Web

Rosati, R. (2006b). Integrating ontologies and Knowledge-Based System: A system that


rules: Semantic and computational issues. In P. exploits knowledge representation and reason-
Barahona, F. Bry, E. Franconi, N. Henze, & U. ing techniques in order to achieve an intelligent
Sattler (Eds.), Reasoning Web, Second Interna- behavior.
tional Summer School 2006, Lissabon, Portugal, Knowledge Representation: It is the process
September 4_8, 2006, Tutorial Lectures. (LNCS of exploiting formal methodologies and languages
4126, pp. 128-151). (e.g., rules formalisms, ontologies) in order to
capture and describe knowledge.
Rosati, R. (2006c). DL+log: Tight integration
Ontology: A formal representation of a do-
of description logics and disjunctive datalog. In
main of discourse that describes concepts and
Proceedings of the Tenth International Conference
relationships among them. This way, it provides a
on Principles of Knowledge Representation and
common vocabulary and allows for the inference
Reasoning (KR 2006) (pp. 68-78). AAAI Press.
of new knowledge.
Rule, M. L. Rule Markup Initiative (2008). Re- Reasoning: The systematic process of infer-
trieved April 22, 2008, from http://www.ruleml. ring new knowledge by applying formal implica-
org/. tion rules to a given knowledge base.
Rule-Based Application: An application
Sirin, E., Parsia, B., & Grau, B., C., Kalyanpur
that takes advantage of rules formalisms (e.g.,
A., & Katz, Y. (2007). Pellet: A practical OWL-
Horn-clauses) in order to represent and infer new
DL reasoner. Journal of Web Semantics, 5(2).
knowledge.
doi:10.1016/j.websem.2007.03.004
Semantic Web: It is an extension of the current
Tsarkov, D., & Horrocks, I. (2006). FaCT++ Web where resources are described through formal
description logic reasoner: System description. syntax and semantics in order to be human- and
In Proc. of the International Joint Conference on machine-readable.
Automated Reasoning (IJCAR 2006).
Wagner, G. (2003). Web rules need two kinds of
Endnotes
negation (LNCS 2901, pp. 33-50).
1
Also called as Datalog view in the litera-
ture
2
Key Terms and definitions http://con.fusion.at/dlvhex/
3
It offers free trials while educational and
Logic Programming: It is a declarative research licenses are available
4
paradigm of programming that is mainly based Universal Resource Identifiers
on first-order logic. Prolog is the most common
logic programming language.

This work was previously published in Handbook of Research on Emerging Rule-Based Languages and Technologies: Open
Solutions and Approaches, edited by A. Giurca, D. Gasevic, and K. Taveter, pp. 371-392, copyright 2009 by Information Sci-
ence References (an imprint of IGI Global).

477
478

Chapter 2.13
Modeling of Web Services
using Reaction Rules
Marko Ribarić
Mihailo Pupin Institute, Serbia

Shahin Sheidaei
Simon Fraser University, Canada

Milan Milanović
University of Belgrade, Serbia

Dragan Gašević
Athabasca University, Canada

Adrian Giurca
Brandenburgische Technische Universität Cottbus, Deutschland

Sergey Lukichev
Brandenburg University of Technology at Cottbus, Germany

Gerd Wagner
Brandenburg University of Technology, Germany

Abstract services in terms of message exchange patterns.


Web services that are being modeled in this way
The development process of Web services needs can easily be integrated in the wider context of
to focus on the modeling of business processes modeling orchestration and choreography. In order
rather than on low-level implementation details of to achieve proposed solution, we have developed a
Web services, and yet it also needs to incorporate plug-in for the Fujaba UML tool (so called Strelka)
the support for frequent business changes. This and a number of model transformations for round-
chapter presents the UML-based Rule Language trip engineering between Web services and reaction
(URML) and REWERSE Rule Markup Language rules. Also, the paper presents mappings of models
(R2ML), which use reaction rules (also known as of Web services with reaction rules into the Drools
Event-Condition-Action rules) for modeling Web rule language, thus enabling the run time execution
semantics for our rule-based models.
DOI: 10.4018/978-1-60566-402-6.ch018

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Modeling of Web Services using Reaction Rules

1. Introduction be more easily integrated in the wider context of


modeling orchestration and choreography.
Web services provide a communication inter- A promising way to solve these problems is
face and a workflow management protocol for to use a high-level modeling approach combined
business systems. Their goal is set to change the with the use of rules. A modeling approach will
Web from a static collection of information, to a allow developers to focus on a problem domain
dynamic place where different software compo- rather than on an implementation technology. This
nents (business logics) can be easily integrated is why we propose using an approach based on
(Arroyo et al., 2004). Some of the factors that Model Driving Engineering (MDE). Although,
constrain web services from achieving that goal there have been several attempts to leverage MDE
can be named as follow: when developers try to principles to model Web services, they are still
enrich the existing Web services with some new very low level oriented, as they again focus on
functionality, they have to implement necessary technical details covered either by WSDL (Bezivin
changes manually, which may lead to potential et al., 2004) (Vara et al., 2005) or OWL-S (Timm
execution errors; furthermore, business rules, due & Gannod, 2005) (Gronmo et al., 2005). The use
to their declarative nature, are used for reflecting of rules implying that they are defined in a declara-
dynamic changes representing desired alteration tive way, they can dynamically reflect business
in the business process. When business rules logic changes at run-time without the need to
are used in a networked business, they are often redesign the whole system. Since Web services
large and complex, which makes them difficult to are used for integration of business processes of
manage and change. There is a lack of automatic various stakeholders, it is important for them to
mechanism for updating Web services based on the reflect changes in business logic, or policies, as
business rules changes (Ribaric et al., 2008). good as possible.
The need for a more straightforward way to ex- The solution that we propose is to use rules
press, and manage business rules as a separate part to represent and model business processes from
of Web services’ composition has been recognized the perspective of message exchange. That is, our
by researchers for the numerous benefits that it modeling approach enables one to model Web
provides (Charfi & Mezini, 2004). However, in- services from the perspective of the underlying
tegrating rule-based systems in a service-oriented business logic regulating how Web services are
environment is a complex task, due to the fact that used regardless of the context where they are used.
both worlds have their own paradigms (Rosenberg To do so, our proposal is to leverage message-
& Dustdar, 2005). Many different factors involve exchange patterns (MEPs) as an underlying
in this integration. However, the most important perspective integrated into a Web service model-
one of these factors is the architecture of service ing language. This perspective has already been
oriented systems. In this architecture, layers play recognized by Web service standards (Chinnici
a great role. The layering architecture is needed to et al., 2007), and our contribution is to raise this
be represented and deployed in the world of rules to the level of models.
which become challenging and could not be simple As the leading Rule Modeling language we
done. This integration approach should be loosely benefit from UML-based Rule Modeling Lan-
coupled, meaning that it is reasonable to expose guage (URML) (Lukichev & Wagner, 2006a)
business rules as services. The way of modeling for our approach. URML having strong ties with
Web services presented in this paper facilitates R2ML (REWERSE Rule Markup Language)
this integration process, and those services can empowers us to model rules in various languages.

479
Modeling of Web Services using Reaction Rules

URML supports modeling of domain vocabular- of Web services, Drools rule engine, and rule
ies in addition to integrity, derivation, production language definitions based on MDE principles
and reaction rules. .Business rules, being similar used in our modeling approach. A complete
to natural language rules, are proposed to be introduction into the ATLAS Transformation
modeled by using reaction rules (also known as Language (ATL) can be found in the chapter en-
Event-Condition-Action, ECA, rules) in this pa- titled “Sharing ontologies and rules using model
per. Reaction rules will perform a task (Action) transformations”.
under some circumstances (Conditions) if some
events (Event) take place and can best describe Web Services
a business rule.
We also have developed a plugin (called A Web service is a loosely coupled component that
Strelka) for the well known UML tool Fujaba, exposes functionality to a client over the Internet
which has a support for URML notation. Strelka (or an intranet) by using web standards such as
also has the support transformation of URML HTTP, XML, SOAP, WSDL, and UDDI (Timm
models into Web Service Description Language & Gannod, 2005).
(WSDL) and different types of rule languages SOAP is an XML-based protocol for exchang-
such as Drools, Jess, and SWRL. In this paper, ing information in a decentralized, distributed
we discuss those transformations that are imple- environment. SOAP builds on XML and common
mented by using a model transformation language Web protocols (HTTP, FTP, and SMTP) (Coyle,
entitled the ATLAS Transformation Language 2002). A SOAP message is the basic unit of com-
(ATL) (ATL, ver. 0.7). munication between SOAP nodes. The “envelope”
The paper is structured as follows: in the next element represents the root of a SOAP message
section we give a bit of the background, including, structure. It contains a mandatory body construct
Web services, R2ML and URML languages with and optional header construct (Gudgin et al., 2007).
the emphasis on reaction rules. In section 3 we The header construct is where meta-information
introduce our approach for modeling business rules can be hosted. In a large number of service-oriented
and services following by section 4 in which we architectures, the header is an important part of
introduce Strelka (Lukichev & Wagner, 2006b), the overall architecture, and although it is optional
a tool we developed that is capable of presenting it is rarely omitted.
URML diagrams, and performing transformations. Web Service Description Language (WSDL)
Section 5 describes mapping process between is a language for describing both the abstract
URML and WSDL that is a part of Strelka, and functionality of a service and the concrete details
also covers a deployment of URML diagrams to of a Web service (Chinnici et al., 2007). At an
the Drools rule engine. Section 6 summarized abstract level, WSDL describes a Web service in
the related work done so far and in Section 7 we terms of interfaces and the operations supported
conclude this work by summarizing the major by the interfaces. An operation is an interaction
points. with the service consisting of a set of (input,
output, infault and outfault) messages exchanged
between the service and other parties involved in
2. Background the interaction (Chinnici et al., 2007). The mes-
sages are described independently of a specific
In this section, we give a brief overview of the wire format by using a type system, typically
technologies and techniques relevant to the prob- XML Schema. WSDL also describes the point
lem under study. This includes a short description of contact for a service provider, known as the

480
Modeling of Web Services using Reaction Rules

endpoint – it provides a formal definition of the • out-in pattern – also consists of exactly two
endpoint interface and also establishes the physical messages (like in-out pattern), but now a
location (address) of the service. service provider is the one that initiates the
Potential requestors need a way to discover exchange of messages;
Web services descriptors. It is necessary that • out-only – consists of exactly one message
these descriptors are collected and stored in a (like in-only pattern), and is most often
central registry. The key part of the Universal used for message notification;
Description Discovery and Integration (UDDI) • robust out-only - presents a variation of the
specification (Clement et al., 2004) presents out-only pattern that provides an option of
standardizing information inside such a registry sending a fault massage;
as well as specifying the way the information can • in-optional-out – is similar to pattern in-out
be searched and updated. with one exception: sending an message
Regardless of how complex tasks performed that represents a response is optional, and
by a Web service are, almost all of them require because of this a requester that has started
the exchange of multiple messages (Erl, 2005). a communication should not expect this
It is important to coordinate these messages in a message;
particular sequence, so that the individual actions • out-optional-in – is opposite to the previ-
performed by the message are executed properly. ous pattern, where the input message is
Message exchange patterns (MEPs) are a set of optional.
templates that provide a group of already mapped
out sequences for the exchange of messages (Erl, Drools
2005). This basically means that MEPs define how
services should be used, as they can coordinate Drools is a business rule management system
input and output messages related to a certain (BRMS) with a forward chaining inference based
operation. rule engine, more correctly known as a production
The WSDL 2.0 specification defines three rule system, using an enhanced implementation
MEPs: of the Rete algorithm. BRMS is a production rule
system with a focus on knowledge representation
• in-only pattern – supports a standard fire- to express propositional and first order logic in
and-forget pattern (i.e., only one message a concise, non ambiguous and declarative man-
is exchanged); ner. The core of a production rules system is an
• robust in-only pattern – presents a varia- inference engine that is able to scale to a large
tion of the in-only pattern that provides number of rules and facts. The inference engine
an option of sending a fault massage, as matches facts and data, against production rules,
a result of possible errors generated while also called productions, to infer conclusions which
transmitting, or processing data; result in actions. A production rule is a two-part
• in-out pattern – presents a request-response structure using first order logic for knowledge
pattern where two messages (input and representation (Proctor et al). Figure 1 presents a
output) must be exchanged. schema of a production rule. It executes the actions
while the conditions are being satisfied. Drools
However, WSDL 2.0 specification offers representation contains the same structure, LHS
the possible use of five more patterns (Lewis, (Left Hand Side) is referring to conditions while
2007): RHS (Right Hand Side) is referring to actions.
Name of the rule which is presented by “name”

481
Modeling of Web Services using Reaction Rules

Figure 1. First order logic representation (left) -


modeling languages such as UML and preferably
Drools representation (right)
defined by following MDE principles. Finally,
the language should be closely related to the
Web standards for defining ontologies (OWL)
and rules (RIF).
We have decided to use REWERSE I1 Rule
Markup Language (R2ML), as it fully satisfies
the above requirements. The language is defined
by a MOF-based metamodel which is refined by
OCL constraints that precisely define relations
between the language’s constructs in addition
to those defined by the metamodel; it has an
is just for better understanding of rules and have XML schema defined concrete syntax; it has a
no functional value. Attributes, marked with at- UML-based graphical concrete syntax, so called
tributes in the rule are used for the rule internal UML-based Rule Modeling Language (URML),
functionality itself. which we will introduce it in the section below;
The process of matching new or existing facts and it has a number of transformations with other
against production rules is called pattern match- rule languages (e.g., JBoss’ Drools and OCL) al-
ing, which is performed by the inference engine lowing us to translate Web service models to the
(part of Drools). Drools implements and extends rule-based languages that can then regulate the
the Rete algorithm The Drools Rete implementa- use of Web services.
tion is called ReteOO, signifying that Drools has R2ML is a rule language that addresses all the
an enhanced and optimized implementation of requests defined by the W3C working group for
the Rete algorithm for object oriented systems the standard rule interchange format (Ginsberg
(Schneier). For better understanding of rules and et al., 2006). The R2ML language can represent
their structure please refer to the chapter “Object different types of rule constructs, that is, it can
Oriented Rule Languages”. represent different types of rules (Wagner et al.,
2006) including: integrity rules, derivation rules,
R2ML production rule, and reaction rules. Integrity rules
in R2ML, also known as (integrity) constraints,
Existing Web rule languages (RuleML (Hirtle consist of a constraint assertion, which is a sentence
et al., 2006) and SWRL (Horrocks et al., 2004)) in a logical language such as first-order predicate
and rule modeling approaches (SBVR (SBVR, logic or OCL. Derivation rules in R2ML are used to
2006) and PRR (Charfi & Mezini, 2004)) very derive new knowledge (conclusion) if a condition
nicely serve the purpose of rule interchange and holds. Production rules in R2ML produce actions
rule representation independent of any specific if the conditions hold, while post-conditions must
platform, respectively. However, none of these also hold after the execution of actions. A reaction
languages offers a suitable modeling foundation rule is a statement of programming logic (Giurca
that can be used for modeling Web services from et al., 2006) that specifies the execution of one
a more abstract perspective of business processes or more actions in the case of a triggering event
and business rules in which they are used. That occurrence and if rule conditions are satisfied.
is, the language should be able to model reaction Optionally, after the execution of the action(s),
rules. Yet, such a modeling language should be post-conditions may be made true. R2ML also
closely related to the already existing software allows one to define vocabularies by using the

482
Modeling of Web Services using Reaction Rules

Figure 2. Definition of reaction rules in the R2ML


modeling approach, here we only describe the
metamodel
details of R2ML reaction rules. Reaction rules
represent a flexible way for specifying control
flows, as well as for integrating events/actions
from a real life (Giurca et al., 2006). Reaction
rules are represented in the R2ML metamodel
as it is shown in Figure 2: triggeringEvent is
an R2ML EventExpression (Fig. 4); conditions
are represented as a collection of quantifier free
logical formulas; producedAction is an R2ML
EventExpression and represents a system state
change; and (optional) postcondition must hold
when the system state changes.
following constructs: basic content vocabulary,
The R2ML event metamodel defines basic
functional content vocabulary, and relational con-
concepts that are needed for dynamic rule be-
tent vocabulary. Here we give short description of
havior (Figure 3). There are two types of events:
vocabulary constructs that we use in this paper.
atomic event (AtomicEventExpression) and
Vocabulary is a concept (class) that can have one
composite events (AndNotEventExpression, Se-
or more VocabularyEntry concepts. Vocabulary-
quenceEventExpression, ParallelEventExpression
Entry is abstract concept (class) that is used for
or ChoiceEventExpression). Characteristic of an
representing other concepts by its specialization.
atomic event is that it has no duration (duration
For example, one of the VocabularyEntry-s is an
= 0). For the sake of modeling Web services,
R2ML Class concept which represents the class
we are using MessageEventExpression for both
element similar to the notion of the UML Class. An
triggering events and produced actions. Messa-
R2ML Class can have attributes (class Attribute),
geEventExpression is used for modeling messages
reference properties (class ReferenceProperty)
that are part of the definition of Web services,
and operations (class Operation).
including, input, output, in-fault, and out-fault
Due to the importance for our Web service
messages. Each MessageEventExpression has its

Figure 3. Event expressions in the R2ML metamodel

483
Modeling of Web Services using Reaction Rules

own type – EventType. In terms of WSDL, mes- the input message (input-Message in Figure 4b of
sage types are defined by XML Schema complex the Web service operation is MessageEventType
types, while in R2ML, EventType is a subclass (a subclass of EventType from Figure 3) and it is
of Class (already defined as a part of the R2ML represented by using the <<message event type>>
Vocabulary). This means that each EventType stereotype on UML classes. The UML class that
has its own attributes, associations, and all other represents the input fault message (inFault in
features of R2ML classes. Figure 4b) of the Web service operation is Fault-
MessageEventType in the URML metamodel.
URML In URML diagrams, FaultMessageEventType is
represented by the <<fault message event type>>
UML-Based Rule Modeling Language (URML) stereotype on UML classes. EventCondition con-
is a graphical concrete syntax of R2ML. URML tains an object variable (ObjectVariable in Figure
is developed as an extension of the UML meta- 4c), which is a placeholder for an instance of the
model to be used for rule modeling. In URML, MessageEventType class. The object variable has
modeling vocabularies is done by using UML a name that corresponds to the arrow annotation,
class models. Rules are defined on top of such which represents EventConditon.
models. The URML reaction rules metamodel, An instance of the RuleAction class is repre-
which we use for modeling services, is shown sented as an outgoing arrow on the URML diagram,
in Figure 4a. The figure shows components of a from the circle that represents the reaction rule to
reaction rule: Condition, Postcondition, RuleAc- the class that represents either an output message
tion and EventCondition. The figure also shows or an output fault message of the Web service op-
that reaction rules are contained inside the UML eration. The UML class that represents the output
package which represents Web services operation. message (outputMessage in Figure 4c) of the Web
This means, that such packages have a stereotype service operation is MessageEventType and it is
<<operation>> in UML diagrams. An instance represented with the <<message event type>>
of the EventCondition class is represented on stereotype on UML classes. The UML class that
the URML diagram as incoming arrow (e.g., see represents the output fault message (outFault) of
Figure 5), from a UML class that represents either the Web service operation is FaultMessageEvent-
an input message or an input fault message of the Type in the URML metamodel and it is represented
Web service operations, to the circle that represents with the <<fault message event type>> stereotype
the reaction rule. The UML class that represents on UML classes. RuleAction contains an object

Figure 4. a) Extension of the URML metamodel for reaction rules; b) Part of the URML meta-model for
EventCondition; c) Extension of the URML metamodel for actions

484
Modeling of Web Services using Reaction Rules

Figure 5. PIM level of the in-out message exchange pattern with out-fault, presented in URML

variable (ObjectVariable), which represents an logical constraint, requiring less amount of items
instance of the MessageEventType class. in stock, than before the rule execution.
There are several general advantages of using
reaction rules for specifying web services:
3. Modeling
• Requirements are often captured in the
In this section the idea behind modeling of rules form of rules in a natural language, formu-
and business processes will be exposed. Modeling lated by business people;
will be performed with the help of URML and • Reaction rules are easier to maintain and
R2ML. In section 5 with the use of modeling, the integrate with other kinds of rules, used
transformation to the web services and automatic in business applications (integrity rules,
generation of codes will be described. which specify constraints the data must
fulfill, derivation rules, which explain how
Rule Modeling a model element can be derived);
• The topic of rules validation and verifica-
As stated, in this paper we are concerned with tion is well studied;
reaction rules. Reaction rule formalizes event- • Reaction rules emphasis on events gives
condition-action behavioral model, where the a flexible way to specify control flow and
action is executed on event with a condition satis- web service interaction patterns.
fied. Let us analyze the following example: On
customer book item request, if the item is avail- Our approach is on the edge of the MDE, Web
able, then approve order and decrease amount Services, and rules. It brings several large com-
of items in stock. In this rule on customer book munities together and helps to facilitate the use
item request is a triggering rule event, if the item of rules and MDE technologies in the growing
is available is a rule condition, approve order is a area of Web Services.
rule action and decrease amount of items in stock
is a postcondition, which can be expressed as a

485
Modeling of Web Services using Reaction Rules

Business Rule Modeling


ON UserMessageArrival (echoIn-
In this section we describe the concept of our ap- Param)
proach talking about modeling of Business rules IF echoInParam.length < 5 THEN
with the help of web services. In our approach, we DO EchoStringResponse (echoIn-
look from the perspective of the potential patterns Param)
of the use of services. That is, we model services ON UserMessageArrival (echoIn-
from the perspective of MEPs. It is important Param)
to point out that we first start from the defini- IF NOT echoInParam.length < 5
tion of a business rule that corresponds to a THEN
MEP under study, but without considering the DO EchoStringError (“message too
Web services that might be developed to support long”)
that rule. In this way, unlike other approaches to
modeling of Web services, we are focused on the In order to have these rules represented using
business rules describing how particular services our modeling notation, and also to be able to relate
are used, but without explicitly stating that we are the rules with the elements of vocabularies, we
talking about Web services. This approach enables model these rules by using URML. It is important
us, not only to focus on the problems under study to stress that all rules-based systems define busi-
and the underlying business logic regulating the ness rules on top of business vocabularies. Thus,
use of Web services, but also now we are able to the URML graphical notation enables us to define
translate such Web service modeling to both Web business rules regulating the use of Web services
service languages and rule-based languages that by leveraging a human-comprehendible, and yet
can regulate how services are used at run-time. formally consistent representation, with the un-
Here, we describe how the in-out MEP is derlying business vocabularies (i.e., UML class
modeled in URML. The in-out MEP consists of models). These abovementioned two reaction rules
exactly two messages: when a service receives represented in URML are shown in Figure 5.
an input message, it has to reply with an output In this particular case, we have the following
message. This means that a fault message must situation: triggering event of either rule (i.e.,
be delivered to the same target node as the mes- UserMessageArrival) maps to the input message
sage it replaces, unless otherwise specified by an of the Web service operation. The action of the
extension or binding extension. To show how this first reaction rule (i.e., EchoStringResponse),
variation of the in-out MEP can be modeled by which is triggered when a condition is true, maps
using reaction rules, we start our discussion from to the output message of the Web service opera-
an example of the in-out MEP with an out-fault tion. The action of the second reaction rule (i.e.,
message. Let us consider the following business EchoStringError), triggered on a false condition,
rule: When a user sends a message, if the length maps to the out-fault message of the Web service
of the message is less than five characters long operation.
return the message to the user, otherwise return As shown, a web service operation may be mod-
an error message. eled by using more than one rule, where rules, that
The business rule is an in-out MEP and can be define an operation, are grouped in one package by
represented by two reaction rules, represented in the name of the operation. Otherwise, for model-
semi-formal pseudo rule syntax: ing the basic in-out MEP, only one (first) reaction
rule would be used. Besides the classes that model
exchanged messages, the URML diagrams also

486
Modeling of Web Services using Reaction Rules

Figure 6. Strelka’s functionality

model conditions (e.g., echoInParam.length < 5). The first part of Figure 6 (leftmost dashed box)
Such condition constructs are modeled by using shows us two important features Fujaba has: it is
OCL filters. OCL filters are based on a part of OCL capable of generating Java code from diagrams,
that models logical expressions, which can be later and it offers easily extensible architecture suitable
translated to R2ML logical formulas, as parts of for plugin implementation. This box also symboli-
reaction rules. However, these OCL filters cannot cally shows that one of the steps that need to be
be later translated to Web service descriptions undertaken when creating a plugin for Fujaba is
(e.g., WSDL), as those languages cannot support to create a metamodel for the new elements Fu-
such constructs. But, we can translate our URML jaba will be extended with. After this first box,
models into rule-based languages (e.g., Jess or in the figure, we have an OCL parser. The OCL
Drools). This means that for each Web service, parser, which we used in this version of Strelka,
we can generate a complementary rule, which is developed at the Technical University of Dres-
fully regulates how the service is used. Another den in Germany2. This parser is widely accepted
benefit of this approach is that our generated Web open source tool – besides Fujaba, it can be also
services, and their regulating rules, are consistent, integrated into other tools: ArgoUML, Borland
as they originate (i.e. they are generated from) the Together, Poseidon. The second dashed box on
same rule-based Web service model. Figure 6 shows that we use JAXB in the process
of mapping URML to R2ML. JAXB guaranties
that the R2ML XML document we get as the
4. Strelka: Our Modeling result of serialization of URML diagram created
Environment in Strelka, is valid according to the R2ML XML
schema. The reason we use JAXB, is that Fujaba
In this section, we describe Strelka – a tool that does not have explicitly defined metamodel in
was implemented as a plugin for the Fujaba tool1. some of the metamodeling languages (e.g. MOF,
Fujaba is a UML modeling tool, developed at the or Ecore). So, we could not use model transforma-
University of Paderborn in Germany. Fujaba al- tions (and ATL), in the mapping process between
lows creation of different types of UML diagrams. URML and R2ML like we did in the process of
It allows generation of Java code based on created mapping between R2ML and WSDL (this map-
model, and it offers a plugin interface that we ping we describe in the next section). And the last
used for Strelka implementation. Figure 6 shows, dashed box on Figure 6 shows that Strelka has an
symbolically, the architecture of Strelka, as well integrated support for calling model transforma-
as the functionality that it provides. tions between R2ML and WSDL.

487
Modeling of Web Services using Reaction Rules

Figure 7. Strelka’s rule metamodel

As already mentioned, one of the steps in the 5. TRANSFORMING WEB


process of creating plug-ins for Fujaba is defin- Service models
ing a metamodel for all the new elements that we
extend Fujaba with. This metamodel is created in In this section, we summarize the translation pro-
Fujaba, and then Java code is generated from it. cess from URML reaction rules to Web services,
This Java code we use afterwards, in the process and then we describe how this URML model can
of generating plugin itself. As the process of be transformed to Drools rule language.
defining metamodel, and Java code generation The process of getting WSDL documents from
is complete, we can create class diagrams that URML models is as follows: first, we create an
can incorporate those newly created elements, as URML diagram in Strelka, after this we serial-
well as UML elements that were already available ize it in the R2ML XML concrete syntax (this
(e.g. classes and associations). Figure 7 presents functionality is built in Strelka, and it relies on
fragment of the Rule metamodel created in Fu- the use of JAXB). Once URML reaction rules are
jaba. Classes that are relevant to the modeling of serialized to R2ML XML, they are translated into
reaction rules are: ReactionRule, MessageAction, WDSL documents (this translation is done at the
EventCondition, MessageEventType, and Fault- level of metamodels as described below). It is
MessageEventType. Figure 7 shows relations important to say that this process allows for two
between the ReactionRule class to the other classes way transformation between URML and WSDL,
in this metamodel. The ReactionRule class is a that is, it is possible to perform reverse engineer-
subclass of Rule class (just like IntegrityRule, ing of existing services, and extract business rules
ProductionRule, and DerivationRule classes, already integrated to service implementation.
which are used for modeling integrity, production, To support this approach, we have implemented
and derivation rules respectively). a number of transformations between different
languages and their representations (these trans-
formations can be found on the ATL ZOO page
(WSDL2R2ML, 2006):

488
Modeling of Web Services using Reaction Rules

• URML and R2ML XML concrete syntax. defining message types and vocabularies, and the
As previously stated, this is the only trans- R2ML vocabulary).
formation that is not implemented by using Figure 9 shows an example of a matched rule
ATL (ATL, ver. 0.7), but instead it is based from our (WSDL2R2ML.atl) transformation (for
on JAXB the Input class in the WSDL metamodel), and
• R2ML XML-based concrete syntax and Figure 10 shows a graphical presentation of this
R2ML metamodel. This transformation is mapping (for this particular class) between WSDL
important to bridge concrete (XML) and and R2ML metamodel. This matched rule initial-
abstract (MOF) syntax of R2ML. This izes “sender”, “objectVariable” and “type” features
is done by using ATL and by leveraging of the target R2ML MessageEventExpression
ATL’s XML injector and extractor. element (we are putting a string “to_be_defined”
• R2ML metamodel and WSDL metamodel. in the “sender” feature, as the user will provide
This transformation is the core of our so- this information if required. This feature refers to
lution and presents mappings between the requester of the service).
R2ML and WSDL at the level of their ab- As the result of the execution of WSDL2R2ML.
stract syntax. atl transformation, we get an R2ML model. This
• WSDL XML-based concrete syntax and model (which is an instance of the R2ML meta-
WSDL metamodel. This transformation is model) can later be serialized into the R2ML
important to bridge concrete (XML) and XML concrete syntax, by our R2ML2XML.atl
abstract (MOF) syntax of WSDL. This transformation and the use of the XMLExtractor
is done by using ATL and by leveraging class.
ATL’s XML injector and extractor. The WSDL2R2ML.atl transformation is
defined as a sequence of ATL rules and ATL
Here, we describe the third and the forth helpers. It has 11 matched rules, 3 lazy rules and
transformation: i.e. transformation between 3 helpers.
R2ML metamodel and WSDL metamodel, and Transformation, in the opposite direction, from
transformation between WSDL XML schema and R2ML metamodel elements to WSDL metamodel
WSDL metamodel. elements (R2ML2WSDL.atl) is based on the same
conceptual mappings presented in the Table 1.
Transformation Between WSDL This transformation uses 9 matched rules, 4 lazy
Metamodel and R2ML Metamodel rules, 1 unique lazy rule and 5 helpers. Figure
11 shows one matched rule from R2ML2WSDL
Figure 8 shows the general principle of mappings transformation. This rule is mapping R2ML Re-
between WSDL metamodel and R2ML metamodel actionRuleSet element into WSDL Service and
(as can be seen on the figure, these mappings can Interface elements.
be done in both directions).
Process of transforming WSDL metamodel Transformation of WSDL XML
elements to R2ML metamodel elements is con- Schema to WSDL Metamodel
ducted in the MOF technical space (TS) (Kurtev
et al., 2002). Table 1 shows a conceptual mapping Because the concrete syntax of WSDL is defined
between these metamodels (this table does not in the XML technical space, and its abstract syntax
show parts related to the mapping between the (metamodel) is defined in the MOF TS, it is neces-
XML Schema language, which WSDL uses for sary to make a translation from concrete syntax

489
Modeling of Web Services using Reaction Rules

Figure 8. Mappings between models that conform to WSDL metamodel to models that conform to R2ML
metamodel

elements to the abstract syntax elements. For this WSDL XML concrete syntax and WSDL abstract
translation we have used XML metamodel (XML syntax. Figure shows that the transformation
metamodel, 2006) as a bridge between XML and can be done in both directions (i.e. we created
MOF TS. bidirectional transformations).
To transform an XML document from XML Before we describe transformation process
TS to the XML model (instance of a XML between WSDL XML schema and WSDL meta-
metamodel) of the same rule in the MOF TS, model, we stress a fact that because WSDL does
we have used the XMLInjector class. By using not define its own language for defining data
a SAX parser, this class creates an XML model types, we are referring to the XSD metamodel
from an XML file. The next figure (Figure 12) that can be obtained from the Eclipse Model
shows general principle of mapping between Development Tool project3.

Table 1. An excerpt of the mapping between the WSDL metamodel and the R2ML metamodel

WSDL metamodel R2ML metamodel


Description RuleBase
ElementType Vocabulary
Interface ReactionRuleSet
Operation ReactionRule
Input MessageEventExpression
Infault MessageEventExpression
Output MessageEventExpression
Outfault MessageEventExpression

490
Modeling of Web Services using Reaction Rules

Figure 9. ATL matched rule that transforms WSDL Input element to R2ML MessageExpression ele-
ment

Transformation from WSDL XML schema to this process we get XML model that can be
the WSDL metamodel has two primary steps: presented in the XML XMI format (step 2
on the Figure 12).
• Step 1: This step is about transform- • Step 2: This step is about transforming
ing WSDL document from XML TS into XML model (EchoService_XML on the
MOF TS. In this step WSDL document Figure 12) into the WSDL based model
(EchoService.xml on the Figure 12) is (EchoService_WSDL on the Figure 12).
transformed into the WSDL metamod- ATL transformation XML2WSDL.atl
el that represents abstract syntax of the (step 3 on the Figure 12) is performing
WSDL language in the MOF TS. By us- this mapping. The output WSDL model
ing XMLInjector class, WSDL XML docu- (EchoService_WSDL) conforms to the
ment is transformed into model that con- WSDL metamodel. The XML2WSDL.atl
forms to the MOF based XML metamodel transformation is executed on the model
(step 1 on the Figure 12). As the result of level (M1 level), by using information

Figure 10. Graphical presentation o mapping shown on the Figure 9

491
Modeling of Web Services using Reaction Rules

Figure 11. ATL matched rule that transforms R2ML ReactionRuleSet element to WSDL Interface ele-
ment

about elements from the (XML and WSDL) Figure 13 shows a matched rule from XM-
metamodel level (M2 level). L2WSDL.atl transformation for the “descrip-
tion” element of the XML metamodel. ATL rule
The next table (Table 2) shows fraction of Description transforms Element of the XML
mappings between WSDL XML schema and model (EchoService_XML) to Description (i.e.
WSDL metamodel. WSDL!Description) of the WSDL metamodel.

Figure 12. Transformation scenario: WSDL XML format into WSDL metamodel

492
Modeling of Web Services using Reaction Rules

Table 2. Fraction of transformation between WSDL XML schema and WSDL metamodel

WSDL XML schema XML metamodel WSDL metamodel Description


description Root name=’description’ Description The Root element that contains
types, binding, service and interface
elements
interface Element name=’interface’ Interface This element presents WSDL inter-
face, it contains operation and fault
elements
operation Element name=’operation’ Operation This element presents WSDL op-
eration, it contains in/outfault and in/
output elements

This rule initializes “types”, “interface”, “bind- Transformation of the


ing” and “service” attributes of Description target WSDL Metamodel into the
model element by calling an ATL helper shown WSDL XML Schema
on the Figure 14.
The XML2WSDL transformation has 18 Transformation process from WSDL abstract
matched rules, one lazy rule, and 17 helpers (we syntax (i.e. WSDL metamodel) into the WSDL
have this number of helpers because we use helpers XML concrete syntax also contains two primary
for mapping enumeration elements to their string steps.
representation – e.g. in-out MEP into the ‘http://
www.w3.org/2006/01/wsdl/in-out’ string). The • Step 1: Transformation of WSDL mod-
biggest problem in the XML2WSDL transfor- el into XML model. Here a WSDL
mation was mapping of XML schema complex model (EchoService_WSDL on Figure
types. For solving this problem we found the 12) is transformed into XML model
(EclipseDoc, 2002) document very helpful. (EchoService_XML) by using the ATL
transformation WSDL2XML.atl (step 5 on
the Figure 12). After this transformation is
executed on the input WSDL models, XML

Figure 13. Description matched rule

493
Modeling of Web Services using Reaction Rules

Figure 14. Helpers we use in the rule in Figure 13

models (EchoService_XML) are placed in When a user sends a request for Social Insurance
the model repository (EchoService_XML. Number (SIN), if the authentication is successful
xmi on the Figure 12). send the SIN to the user.
• Step 2: Transformation from MOF TS
into the XML TS, where an XML model This business rule is presented with one reac-
(EchoService_XML on the Figure 12) that tion rule that has a triggering event SINrequest,
conforms to the MOF based XML meta- condition Person.Address=SINrequest.Address
model (and that is created in the step 1), AND Person.Name=SINrequest.Name, and ac-
is transformed into the EchoService.xml tion SINresponse:
document (step 7 on the Figure 12).

In the WSDL2XML transformation, we have ON SINrequest (Address, Name)


17 matched rules, 3 lazy rules, as well as 5 helpers. IF Person.Address=SINrequest.
The WSDL2XML transformation (i.e. transfor- Address AND Person.
mation from WSDL metamodel into the WSDL Name=SINrequest.Name
XML schema) is simpler then XML2WSDL trans- DO SINresponse(SIN)
formation (WSDL elements are directly mapped
to their corresponding representation from the An URML diagram that models the corre-
WSDL XML schema, and the other way around sponding Web service operation is presented in
is also true). Figure 15, and Figure 16 illustrates the R2ML
presentation of it that Strelka generated for us.
Deploying URML Models on The first step in getting Drools rules, from the
a Rule-Based Engine generated R2ML file is transformation of R2ML
reaction rules to R2ML production rules. The
Here, we describe the transformation process second step is transformation of R2ML produc-
from a URML model to the Drools rule engine. tion rules to Drools rules, where we are using
For this purpose we are using an example of the the translation provided by (Giurca & Werner,
in-out MEP, presented with the following busi- 2006).
ness rule: In order to successfully realize the first step
one must address the major difference between
reaction rules and production rules: reaction rules

494
Modeling of Web Services using Reaction Rules

Figure 15. URML model of the in-out MEP

have triggeringEvent element, while production the condition element in production rule. Figure
rules do not have this element. So there is a need 17 shows the corresponding R2ML production
for representing triggeringEvent in production rule document that we get as the result of our
rules. If only atomic events are considered, it is transformation.
legible to map the triggeringEvent element to As Figure 17 shows, the producedAction part

Figure 16. R2ML serialization of URML model from Figure 15

495
Modeling of Web Services using Reaction Rules

Figure 17. Generated R2ML production rule document – part of rule containing conditions is shown on
left – vocabulary part can be seen on the right

of the generated production rule is the same as sponse is fired as the response to the information
in the reaction rule, while the conditions part is requested.
constituted from conditions of a reaction rule along Table 4 presents the concept of transformation
with the triggeringEvent of reaction rule. Also, a between R2ML production rule elements and the
vocabulary part of production rule is updated with Drools generated code. The table is not present-
the adequate class generated from triggeringEvent ing all steps of transformation as some of them
of reaction rule. (i.e. if no such class is already are not a simple mapping between elements but
defined. This situation happens with the presenta- gives a general idea of what is done.
tion of rules in older versions of R2ML). Table 3
provides a brief overview of the transformation
from Reaction Rules to Production Rules. 6. Related Work
Once we have generated a R2ML production
rule from the R2ML reaction rule, we use the In this section, we compare the proposed approach
transformation provided by (Giurca & Werner, with some relevant solutions to the modeling of
2006) to transform R2ML production rule to a Web services that are based on the MDE prin-
Drools rule. Figure 18 shows the generated Drools ciples.
rule from the production rule presented in Figure (Bezivin et al., 2004) demonstrate how one can
17. The left part of this figure shows the classes take advantage of MDE to develop e-business ap-
generated for Drools engine, and the right part plications. In their approach, they start from UML
of the figure shows the rule itself. Each rule fires and the UML profile for Enterprise Distributed
an action when conditions are met - in the case Object Computing (EDOC) to define platform-
below, SINrequest trigger this rule, while SINre- independent models of e-business applications.

496
Modeling of Web Services using Reaction Rules

Table 3. Presenting the basic of transformation from reastion rule to production rule

Reaction Rule Production Rule


ReactionRuleSet ProductionRuleSet
ReactionRule ProductionRule
TrigerringEvent & Conditions Conditions (There will be only one Conditions element generated although both TrigerringEvent
and Conditions could be present in the Reaction Rule)
MessageEventExpression ObjectClassificationAtom

In the next step, they translate such models into business processes, while OCL filter expressions
models that are based on metamodels of Java, can even further specify conditions under which
WSDL, and JWSDP. Although this approach uses a message can happen. This is very useful for
ATL, similar to what we do, it does not provide potential integration of business process based on
two way transformations between models and the use of Web services (Milanovic et al., 2006).
service implementation. This is unlike our solu- Of course, such filter expressions overcome the
tion, because we support two-way transformations, potentials of the Web service technology, as they
and thus we enable reverse engineering of the presume that there should also be a rule-based
existing Web services. Next, the translation of engine, which is able to interpret such conditions.
regular UML models, i.e., UML class related ele- However, the research on semantic Web services
ments into WSDL is limited, as one does not have (Sheth et al., 2006) and Web rules (Nagl et al.,
enough expressivity in UML diagrams to define 2006) demonstrates that this issue attracts more at-
all details of WSDL (i.e., one can not distinguish tention in the Web service research community.
between input and output messages). This issue (Vara et al., 2005) define a metamodel for
is addressed by using the UML profile for EDOC, WSDL and its corresponding UML profile that is
but this approach is more focused on modeling used for modeling of Web services in the MIDAS-
distributed components rather on modeling busi- CASE MDA-based tool for Web information
ness logics. In our approach, the use of reaction system development. They also support automatic
rules in URML models enables us to be closer to generation of the respective WSDL description for

Figure 18. Generated Drools rule (right) and generated class of it (left)

497
Modeling of Web Services using Reaction Rules

Table 4. Presenting transformation concept of R2ML production rule to Drools

R2ML Drools
Class Class
Attribute Integer, String etc (Based on the value of r2mlv:ID)
Conditions Conditions (and based on the object within this element the condition element will be generated)
producedAction Actions (based on the elements residing within this element proper code with regards to the previously
generated classes will be generated)

the Web services that is being modeled. Although from their UML definitions (i.e., OCL) to the
the MIDAS framework supports platform-inde- languages of the potential target platforms (e.g.
pendent models of services and service composi- Jess). This approach does not consider modeling
tions, their definitions are very incomplete and usage patterns or MEPs or error handling (i.e.,
one can hardly generate complete service models business logic) like we do, but instead it focuses
automatically. In our approach, we do not strictly on the details specific for service platforms.
base models of services on workflows in which (Manolescu, et al., 2005) propose a model-
services will be used, as we wanted to focus on driven approach to designing and deploying service-
how services are used (i.e., MEPs). enabled web applications. This work is different
(Gronmo et al., 2005) and (Timm & Gannod, from the ones above in the following aspects: i) it
2005) propose an approach to modeling semantic extends WebML, a well-known modeling language
Web services by using UML profiles. Gronmo et for developing Web applications; ii) it uses E-R
al. have the goal to define platform-independent models for defining data types; and iii) it focuses
models of services, which can be translated to on the usage patters (MEPs) of Web services in
semantic Web services (OWL-S and WSMO). In order to define WebML extensions. Their modeling
their work, they abstract concepts from OWL-S approach is based on the use of MEPs, but they are
considering MEPs used in WSDL 1.1, and for each
and WSMO, and extend UML activity diagrams
MEP they define corresponding WebML primi-
(i.e., they define a UML profile) accordingly, while
tive. These new WebML primitives can be used
for defining vocabularies they use the Ontology
in hypertext models of Web applications. This ap-
Definition Metamodel, and its corresponding
proach is the most similar to ours; both approaches
UML profile defined in (Gasevic et al., 2005).
fully focus on modeling business processes and
Since both these approaches use XSLT, they can potential usage patterns. However, the WebML
hardly translate service pre- and post-conditions approach does not explicitly define a Web service

Table 5. Summary of solutions for modeling of Web services based on the MDE principles

Approach Used metamodel Model description transformation output


(Timm & Gannod, 2005) UML profile UML XMI XSLT OWL-S
(Bezivin et al., 2004) MOF MOF XMI ATL WSDL, Java, JWSDP
(Manolescu, et al., 2005) Extended WebML WebML model XSLT (through visual WebRatio output
editor)
(Vara et al., 2005) UML profile UML XMI MIDAS-CASE transfor- MIDAS output
mation unit
Our approach MOF MOF XMI ATL R2ML, WSDL

498
Modeling of Web Services using Reaction Rules

description, but it infers it from the definitions By leveraging MDE, we have defined a rule-
of hypertext models in which the service is used, based modeling language that can be managed by
which makes this process context dependent. That universal MDE tools. The use of model transforma-
is, services are defined inside specific WebML tions allows for transforming platform independent
workflows unlike our approach where we define models of business logic to specific platforms such
workflow-independent services. In addition, they as Web services. Moreover, for each Web service,
do not consider the use of preconditions of Web we can also generate a rule that will fully regulate
services, which is important for the reusability of the behavior of the service (i.e., how the service is
services in different contexts. Although WebML has used), and thus make sure that the business logic is
support for exception handling (Brambilla et al., fully followed. In this paper, we have not explained
2005), this is also focused on the workflow level that part of the solution, but we are going to report
rather than on a service level. Finally, the WebML on that in our future papers.
approach does not support reverse engineering of Our solution has much broader potentials that
Web services. overcome the pure translation between rule-based
The following table (Table 5) summarizes these business models and Web services. Unlike the
approaches of modeling of Web services, and WSDL definition of Web services, our models
shows their relation to our approach we describe also have an option for defining pre- and post-
in this paper. conditions of services. However, WSDL can not
express pre- and post-conditions in the descriptions
of Web services. That is, Web service tools can not
7. Conclusion automatically support conditions under which some
services can be used (as defined by OCL filters in
In this paper, we have demonstrated how the use of Section 4). To address this problem, we expand
MDE principles can enable for rule-based modeling our approach on W3C’s Semantic Annotations
of Web services. By using the MDE principles, we for WSDL (SAWSDL) recommendation. In our
have been able to develop a framework for modeling current activities, we have developed transforma-
Web services from the perspective of how services tions from our R2ML reaction rule-based models
are used in terms of message-exchange patterns to several production rule languages (e.g., Drools,
(MEPs). Our approach enables developers to focus Jena2, and Jess). Our particular focus is on Drools,
on the definition of business rules, which regulate as the use of Drools allows us to directly enforce
MEPs, instead of focusing on low level Web service business rules to regulate the use of Web services
details or on contexts where services are used (i.e., deployed on JBoss’ application server.
workflows). Business rules are important assets
of a business organization that embody valuable
domain knowledge. They are stated in a way, which References
is very close to how users think and talk. Due to
the declarative nature of rules, the business logic Arroyo, S., Toma, I., Roman, D., Drumm, C.,
can easier be updated without the need to change Dimitrov, M., Spork, M., et al. (2004). Report
to whole system. Furthermore, integrating business on state of the art and requirements analysis
rules in process-oriented Web service composition (Deliverable 3.1). DIP EU FP6 Integrated Proj-
can improve the quality and ease development. ect. ATLAS group, Nantes. (ver. 0.7) (n.d.).
Once business rules are expressed explicitly as Atlas transformation language - User manual.
first-class entities, they can be reused across several Retrieved from http://www.eclipse.org/gmt/atl/
compositions i.e., the same rule can be applied to doc/ATL_User_Manual[v0.7].pdf
many web service compositions.

499
Modeling of Web Services using Reaction Rules

Bézivin, J. (2005). On the unification power of EclipseDoc. (2002). How abstract components
models. Software and System Modeling, 4(2), are created when reading a schema – an anima-
171–188. doi:10.1007/s10270-005-0079-0 tion in 9 slides (complexType). Retrieved from
http://help.eclipse.org/help33/index.jsp?topic=/
Bezivin, J., Hammoudi, S., Lopes, D., & Jouault, F.
org.eclipse.xsd.doc/references/diagrams/images/
(2004). Applying MDA approach for Web Service
xsd-diagram-ct02/img0.html
platform. In Proc. of the 8th IEEE Int. Conf. on
Enterprise Distributed Object Computing Conf. Erl, T. (2005). Service-oriented architecture: Con-
(pp. 58–70). cepts, technology, and design. Prentice Hall.
Brahe, S., & Osterbye, K. (2006). Business process Falkovych, K., Sabou, M., & Stuckenschmidt,
modeling: Defining domain specific modeling H. (2003). UML for the Semantic Web: Trans-
languages by use of UML profiles. In ECMDA-FA formation-based approaches. In B. Omelayenko
2006 (LNCS 4066, pp. 241-255). 2006. & M. Klein (Eds.), Knowledge transformation
for the Semantic Web (pp. 92-106). Amsterdam:
Brambilla, M., Ceri, S., Comai, S., & Tziviskou,
IOS Press.
C. (2005). Exception handling in workflow-driven
Web applications. In Proc. of the 14th International Gasevic, D., Djuric, D., & Devedzic, V. (2005).
WWW Conference (pp. 170–179). Bridging MDA and OWL ontologies. Journal of
Web Engineering, 4(2), 119–134.
Charfi, A., & Mezini, M. (2004). Hybrid Web
service composition: Business processes meet Ginsberg, A., Hirtle, D., McCabe, F., & Patran-
business rules. In Proc. of the 2nd International jan, P. (2006). RIF Use cases and requirements.
Conference on Service Oriented Computing (pp. W3C Working Draft. Retrieved from http://www.
30-38). w3.org/TR/rif-ucr.
Chinnici, R., Moreau, J., Ryman, A., & Weer- Giurca, A., Lukichev, S., & Wagner, G. (2006).
awarana, S. (2007). Web Services Description Modeling Web services with URML. In Proc.
Language (WSDL) Ver. 2.0 Part 1: Core Language. of Workshop Semantics for Business Process
W3C Recommendation. Retrieved from http:// Management.
www.w3.org/TR/2007/REC-wsdl20-20070626.
Giurca, A., & Werner, J. (2006). Translator: R2ML
Clement, L., Hately, A., Riegen, C., & Rogers, T. to JBoss rules. Retrieved from http://oxygen.
(2004). UDDI Ver. 3.0.2. Retrieved from http:// informatik.tu-cottbus.de/translator/R2MLtoJ-
www.oasis-open.org/committees/uddi-spec/doc/ BossRules/
spec/v3/uddi-v3.0.2-20041019.htm.
Gronmo, R., Jaeger, M. C., & Hoff, H. (2005).
Coyle, F. (2002). XML, Web services, and the data Transformations between UML and OWL-S. In
revolution. Addison Wesley. Proc. of the 1st European Conf. on Model Driven
Architecture - Foundations and Applications (pp.
Czarnecki, K., & Helsen, S. (2006). Feature-based
269–283).
survey of model transformation approaches. IBM
Systems Journal, 45(3), 621–645. Gudgin, M., Hadley, M., Mendelsohn, N., Moreau,
J., Nielsen, H. F., Karmarkar, A., & Lafon, Y.
(2007). SOAP Ver. 1.2 Part 1: Messaging frame-
work. W3C Recommendation. Retrieved from
http://www.w3.org/TR/soap12-part1/

500
Modeling of Web Services using Reaction Rules

Hirtle, D., Boley, H., Grosof, B., Kifer, M., Sin- Miller, J., & Mukerji, J. (Eds.). (2003). MDA
tek, M., Tabet, S., & Wagner, G. (2006). Schema Guide Version 1.0.1. OMG.
specification of RuleML 0.91. Retrieved from
MOF. (2005). Object Management Group, Meta
http://www.ruleml.org/spec/
Object Facility (MOF) Core, v2.0. OMG Docu-
Horrocks, I., Patel-Schneider, P. F., Boley, H., Ta- ment formal/06-01-01. Retrieved from http://
bet, S., Grosof, B., & Dean, M. (2004). SWRL: A www.omg.org/cgi-bin/doc?formal/2006-01-01.
Semantic Web rule language combining OWL and
Nagl, C., Rosenberg, F., & Dustdar, S. (2006).
RuleML. W3C Member Submission. Retrieved
VIDRE– A distributed service oriented business
from http://www.w3.org/Submission/SWRL/
rule engine based on RuleML. In Proc. of the 10th
Jovanović, J., & Gašević, D. (2005). XML/ IEEE International Enterprise Distributed Object
XSLT-based knowledge sharing. Expert Systems Computing Conference (pp. 35-44).
with Applications, 29(3), 535–553. doi:10.1016/j.
Object Management Group. (2008). Business
eswa.2005.04.024
Process Model and Notation (BPMN) specification
Kurtev, I., Bezivin, J., & Aksit, M. (2002). Techno- 2.0. Retrieved from http://www.omg.org/cgi-bin/
logical spaces: An initial appraisal. In Proceedings doc?bmi/08-02-06
of the Confed. Int’l Conferences CoopIS, DOA,
OCL (2006). Object constraint language.
and ODBASE.
OMG Available Specification, Version 2.0, for-
Lewis, A. (2007). Web Services Description mal/06-05-01.
Language (WSDL) Version 2.0: Additional MEPs.
Proctor, M., Neale, M., Frandsen, M., Griffith, S.,
W3C Working Group Note. Retrieved from http://
Jr., Tirelli, E., Meyer, F., & Verlaenen, K. (n.d.).
www.w3.org/TR/2007/NOTE-wsdl20-additional-
Drools documentation. Retrieved from http://
meps-20070626/
downloads.jboss.com/drools/docs/4.0.7.19894.
Lukichev, S., & Wagner, G. (2006). UML-based GA/html/index.html
rule modeling with Fujaba. In Proc. of the 4th
Ribaric, M., Gasevic, D., & Milanovic, M. (n.d.).
International Fujaba Days (pp. 31-35).
R2ML: A rule-based approach to modeling of
Lukichev, S., & Wagner, G. (2006a). Visual rules semantically-enriched Web services.
modeling. In Proceedings of the 6th Inter-national
Rosenberg, F., & Dustdar, S. (2005). Business
Conference Perspectives of Systems Informatics
Rules Integration in BPEL – A service-oriented
(pp. 467-673).
approach. In Proc. of the 7th International IEEE
Manolescu, I., Brambilla, M., Ceri, S., Comai, S., Conference on E-Commerce Technology (pp.
& Fraternali, P. (2005). Model-driven design and 476-479).
deployment of service-enabled Web applications.
Rupp, N. A. (n.d.). The login of the bottom line:
ACM Transactions on Internet Technology, 5(3),
An introduction to The Drools Project. Retrieved
439–479. doi:10.1145/1084772.1084773
from http://www.theserverside.com/tt/articles/
Milanovic, M., Gasevic, D., Giurca, A., Wagner, article.tss?l=Drools
G., & Devedzic, V. (2006). On interchanging
SBVR. (2006). Object Management Group. Se-
between OWL/SWRL and UML/OCL. In Proc.
mantics of business vocabulary and business rules
of the OCLApps Workshop (pp. 81-95).
(SBVR) – Interim specification. Retrieved from
http://www.omg.org/docs/bei/05-08-01.pdf

501
Modeling of Web Services using Reaction Rules

Schneier,B.(1992). The Rete matching algorithm: Key Terms and definitions


A description of the pattern-matching algorithm
underlying Drools. Business Rules: Business rules or business
rule sets describe the operations, definitions
Seidewitz, E. (2003). What models mean. and constraints that apply to an organization in
IEEE Software, 20(5), 26–32. doi:10.1109/ achieving its goals.
MS.2003.1231147 Drools: A business rule management system
Sheth, A., Verma, K., & Gomadam, K. (2006). (BRMS) with a forward chaining inference based
Semantics to energize the full services spec- rules engine, more correctly known as a production
trum. Communications of the ACM, 49, 55–61. rule system, using an enhanced implementation
doi:10.1145/1139922.1139949 of the Rete algorithm.
Reaction Rules: Also known as Event-
Tabet, S., Wagner, G., Spreeuwenberg, S., Vin- Condition-Action Rules, are three components
cent, P., Jacques, G., de Sainte Marie, C., et al. rules i.e. the event part specifies the signal that
(2007). Object Management Group: Production triggers the invocation of the rule, the condition
Rule Representation (PRR). Retrieved from http:// part is a logical test that, if satisfied or evaluates
www.w3.org/2004/12/rules-ws/paper/53/ to true, causes the action to be carried out and the
Timm, J., & Gannod, G. (2005). A model-driven action part consists of updates or invocations on
approach for specifying Semantic Web services. the local data.
In Proc. of IEEE International Conference on R2ML: REWERSE I1 Rule Markup Lan-
Web Services (pp. 313-320). guage, a rule markup language allowing rules
interoperability.
UML (2005). Object Management Group, Unified URML: UML-based Modeling Language, a
Modeling Language 2.0. Docs. formal/05-07-04 visual language for rules modeling.
& formal/05-07-05. Web Service: A software system designed
Vara, J., Castro, V., & Marcos, E. (2005). WSDL to support interoperable machine-to-machine
Automatic Generation from UML Models in a interaction over a network.
MDA Framework. International Journal of Web WSDL: Web Service Description Language
Services Practices, 1(1-2), 1–12. is a language for describing both the abstract
functionality of a service and the concrete details
Wagner, G., Giurca, A., & Lukichev, S. (2006). of a Web Service.
R2ML: A general approach for marking up rules.
In Dagstuhl Seminar Proc. 05371.
WSDL2R2ML (2006). The WSDL to Endnotes
R2ML transformation scenario. Retrieved 1
http://www.Fujaba.de
f r o m h t t p : / / w w w. e c l i p s e . o rg / m 2 m / a t l / 2
http://dresden-ocl.sourceforge.net/
atlTransformations/#WSDL2R2ML 3
http://www.eclipse.org/xsd/s
XML metamodel (2006). Retrieved from http://
mdr.netbeans.org/mdrxml.html

This work was previously published in Handbook of Research on Emerging Rule-Based Languages and Technologies: Open
Solutions and Approaches, edited by A. Giurca, D. Gasevic, and K. Taveter, pp. 422-446, copyright 2009 by Information Sci-
ence Reference (an imprint of IGI Global).

502
503

Chapter 2.14
A Semantic Web-Based
Approach for Building
Personalized News Services
Flavius Frasincar
Erasmus University Rotterdam, The Netherlands

Jethro Borsje
Erasmus University Rotterdam, The Netherlands

Leonard Levering
Erasmus University Rotterdam, The Netherlands

Abstract INTRODUCTION

This article proposes Hermes, a Semantic Web- The simplicity, availability, reachability, and
based framework for building personalized news reduced exploitation costs have made the Web
services. It makes use of ontologies for knowledge one of the most common platforms for infor-
representation, natural language processing tech- mation publishing and dissemination. This is
niques for semantic text analysis, and semantic particularly true for news agencies that use Web
query languages for specifying wanted informa- technologies to present emerging news regarding
tion. Hermes is supported by an implementation different types of events as for example business,
of the framework, the Hermes News Portal, a tool cultural, sport, and weather events. Most of this
which allows users to have a personalized online information is published as unstructured text that
access to news items. The Hermes framework and is made available to a general audience by means
its associated implementation aim at advancing the of Web pages.
state-of-the-art of semantic approaches for per- The heterogeneity of the Web audience and
sonalized news services by employing Semantic the diversity of the published information asks
Web standards, exploiting domain information, for more refined ways of delivering information
using a word sense disambiguation procedure, that would enable users to access news items
and being able to express temporal constraints that interest them. For this purpose the Really
for the desired news items. Simple Syndication (RSS) (Winer, 2003) standard

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
A Semantic Web-Based Approach for Building Personalized News Services

was developed that publishes information in a or Yahoo! Finance are developed to meet these
semi-structured format that supports machine personalization needs by supporting automatic
processing. This format is based on metadata that news filtering on the Web.
(1) associates news items with channels (feeds) Current approaches to news filtering are able
that have properties like categories (e.g., business, to retrieve only the news that explicitly mention
sport, politics, etc.), title, publication date, etc., the companies involved, failing to deliver indirect
and (2) describes news items by means of their information which is also deemed relevant for the
properties as categories (e.g., online business, considered portfolio. For example, for a portfolio
business system, Internet marketing, etc.), release based on Google shares, such systems fail to de-
time, title, abstract, link to the original published liver news items related to competitors of Google,
information, etc. such as Yahoo! or Microsoft, which might have an
Most of the annotations supported by the RSS indirect influence on the share price of Google.
feeds are coarse-grained in the sense that they fail Exploiting the semantic contextual information
to identify the different topics addressed in a cer- related to companies such as its competitors,
tain news item. Also, the current annotations are CEO’s, alliances, products, etc., enables a more
only partially processable by machines as the tags comprehensive overview of relevant news with
do not have unique semantic meaning associated to respect to a certain portfolio.
them and thus have different interpretations. Being Another limitation of existing news filtering
able to understand the semantic content of a news systems is their inability to cope with delivering
item would enable a fine-grained categorization of news items satisfying temporal constraints. The
this information, thus better supporting the users time aspect is of utmost importance when, for
(casual users, media analysts, stock brokers, etc.) example, one considers the fact that news items
information needs. usually have an immediate impact on stock prices,
In order to make the Web data not only ma- or when one desires to do a historical analysis
chine readable but also machine understandable of past news and stock price evolutions. Being
the World Wide Web Consortium proposes the able to exploit the timestamps associated to news
Semantic Web (Berners-Lee, Hendler, & Lassila, items enables retrieving only news that obey user-
2001), a sequence of technologies that allow for determined time-related constraints.
self-describing content. On the Semantic Web In this article we propose the Hermes frame-
metadata is defined using semantic information work, a semantic-based approach for retrieving
usually captured in ontologies. Some of the most news items related, directly or indirectly, to the
popular formats to describe ontologies on the Se- concepts of interests from a domain ontology. In
mantic Web are RDF(S) (Klyne & Carroll, 2004) addition these news items might need to satisfy
(Brickley & Guha, 2004) and OWL (Bechhofer temporal constraints. For illustration purposes
et al., 2004). we focus here on the NASDAQ stock market do-
A special class of users who make daily use of main (Kandel & Marx, 1997), but the genericity
(emerging) news is that of stock brokers. Because of our approach makes it applicable also to other
news messages may have a strong impact on stock domains, as, e.g., tourism or scientific domains.
prices, stock brokers need to monitor these mes- The Hermes News Portal (HNP) is an implemen-
sages carefully. Due to the large amounts of news tation of the Hermes framework, which allows
information published on a daily basis, the manual the user to specify queries for the concepts of
task of retrieving the most interesting news items interest and temporal constraints, and retrieve
with respect to a given portfolio is a challenging the corresponding news items.
one. Existing approaches such as Google Finance

504
A Semantic Web-Based Approach for Building Personalized News Services

For HNP we make use of Semantic Web tech- which attributes define different components, e.g.,
nologies like OWL (Bechhofer et al., 2004) for title, abstract, text, photos, videos, commentaries,
formally defining the semantics of the concepts etc. The system employs a user model initialized
of interest in the ontology. We employ natural using orthogonal stereotypes (interests, domain
language processing (NLP) technologies as, e.g., expertise, cognitive characteristics, and life styles)
lexical analysis, gazetteering, word sense disam- for which the user is asked to provide input and is
biguation, etc., for indexing news items based on further updated using rules that exploit the user
ontology terms. The most popular Semantic Web behavior with the application. Taking into account
query language SPARQL (Prud’hommeaux & the user model, the system builds a presentation
Seaborne, 2008) is used for expressing queries based on relevant news items, each news item be-
using the previously identified concepts. In order to ing shown at an appropriate level-of-detail (based
simplify the representation of temporal constraints on the user model). Differently than SeAN, our
we propose time-related extensions to SPARQL. framework uses standard Semantic Web technolo-
HNP is a generic platform that could easily be gies for representing knowledge and employs NLP
applied to other domains than the financial one. techniques for automatic annotation of news items,
The structure of the article is defined as follows. instead of using a manual approach.
The first section discusses related approaches for YourNews (Ahn, Brusilovsky, Grady, He, &
personalized news services. The second section Syn, 2007) proposes an open and editable user
presents the Hermes framework identifying the model for personalizing news items. The user
proposed methodological steps. The third section model is open in the sense that users can view the
describes HNP, an implementation of the proposed list of keywords stored in the individual profiles.
framework. The last section concludes the article The user model is also editable as it allows users to
and discusses future work. add/delete keywords from their associated profiles.
As an additional feature which also contributes
to the transparency and control over adaptation,
RELATED WORK YourNews shows the key terms present in news
items. The representation of news items is given
Among the methods that aim at personalizing by weighted vectors of terms (Salton, 1971), where
news information we distinguish two types: non- the weights are computed using TF-IDF (Salton
semantic approaches and semantic approaches. & McGill, 1983). The visited news items are used
In the followings we will present short descrip- for building a weighted term vector which is the
tions of two non-semantic methods: Server for user model. The similarity between a news item
Adaptive News and YourNews, and two semantic and the user model is defined by the cosine metric
methods: MyPlanet and SemNews. For each pre- between their associated vectors. This measure
sented method we give the differences compared allows the system to recommend news items that
to our approach and at the end of this section we are considered relevant for the user. Despite the
highlight the main contributions of the Hermes users’ interest to view and edit their profiles, there
framework. is a decrease in performance (e.g., precision, recall,
Server for Adaptive News (SeAN) (Ardissono, etc.) for recommended items compared to the same
Console, & Torre, 2001) enables a personalized system using a closed user model (where the user
access to news servers on the Web. The generated is not able to view/edit the user model). While
views are composed of sections, as in newspapers, YourNews uses a keyword-based approach for
on which customized news items are embedded. modeling news items and user interests, Hermes
The news items are viewed as complex entities in

505
A Semantic Web-Based Approach for Building Personalized News Services

employs a semantic approach based on ontology The contributions that Hermes brings to
concepts for modeling similar aspects. building personalized news services compared
MyPlanet (Kalfoglou, Domingue, Motta, to existing approaches are fivefold. First, Hermes
Vargas-Vera, & Shum, 2001) aims at providing makes a strict distinction between the framework
users with news items relevant for their topics of (Hermes framework) and its implementation
interest. MyPanet is an extension of PlanetOnto, (HNP), allowing for possible different technolo-
an integrated suite of tools used to create, deliver, gies (as these evolve) to be used with the same
and query internal newsletters of the Knowledge framework. Second, Hermes uses an advanced
Media Institute (KMi). Similar to our approach NLP methodology (e.g., tokenization, part-of-
an ontology is used for classifying news items speech tagging, word sense disambiguation, etc.)
and allowing the user select his topics of interest. for news understanding employing a semantic
Nevertheless, the classification process is based lexicon (e.g., WordNet). Third, the implementa-
on the heuristics of cue phrases attached to ontol- tion is based on the most up-to-date Semantic
ogy concepts, while we have a more systematic Web standards (OWL and SPARQL). Fourth, we
approach to classification by employing NLP allow news querying using temporal constraints
techniques (e.g., exploiting the WordNet term syn- by providing temporal extensions to SPARQL.
onyms, performing word sense disambiguation, Fifth, and the last contribution, the user is pro-
etc.) that improve classification results. In addi- vided with a graphical query interface to specify
tion, our implementation is based on the standard the concepts of interest in a direct (using concept
ontology language OWL instead of the specific selections) or indirect manner (using relationship
ontology language, OCML (Motta, 1999), used selections).
in myPlanet. We also did choose to present the
ontology as a graph instead of a tree as it allows
the user to have a more comprehensive overview HERMES FRAMEWORK
of the ontology structure.
SemNews (Java, Finin, & Nirenburg, 2006) The Hermes framework proposes a sequence of
proposes a framework for understanding and steps to be followed in order to build a personal-
querying news items. As in Hermes, the moni- ized news service. The input for the constructed
tored news are made available by RSS feeds. The system comprises RSS news feeds and the output
news items are analyzed by OntoSem (Nirenburg consists of news items fulfilling user needs. The
& Raskin, 2001), SemNews’ natural language Hermes framework is centered around a domain
processing engine. OntoSem converts the textual ontology which is used for indexing news items
representation of news into Text Meaning Repre- and helping the user formulate queries based on
sentation (TMR), a specific format for knowledge his concepts of interest. In addition, the user can
representation. The TMR descriptions are sub- specify temporal constraints that news items need
sequently converted to OWL and published on to satisfy. The resulting news items are sorted
the Web. The OWL news representation can be based on their relevance for the user queries.
used for querying using RDQL (Seaborne, 2004), For illustrative purposes we chose a financial
one of the precursors of the SPARQL query lan- domain example, i.e., a personalized news service
guage. Differently than SemNews, Hermes uses which can help the stock brokers in their daily deci-
a semantic lexicon (e.g., WordNet) for performing sions. More precisely we opted for portfolios based
word sense disambiguation, and allows for a more on stocks of NASDAQ companies. For this purpose
intuitive way of building queries by letting the we developed a domain ontology, which captures
user make his selections in a graphical way. companies, products, competitors, CEO’s, etc. In

506
A Semantic Web-Based Approach for Building Personalized News Services

addition we have developed a news ontology able The architecture of the Hermes framework is
to store news items and their metadata such as described in Figure 1 and is composed of three
title, abstract, time stamp, etc. main steps: news classification, news querying,
The domain ontologies are developed by and results presentation. News classification is
domain experts. The process of developing the responsible for indexing the news items based
ontology is an incremental middle-out approach. on ontology concepts. News querying consists
First the most salient concepts are defined and of two substeps: query formulation, i.e., helping
then these are refined using generalization/spe- the user build the query that expresses the items
cialization towards the top/bottom of the ontology. of interest, and query execution, i.e., computing
As the news information can contain additional the results of query evaluation. In the last step,
concepts not a priori known, the ontology needs results presentation the computed news items
to be regularly maintained by the domain experts. are presented based on their relevance to the
We validated our domain ontology using the On- user interests.
toClean methodology (Guarino & Welty, 2002).
As news items might come from different
RSS feeds, it is possible that the same news item NEWS CLASSIFICATION
has been published through different channels.
After aggregating the news items one needs to The ontology concepts used for news items clas-
remove the duplicate ones. In order to speed-up sification are classes and individuals from the
this process we employed the heuristics to use only domain ontology. Concepts are linked to synsets,
the title for identifying news items, even so we i.e., sets of synonyms, from a semantic lexicon,
acknowledge that in few cases news items might which identify their unique meaning. The syn-
have the same title and still have different content, onyms stored in a synset in the semantic lexicon
i.e., they represent different news items. are used as lexical representations of the associated

Figure 1. Hermes architecture

Semantic
Lexicon

Ontology
News Querying

News News Classification Query Formulation


Items

Query

Classified Query Execution


News Items

News Item 1
Filtered Results News Item 2
News Items Presentation News Item 3
...

507
A Semantic Web-Based Approach for Building Personalized News Services

concept. In this case a lexical representation has a representations of concepts stored in the domain
sense associated with it, i.e., the one given by the ontology. For lexical representation identification
corresponding synset. As the ontology is specific we use the maximal group of words (compound
to a domain, while the semantic lexicon is domain words) found in sequence in a news item that
independent, we associate additional domain stand for a concept’s lexical representation. For
specific lexical representations to the concepts example “European Central Bank” would be
in our ontology. The lexical representations are identified as a compound word representing the
composed only of word lemmas (the canonical ECB concept, e.g., a longer match supersedes a
word form appearing in dictionaries). shorter match.
In addition, for classes without subclasses Each time a lexical representation of a concept
and individuals we decided to consider the hyp- is matched a word sense disambiguation procedure
onyms associated to their corresponding semantic takes place. As the same lexical representation
lexicon synset. For these classes and individuals, can belong to different concepts (present or not in
the domain expert, who devised the ontology, is the ontology) this procedure checks if the match
probably not interested in more refined definitions indeed corresponds to the meaning of the found
of these. However, the lexical representations of ontology concept. If the check is positive a hit is
these concepts can be enlarged by considering stored in the ontology, i.e., a link between the news
also the corresponding hyponyms synsets from item and the corresponding concept is defined.
the semantic lexicon. The hit also stores the found lexical representa-
The classification approach is ontology-centric, tion, as classification evidence.
in the sense that the ontology concepts are loaded For word sense disambiguation we use a
one at-a-time and their lexical representations are variant of the SSI algorithm (Navigli & Velardi,
matched against the news items. This approach 2005). In this process we also consider lexical
is opposed to a news items-centric one where representations for concepts that are not stored
the words in the news items are matched against in the ontology but are present in the semantic
the lexical representations of the concepts from lexicon as these are relevant when computing the
the ontology. We opted for an ontology-centric sense of a found lexical representation from the
approach in order to speed-up the classification ontology. These lexical representations help in
process as in this case we need to traverse the better determining the context of a sentence and
ontology only once. The number of concepts in the thus computing the sense of an ontology-based
ontology is considerably larger than the number lexical representation.
of words in the news items. The algorithm determines, per news sentence,
First the tokenization, sentence splitting, part- the sense of a lexical representation (lex) by com-
of-speech tagging, and morphological analysis are puting the sum of the distances between one of the
performed. The tokenization precedes sentence senses of the considered lexical representation (sj)
splitting as sentence splitting needs the identifica- and the senses of the previously disambiguated
tion of the punctuation signs from tokenization lexical representations from the context sentence
(as “.”, “,”, etc.). Morphological analysis follows (sci). The sense corresponding to the smallest sum
part-of-speech tagging because the lemma of is the chosen one (selectedSense). The algorithm
a word depends on its part-of-speech tag. For starts with monosemous lexical representations
example “reading” as a verb has lemma “read” (i.e., lexical representations which correspond to
but as noun it has the lemma “reading”. In this only one concept) and in case that such representa-
way, all words in a news item are reduced to their tions do not exist a guess is made by picking the
canonical form, a form shared also by the lexical most common sense for one of the found lexical

508
A Semantic Web-Based Approach for Building Personalized News Services

Box 1.

selectedSense(lex) = arg min


s j ∈senses (lex )
∑ d (s , sc )
sci ∈I
j i (1)

representations. These senses are added to the occurrences in a corpus (e.g., pairwise mutual
context (I) of the sentence. For each remaining information, Google distance, etc.). Nevertheless,
polysemous lexical representations two steps many of these methods are less precise than the
take place: a disambiguation step to find the cor- graph-based method used here, as they do not
rect sense and an insertion step which adds the take in account senses, comparing only the lexical
newly computed sense to the context. In this way representations of concepts.
the context gets enlarged with new senses that
model the meaning of the sentence. Formula (1)
specifies what sense is selected for each lexical NEWS QUERYING
representation (See Box 1).
The distance between senses is defined as The user expresses the topics of interests by posing
being inverse proportional with the length of the queries using concepts from the domain ontology.
shortest path between the corresponding synsets In addition the user can express constraints that
(sensei and sensej) in the semantic lexicon graph. the timestamps associated to news items need to
The graph is based on the hypo/hypernyms re- satisfy. The news querying step consists of two
lationships stored in the semantic lexicon. Path substeps: query formulation, i.e., supporting the
lengths larger than a predefined constant (e.g., 4 user to build queries, and query execution, i.e.,
or 5) are not used (for these cases the distance is computing the results of query evaluation. Below
considered infinite). In this way we employ only we present in detail each of these substeps.
semantically close concepts that truly help in the
disambiguation procedure and also improve the Query Formulation
speed of the process by not using arbitrarily large
paths. Formula (2) shows how to compute the In order to assist the query construction process
distance between two senses (See Box 2). we present to the user the ontology graph. We
As the distances between synsets are not decided to show a graph-based representation of
changing, one can pre-compute these, thus further the ontology instead of a tree-based representation,
reducing the time needed for the disambigua- as it gives more insight in the overall structure of
tion step. The Hermes framework can be used our domain. For example a graph representation
with other methods for computing the similarity captures more relationship types instead of a
between concepts as for example string metrics singular relationship type, often the subsumption
(e.g., Levenshtein, Editex, etc.) or lexical co- relationship, from a tree-based representation. The

Box 2.

1
d ( sensei , sense j ) =
length( shortestPath( synset ( sensei ), synset ( sense j ))) (2)

509
A Semantic Web-Based Approach for Building Personalized News Services

user needs to understand the different relationship The original graph of the domain ontology is
types in order to be able to build his query. also called the conceptual graph. Based on the user
By using the ontology graph the user can selection, a new graph is generated, the so-called
select the direct concepts of interest. In addition search graph that contains only the concepts and
he is able to specify concepts of interest using a concept relationships relevant for the query. The
keyword search facility through the graph. For user can go back and forth between the two graphs
this purpose the input keyword is checked for performing new selections and thus updating the
possible inclusion in the lexical representations search graph with new concepts.
of concepts. If such inclusions are found the The search graph is given by the subgraph of
corresponding concepts are being returned as the conceptual graph that models the user interests
possible direct concepts of interest. It is the task which is equivalent to the answer of a conjunctive
of the user to accept these as direct concepts of query based on selected binary predicates and
interest or to reject them. concepts (that form graph patterns) while keeping
One of the important functionalities of the the selected relationships between the concepts
Hermes approach is that it allows for the selec- in the result set. The search graph has disjunctive
tion of concepts indirectly linked to the selected semantics with respect to the included concepts
ones, concepts which are not a priori known to the which means that the user is interested in any of
user. We call these concepts indirect concepts of the search graph concepts.
interest. For the selection of the indirect concepts Another crucial functionality of the Hermes
of interest the user can state the type of the rela- approach is that it allows the specification of
tionships that links the direct concepts of interest temporal constraints for news items. As news
to the indirect concepts of interests or leave this items appear at a certain moment in time and
undefined in which case all relationship types are have certain time validity, it is important to be
being considered. able to restrict the timestamps associated with
Suppose that the user has selected the direct the news. For this purpose the user can employ
concept of interest Google from the NASDAQ time comparison/arithmetic operators and retrieve
domain ontology. The user also specifies that he the current time in order to build complex time
is interested in news related to the competitors of expressions. In addition the system provides
Google by selecting the hasCompetitor rela- predefined temporal constraints such as: last day,
tionship. That means that Yahoo!, Microsoft, last week, last two weeks, last three months, last
and EBay will be selected as indirect concepts quarter, last half year, and last year. The tempo-
of interests, without the user having to know the ral conditions that model these constraints have
exact names of Google’s competitors. All this conjunctive semantics as they need to be fulfilled
background information is being extracted from in the same time.
the ontology.
The direct, indirect, keyword-based search Query Execution
concepts of interest, and the other concepts from
the ontology need to be emphasized in a graph Based on the previously selected concepts and
by using for example different colors. In this way specified temporal constraints the system can
the user is able to know, by analyzing the graph, support the generation of the corresponding
why a certain concept is being highlighted. As query in a semantic query language. This trans-
the size of the graphs can be very large the user lation process involves mapping concepts and
is provided with zooming/panning facilities for temporal constraints to query restrictions. After
visualizing the ontology. that, the user can trigger the query evaluation

510
A Semantic Web-Based Approach for Building Personalized News Services

and the relevant news items are retrieved. The news item text, thereby offering to the user an
order of the retrieved news items is not relevant, explanation of why a certain news item is con-
at this stage. sidered to be relevant.

RESULTS PRESENTATION HERMES NEWS PORTAL

The results returned from query evaluation are The Hermes News Portal (HNP) is an implementa-
presented in the order of their relevance for the tion of the Hermes framework, which allows the
user query. For this purpose, for each returned user to specify queries on the considered domain
news item a relevance degree is computed based using temporal constraints and subsequently re-
on all the hits between the news item and the trieve the relevant news items. The presentation
query concepts. News items with high relevance of HNP follows closely the steps proposed by
degree are placed at the top of the retrieved news the Hermes framework. Note that HNP is one
items list. of the possible implementations of the Hermes
Based on previous work (Micu, Mast, Milea, framework, with specific design choices, query/
Frasincar, & Kaymak, 2008) the relevance de- programming languages, and libraries used.
gree is defined as a weighted sum of the number Operating on the Semantic Web we chose as
of hits (n(ci)), where the weights (w) depend on ontology language OWL due to its expressivity
hits location (title or body of a news item). From and standard status. We did not opt for RDFS
our experimental results we have determined as because OWL specific features, as for example
acceptable values for wtitle to be 2 and for wbody to relationships inverses (hasCompetitor has
be 1. Formula (3) presents how to compute the as inverse isCompetitorOf), are exploited in
relevance degree (See Box 3). the conceptual graph. Lacking a true OWL query
News items that have the same relevance language we used SPARQL as the query language,
degree are sorted in descending order based on an RDF query language that we extended with
the associated timestamps (the most recent news time-related functionality. This functionality is
items are presented first). provided by implementing comparison/arithmetic
In addition to presenting the relevant news time operators and functions for retrieving current
items, the system shows the query concepts in time information.
order to provide cues of the current query for The chosen implementation language is Java
which the results are computed. Also, for each due to the availability of powerful libraries for
returned news item, the found lexical represen- manipulating, reasoning with, querying, and
tations stored in the hits are emphasized in the visualizing OWL ontologies. For manipulating

Box 3.

relevanceDegree (news item) = ∑


ci found in title
wtitle n(ci ) + ∑
ci found in body
wbody n(ci )
ci ∈O ∩ newsitem ci ∈O ∩ newsitem

(3)

511
A Semantic Web-Based Approach for Building Personalized News Services

Figure 2. News item example

and reasoning with OWL ontologies we used Jena timestamp, and the remaining text represents the
(Jena Development Team, 2008a). For querying we content of the news item.
employed ARQ (Jena Development Team, 2008b), The news classification step starts by iden-
the SPARQL implementation available in Jena. tifying lexical representations of the ontology
For visualizing ontologies we adapted the generic concepts in the news item. First the tokenization,
graph visualization library Prefuse (The Berkeley sentence splitting, part-of-speech tagging, and
Institute of Design, 2008) for visualizing OWL morphological analysis take place. Then, the
graphs (Borsje & Giles, 2008). For part-of-speech concepts from the ontology are traversed once and
tagging we used the Stanford parser (The Stanford their lexical representations are matched against
Natural Language Processing Group, 2008). As a the content of the news item. The following lexical
semantic lexicon we employed WordNet (Princ- representations “Google”, “extend”, and “com-
eton Cognitive Science Laboratory, 2008), the pany” are found. Next, per sentence, for each of
largest database available online for the English the lexical representations with multiple senses,
language. JWI (Finlayson, 2008) was used for the the word sense disambiguation procedure takes
morphological analysis (finding lemmas of words) place in order to identify the used senses. The
and the communication with WordNet. noun “Google”, having only one sense, doesn’t
We illustrate the HNP by means of the follow- undergo the word sense disambiguation procedure
ing user query: retrieve all news items related to and is mapped to the Google concept.
Google or one of its competitors that appeared in For “extend” and “company” a word sense
the last three months. For this query we will go disambiguation procedure is needed. In this pro-
through all the different phases of Hermes: news cess we do consider also lexical representations
classification, news querying, and results presenta- of concepts outside the ontology, as for example
tion. In the current HNP the news items duplicates the nouns “customer” and “business”, or the
removal has not been yet implemented. verb “buy” that do appear in the news item. We
select the sense that yields the smallest sum of
similarities to previously disambiguated lexical
NEWS CLASSIFICATION representations. For “extend” it is determined as
representing the extend-verb-#1 concept with
The news classification step is responsible for lemma extend, part-of-speech tag verb, and the
indexing news items based on the domain ontol- first sense from WordNet. For “company” the
ogy. We present this process by means of the corresponding concept is company-noun-#1.
news item example depicted in Figure 2. The first “customer”, a lexical representation outside the
line describes the title, the second line shows the ontology, is determined as having the sense
customer-noun-#1.

512
A Semantic Web-Based Approach for Building Personalized News Services

Figure 3. Conceptual graph example

After identifying an ontology concept in a The concepts of interest can only be the current
news item, a hit is stored. This hit is defined as concept, all related concepts including the current
a link between the news item and the concept one, all related concepts excluding the current
together with the found lexical representation. one, or all related concepts by means of speci-
For this purpose we decided to model a hit as fied relationships including/excluding the current
an instance of the Relation class, which uses concept. The node info panel shows information
different properties for storing the involved news regarding the selected concept.
item, concept, and found lexical representation. The local name of the different concepts is
This modeling choice is based on a best practice displayed using ovals. In order to emphasize the
for modeling N-ary relations on the Semantic different types of concepts we use the following
Web (Noy & Rector, 2006). coloring scheme for ovals. The selected concepts
are displayed in red, the concepts related to the
selected one are shown in green, and the ones
NEWS QUERYING returned by the keyword search are presented in
pink. The other concepts are displayed in yellow
Query Formulation for classes and magenta for individuals.
In the example from Figure 3 the user has
Figure 3 shows the conceptual graph from which directly selected the Google concept. In the
the user can select concepts of interest. Once a user node info panel all the information related to the
selects a concept, the control panel gets activated Google concept is displayed: name, competitors,
by means of which the user can add to the search CEO, etc. Then, the user can select the indirect
graph his concepts of interest. concepts by specifying the hasCompetitor

513
A Semantic Web-Based Approach for Building Personalized News Services

Figure 4. Search graph example

relationship between the direct concept and the differently than in the previous situation, only the
indirect ones. The user also specifies that the information given by the specified relationships,
Google concept should be kept in the search in this case the hasCompetitor relationship,
graph. is displayed.
After selecting the concepts of interest the In this example the user has specified that the
user can visualize the search graph. The user can news items have to be between 1st of September
refine its query by deleting some of the concepts, 2006 and 1st of December 2006, where the current
resetting the search, or adding new concepts from day is 1st of December 2006 (the last three months
the conceptual graph to the search graph. After in the user query). Alternatively the user could
a number of iterations the user has added all the have selected the past three months option from
concepts of interests to the search graph. the predefined temporal constraints.
Figure 4 shows the search graph, which pres-
ents the selected concepts, in this case Google Query Execution
and its competitors: Microsoft, Yahoo!, and
EBay. In the time constraint panel the user can For each search graph a SPARQL query is gener-
specify the desired time restriction. The user can ated. This query is a SELECT query as it retrieves
choose between predefined temporal constraints the news items in which any of the search graph
as the past hour, past day, past week, past two concepts are present. The disjunctive semantics
weeks, past three months, past quarter, past of the search graph with respect to its embedded
half year, and past year, and specific time/date concepts is naturally specified as an ‘or’ filter in
constraints. the SPARQL query. Also, we have decided to use
As before the node info panel displays informa- filters to specify the time restrictions that news
tion about the currently selected node. However, timestamps need to satisfy. Due to the conjunctive

514
A Semantic Web-Based Approach for Building Personalized News Services

Figure 5. SPARQL query example


PREFIX hermes: <http://hermes-news.org/news.owl#>
SELECT ?title
WHERE {
?news hermes:title ?title .
?news hermes:time ?date .
?news hermes:relation ?relation .
?relation hermes:relatedTo ?concept .
FILTER (
?concept = hermes:Google ||
?concept = hermes:Microsoft ||
?concept = hermes:Ebay ||
?concept = hermes:Yahoo
) .
FILTER (
?date > “2006-09-01T00:00:00.000+01:00” &&
?date < “2006-12-01T00:00:00.000+01:00”
)
}

Figure 6. Custom time functions


xsd:date currentDate()
xsd:time currentTime()
xsd:dateTime dateTime-add(xsd:dateTime A, xsd:duration B)
xsd:dateTime dateTime-substract(xsd:dateTime A, xsd:duration B)
xsd:duration dateTime-substract(xsd:dateTime A, xsd:dateTime B)

Figure 7. tSPARQL query example


PREFIX hermes: <http://hermes-news.org/news.owl#>
SELECT ?title
WHERE {
?news hermes:title ?title .
?news hermes:time ?date .
?news hermes:relation ?relation .
?relation hermes:relatedTo ?concept .
FILTER (
?concept = hermes:Google ||
?concept = hermes:Microsoft ||
?concept = hermes:Ebay ||
?concept = hermes:Yahoo
) .
FILTER (
?date > hermes:dateTime-substract(hermes:now(), P0Y3M) &&
?date < hermes:now()
)
}

semantics of the time restrictions we modeled them The first part of the SPARQL query defines that
as an ‘and’ filter in the SPARQL query. the returned news items should be related to the
Figure 5 shows the SPARQL query correspond- concepts of interest. The second part of the query
ing to the search graph given in Figure 4. This is composed of two filters. The first SPARQL filter
query is hard-coded with XML Schema date- defines the concepts of interest to be Google,
Times specifying the desired temporal boundar- Microsoft, Ebay, and Yahoo!. The second
ies of the interval in which the timestamps of the SPARQL filter specifies that the timestamp of news
desired news items need to be contained. items should be between 1st of September 2006

515
A Semantic Web-Based Approach for Building Personalized News Services

and 1st of December 2006, where 1st of December of the query is closer to its representation, in our
2006 is the current day. Both dates are specified current example that is retrieving the news items
using XML Schema dateTime format. that appeared in the last three months.
In order to ease the specification of temporal The tSPARQL query uses the dateTime-
constraints in queries we have extended SPARQL substract() to determine the dateTime of
with custom functions. We call the SPARQL lan- three months ago, now() is used to obtain the
guage extended with time functions tSPARQL. current dateTime, and it specifies that the news
Please note that SPARQL does naturally support timestamps should be between these two date-
such extensions, tSPARQL being backwards Times. P0Y3M is an XML Schema duration
compatible with SPARQL. Figure 6 shows the constant that specifies a period with 0 number of
signature of the time functions that we have added. years and 3 months.
These functions relate to retrieving the current After its creation, the tSPARQL query is
date and time, the current dateTime instance, executed. As a result, the news items that match
adding/substrating to a dateTime instance a the query constraints (concepts of interest and
duration, and subtracting two dateTime temporal constraints) are being returned. The
instances (date, time, dateTime, and dura- order of the results is not relevant here.
tion are defined by XML Schema).
Figure 7 depicts the same query as in Figure
5 but now written in tSPARQL. Different than RESULTS PRESENTATION
in the previous case the tSPARQL query is not
hard-coded with times and dates, but makes use Figure 8 shows the results after the query execu-
of custom functions and durations. The semantics tion. It lists the concepts of interest from the search

Figure 8. Results presentation example

516
A Semantic Web-Based Approach for Building Personalized News Services

graph (top) and the retrieved news items (bottom). at providing meaning for each (compound) word
In addition, the system shows the found lexical in news messages but only for the ones that cor-
representations of the concepts of interests in the respond to lexical representations of ontology
returned news items using different colors. The concepts. The meanings of the news’ (compound)
user is able to deselect some of the concepts of words that are not present in the ontology are used
interest in order to refine his query and thus limit only to help the disambiguation process of found
the result set. At the current moment the relevance concept lexical representations.
degree and timestamps-based sorting of relevant A different metric for the performance evalu-
news items as proposed in the Hermes framework ation is the latency of a news item in the concept
have not been yet implemented in the HNP. identification phase. The obtained average latency
time is around 30s which represents the time
needed to process a news item from tokenization
EVALUATION to concept recognition. The bottleneck lies in the
disambiguation step for which distances between
In order to evaluate the performance of the imple- synsets need to be computed. As identified in the
mentation we measured the concept identification Hermes framework these distances can be pre-
precision. Precision was defined as the number computed (given a certain limit for the shortest
of concepts correctly identified in the news items path length between synsets) which would further
divided by the number of concepts identified in reduce the disambiguation time.
news items. If we define recall as the number of Regarding usability we have asked 9 users
concepts correctly identified in the news items (students at Erasmus University Rotterdam fol-
divided by the number of concepts that should lowing a course on Semantic Web technologies
have been identified in the news item, one can including RDF(S), OWL, and SPARQL) to find
notice that for our application precision and re- news items for 3 given natural language queries
call are the same. The reason is that we are only in two ways: (1) using the Hermes implementation
looking for given concept lexical representation and (2) a SPARQL engine. Most students were able
in news items which means that we identify all to correctly build the search graph and specify the
the concepts that should be recognized possibly temporal constraints, as well as the corresponding
with some errors. For our current implementation SPARQL query. All queries were faster specified
the precision (recall) is 85% for a given repository using the Hermes framework than using SPARQL.
of around 200 news items. Note that we do not claim that it is easier to use
Precision (recall) are based on cumulative er- Hermes instead of SPARQL for querying RDF
rors through our application pipeline based on the graphs, but for expressing a certain set of RDF
HNP’s part-of-speech recognition, morphological queries (the ones supported by the search graph
analysis, and word sense disambiguation algo- with temporal restrictions, which we consider
rithm. Despite using only WordNet as a semantic typical for news querying) Hermes seems to be
lexicon (compared to other approaches which easier to use than SPARQL. Among the features
combine several semantic lexicons, some domain mostly appreciated by students in Hermes were
specific (Navigli & Velardi, 2005)) we obtained the graphical representations of the conceptual
high values for precision as many of our concepts’ graph, the predefined time functions, and the
lexical representations are named entities (names visual cues employed for emphasizing concepts
of companies, CEO’s, locations, etc.) that usually in returned news items.
have only one meaning. The high value of recall Compared with traditional keyword-based
can be explained by the fact that we do not aim search engines for news items (e.g., YourNews,

517
A Semantic Web-Based Approach for Building Personalized News Services

SeAN, Google News, Yahoo! News, etc.) our as a query language we used SPARQL. As a
semantic approach benefits from better precision, semantic lexicon we employed WordNet, one of
as it is able to disambiguate (compound) words, the most popular English dictionaries available
more query expressive power, because it allows the online. For representing temporal constraints
selection of indirect concepts (i.e., concepts not we have extended the SPARQL language with
directly related to the items of interest), and the temporal functions.
support for temporal constraints. A quantitative Differently than Google News and Yahoo!
comparison with non-semantic based approaches News, Hermes is able to exploit the background
is difficult to achieve due to the query limitations information stored in ontologies for retrieving
that these systems have and the impossibility of user’s items of interest. In this way the user doesn’t
using the same news items as inputs in the com- need to explicitly define all the instances involved
pared systems. in the query by making use of the concept rela-
In the tradeoff between expressivity and us- tionships for specifying his concepts of interest.
ability we decided to keep our queries simple with In addition to the concepts of interest, the user is
intuitive semantics so that a broad range of users able to specify temporal constraints in his query.
(casual users, media analysts, stock brokers, etc.) Another key feature of Hermes is the word sense
should be able to use Hermes. Nevertheless, we disambiguation procedure, which is not used in re-
acknowledge that an expert user might need more lated approaches as SeAN, YourNews, MyPlanet,
query expressivity (e.g., optional graph patterns, or SemNews. The word sense disambiguation step
disjunctive semantics for temporal constraints, increases the accuracy of news classifications, by
etc.) which contributes to the increase in complex- making sure that the found lexical representations
ity (and thus to the decrease in usability) of the indeed correspond to the meaning of the domain
framework. For this purpose we plan to extend ontology concepts.
the Hermes framework in the future with addi- As future work, we would like to extend
tional powerful functionality that would enable the Hermes framework by employing multiple
the generation of a news personalization service semantic lexicons and adding specific concepts
family (services targeting novice, average, or to the ontology that are not captured in existing
expert users). semantic lexicons. Some domain specific concepts
(e.g., domain neologisms) are used in news items
while current semantic lexicons, which are not
CONCLUSION up-to-date, do not include them. We also plan to
exploit the structure of the domain ontology in
The Hermes framework proposes a sequence of order to compute the similarity between concepts.
steps to be followed for building personalized news In this way we enrich our knowledge base and
services. The input for these systems comprises thus are better equipped in determining concept
RSS news feeds and the output are news items (synset) similarity.
fulfilling user needs. The Hermes approach is In addition, we would like to introduce a learn-
based on a domain ontology used for classify- ing step in which new ontology instances and
ing news items and to support the user define relations are extracted from news items. For this
his concepts of interest. In addition the user can purpose we envisage the use of lexico-semantic
specify temporal constraints that the news item rules that would extract the relevant information
needs to obey. The Hermes News Portal (HNP) from news items. For example a proper noun
is an implementation of the Hermes framework. followed by “Inc.” in a news item as in “Clear-
The domain ontology is specified in OWL and Forest Inc.” would indicate that the proper noun

518
A Semantic Web-Based Approach for Building Personalized News Services

is a company which needs to be inserted in the in a news-centric approach where the concept
ontology if it is not already present there. In ad- lexical representations are identified during a
dition we would like to explore the possibilities single news item traversal. Having a constant
to redefine the domain ontology as a time-based access time to concept lexical representations and
representation, where instances have a certain time taking in consideration that the number of lexical
validity associated with them (Milea, Frasincar, representations in a news item is smaller than in
& Kaymak, 2008). These temporal extensions an ontology might reduce the time needed for the
to the ontology would enable us to better reason news classification step.
with the temporal contextual information avail- Additionally, we would like to conduct a more
able for our domain. extensive evaluation procedure of the Hermes
Another direction that we would like to pur- implementation. Based on detailed question-
sue is that of semantic adaptation of news items naires and measuring the time spent on building
based on a user model. The user preferences now queries given in natural language, we can obtain
represented in the (temporary) search graph would more empirical evidence on the system usability.
be represented in a (stored) user model which is The accuracy of the proposed relevance degree
continuously adapted based on user behavior. In (based on concept identification) could be deter-
order not to bother the user with already seen mined by measuring the access order and reading
content, we would like to be able to filter news time of news item in the result list (accessing the
items that provide new information by using a first items first and spending substantial time
novelty control mechanism (Gabrilovich, Dumais, for reading them are good indications that the
& Horvitz, 2004). We believe that our semantic returned items are relevant). In addition, we want
approach can be successfully applied for model- to experiment with other domains (e.g., politics,
ing dissimilarities between news items and thus sports, etc.) and analyze the precision and latency
be able to recommend only news carrying novel of our implementation for these new fields. The
content. In a different scenario, by measuring genericity of our approach only asks for the
the similarities between news items, we would definition of a new domain ontology and domain-
be able to recommend news items related to the specific news feeds that need to be plugged into
same story but issued at different moments in our implementation.
time and/or by different institutions.
Regarding HNP we would like to implement
news items duplicates removal, and the relevance ACKNOWLEDGMENT
degree and timestamps-based sorting of relevant
news items, as proposed in the Hermes frame- The authors are supported by the EU funded IST-
work. Additionally, we also wish to implement STREP Project FP6-26896: Time-Determined
the previously proposed extensions to the Hermes Ontology-Based Information System for Real
framework: enriching our knowledge sources with Time Stock Market Analysis (TOWL). More
multiple lexicons and domain-specific concepts, information is available on the official website of
updating the domain ontology based on news the TOWL project (TOWL Consortium, 2008).
information, adding a user model and employ- Also, we would like to thank Wouter Rijvordt,
ing it to adapt system functionality, and filtering Maarten Mulders, and Hanno Embregts for their
news that provide novel content. Also, we would contribution to the Hermes framework.
like to test the usage of data structures for fast
data access (e.g., hash maps) for ontology access

519
A Semantic Web-Based Approach for Building Personalized News Services

REFERENCES Science (HICSS 2006) (Vol. 3, pp. 62.62). Wash-


ington, DC: IEEE Computer Society.

Ahn, J.-w., Brusilovsky, P., Grady, J., He, D., & Jena Development Team. (2008a). A Semantic
Syn, S. Y. (2007). Open User Profiles for Adaptive Web Framework for Java (Jena). from http://jena.
News Systems: Help or Harm? In 16th Interna- sourceforge.net/
tional Conference on World Wide Web (WWW
Jena Development Team. (2008b). A SPARQL
2007) (pp. 11-20). New York, NY: ACM.
Processor for Jena (ARQ). from http://jena.source-
Ardissono, L., Console, L., & Torre, I. (2001). An forge.net/ARQ/
Adaptive System for the Personalized Access to
Kalfoglou, Y., Domingue, J., Motta, E., Vargas-
News. AI Communications, 14(3).
Vera, M., & Shum, S. B. (2001). myPlanet: An
Bechhofer, S., Harmelen, F. v., Hendler, J., Hor- Ontology-Driven Web-Based Personalized News
rocks, I., McGuinness, D. L., Patel-Schneider, P. Service. Paper presented at the Workshop on On-
F., et al. (2004). OWL Web Ontology Language tologies and Information Sharing (IJCAI 2001).
Reference W3C Recommendation 10 February
Kandel, E., & Marx, L. M. (1997). NASDAQ
2004.
Market Structure and Spread Patterns. Journal
Berners-Lee, T., Hendler, J., & Lassila, O. (2001). of Financial Economics, 45(1), 61-89.
The Semantic Web. Scientific American, 284(5),
Klyne, G., & Carroll, J. J. (2004). Resource
34-43.
Description Framework (RDF): Concepts and
Borsje, J., & Giles, J. (2008). OWL2Prefuse. from Abstract Syntax: W3C Recommendation 10
http://owl2prefuse.sourceforge.net/index.php February 2004.

Brickley, D., & Guha, R. V. (2004). RDF Vocabu- Micu, A., Mast, L., Milea, V., Frasincar, F., &
lary Description Language 1.0: RDF Schema: Kaymak, U. (2008). Financial News Analysis
W3C Recommendation 10 February 2004. Using a Semantic Web Approach. In A. Zilli,
E. Damiani, P. Ceravolo, A. Corallo & G. Elia
Finlayson, M. (2008). The MIT Java Wordnet
(Eds.), Semantic Knowledge Management: An
Interface (JWI). from http://www.mit.edu/~markaf/
Ontology-Based Framework (pp. 311-328). Her-
projects/wordnet/
shey, Pennsylvania: IGI Global.
Gabrilovich, E., Dumais, S., & Horvitz, E. (2004).
Milea, V., Frasincar, F., & Kaymak, U. (2008).
Newsjunkie: Providing Personalized Newsfeeds
Knowledge Engineering in a Temporal Seman-
via Analysis of Information Novelty. In 13th
tic Web Context. In The Eighth International
International Conference on World Wide Web
Conference on Web Engineering (ICWE 2008)
(WWW2004) (pp. 482-490). New York, NY:
(pp. 65-74). Washington, DC: IEEE Computer
ACM.
Society Press.
Guarino, N., & Welty, C. A. (2002). Evaluating
Motta, E. (1999). Reusable Components for
Ontological Decisions with OntoClean. Com-
Knowledge Modelling: Case Studies in Parametric
munications of the ACM 45(1), 61-65.
Design Problem Solving (Vol. 53). Amsterdam,
Java, A., Finin, T., & Nirenburg, S. (2006). Text the Netherlands: IOS Press.
Understanding Agents and the Semantic Web. In
Navigli, R., & Velardi, P. (2005). Structural
39th Hawaii International Conference on Systems
Semantic Interconnections: a Knowledge-Based

520
A Semantic Web-Based Approach for Building Personalized News Services

Approach to Word Sense Disambiguation. IEEE Salton, G., & McGill, M. J. (1983). Introduction
Transactions on Pattern Analysis and Machine to Modern Retrieval. New York, NY: McGraw-
Intelligence, 27(7), 1063-1074. Hill.
Nirenburg, S., & Raskin, V. (2001). Ontological Seaborne, A. (2004). RDQL - A Query Language
Semantics, Formal Ontology, and Ambiguity. In for RDF: W3C Member Submission 9 January
Formal Ontology in Information Systems (FOIS 2004.
2001) (pp. 151-161). New York, NY: ACM.
The Berkeley Institute of Design. (2008). The
Noy, N., & Rector, A. (2006). Defining N-ary Prefuse Visualization Toolkit. from http://prefuse.
Relations on the Semantic Web: W3C Working org/
Group Note 12 April 2006.
The Stanford Natural Language Processing
Princeton Cognitive Science Laboratory. (2008). Group. (2008). The Stanford Parser: A Statistical
A Lexical Database for the English Language Parser. from http://nlp.stanford.edu/software/lex-parser.
(WordNet). from http://wordnet.princeton.edu/ shtml

Prud’hommeaux, E., & Seaborne, A. (2008). TOWL Consortium. (2008). Time-Determined


SPARQL Query Language for RDF: W3C Rec- Ontology-Based Information System for Real
ommendation 15 January 2008. Time Stock Market Analysis (TOWL). from http://
www.towl.org/
Salton, G. (1971). The SMART Retrieval System—
Experiments in Automatic Document Processing. Winer, D. (2003). RSS 2.0 Specification: Harvard
Upper Saddle River, NJ: Prentice-Hall. Law School.

This work was previously published in IGI Publishing, edited by I. Lee, pp. 35-53, copyright 2009 by IGI Publishing (an
imprint of IGI Global).

521
522

Chapter 2.15
A Service Oriented Ontological
Framework for the Semantic
Validation of Web Accessibility
Rui Lopes
LaSIGE, University of Lisbon, Portugal

Konstantinos Votis
Pattern Recognition Laboratory, University of Patras, Greece Informatics and Telematics Institute,
CERTH, Greece

Luís Carriço
LaSIGE, University of Lisbon, Portugal

Spiridon Likothanassis
Pattern Recognition Laboratory, University of Patras, Greece

Dimitrios Tzovaras
Informatics and Telematics Institute, CERTH, Greece

Abstract according to guidelines such as WCAG). In this


Chapter the authors present SWAF, the Semantic
The Web serves as the principal mediator for infor- Web Accessibility Framework, a base framework for
mation sharing and communication on a worldwide supporting the integration of accessibility services
scale. Its highly decentralized nature affords a into Web design and development processes. SWAF
scale free growth, where each endpoint (i.e., Web affords both tailoring accessibility to user needs and
site) is created and maintained independently. Web specifying the semantic validation of accessibility
designers and developers have the onus of making guidelines in different application situations.
sure that users can interact without accessibility
problems. However, coping with users with dis-
abilities poses challenges on how to ensure that a Introduction
Web site is accessible for any kind of user. When
(and if) this is done, designers and developers do it The increasing adoption of technologies from users
in a post-hoc way, (i.e., verify and tweak Web sites puts the Internet in a central spotlight. The Web, as
its major application, is accessed and interacted by
DOI: 10.4018/978-1-60566-650-1.ch003 users at constantly increasing pace, allowing them to

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

quickly seek information, interact with their peers totally dismiss the adequacy of Web sites to the
through social networks, or perform transactions different requirements of accessibility-dependent
from the comfort of their homes. For this reason, audiences, despite the fact that legislation is being
the way information is structured and presented pushed in several countries, in order to promote
is critical for the success of accessing it. the rights of people with disabilities.
However, users have their own specific charac- The dismissal of accessibility from information
teristics (e.g., abilities, impairments, preferences, providers leverages the fact that such guidelines
knowledge). Consequently, the accessibility of and standards for accessibility have inherent
each piece of information – such as a Web page problems. Since they are specified in such a way
– can differ significantly from user to user. While that they require manual inspection of their con-
often dismissed in the Internet realm and, more formance, developers have an increased effort on
specifically, on the Web, people with disabilities coping with accessibility issues. Furthermore, by
are not just a small population minority. If one being informally described (i.e., in natural lan-
takes into account people with mild disabilities, guage), they tend to lead to different interpretations
the slice of the population that requires some sort from developers and accessibility experts and,
of software-based accessibility ramp is of the ut- for this reason, different and incoherent ways to
termost importance. ensure that a given Webpage is accessible.
The most important way to mitigate this prob- When Webpages and Websites require an
lem is making sure that information providers additional effort of supporting more specific
(ranging from the individual to large corporations) and fine-grained audiences, the development of
do not overlook such accessibility issues. Inter- accessibility-centric solutions becomes cumber-
nationally recognized organizations such as the some. Guidelines have an implicit assumption of
World Wide Web Consortium (W3C, n.d.) play a which audiences they target to (and more often than
critical role on helping information providers to not, in a very informal and loose way). Typically,
cope with accessibility. Traditionally, this goes in these are often geared towards people with visually
the form of specifying accessibility-centric best impairments (e.g., different kinds of blindness).
practices, guidelines, and vocabularies to augment The way users from this audience interact with a
already existing Web languages. webpage is also left implicit, but often assumed
Evangelization of accessibility practices, coped to include specific devices (e.g., screen readers).
with the progressive intertwining of accessibility This leads to a lack of understanding what is
features in Web languages has brought Web ac- the role of guideline checkpoints for each user
cessibility more close to information providers. and device characteristic they are tackling, thus
Consequently, each day, Web accessibility is posing more difficulties on developers on how
gaining awareness. Guidelines such as WCAG, to tackle fine-grained accessibility analysis and
the Web Content Accessibility Guidelines (Chish- consequent development of accessible Websites
olm, Vanderheiden, & Jacobs, 1999), are being and Webpages. Consequently, software developers
followed more often, leading more users with need to have a conceptual framework in which
disabilities to have access to information without to situate disabled-related guidelines, which they
barriers. In order to be so, these guidelines are often do not have due to lack of experience with
presented as straightforward as possible, geared disabled people and their technologies.
towards the largest set of Web designers and This Chapter proposes SWAF (Semantic
developers. However, mostly due to financial, Web Accessibility Framework) as an ontological
human resources, and technological expertise framework targeted to accessibility-aware Web
problems, several companies (and individuals) design and development processes. This will

523
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

enable large organizations, small enterprises, or Web user agents (e.g., Web browser), authoring
even individuals (developers, designers, etc.) to tools, and content creators. However, it is not
produce Web sites of superior accessibility and clear whether providing guidelines is an effective
usability, accompanied with appropriate measures, method for ensuring usable designs, since these
technologies and tools that improve their overall might be differently interpreted by developers and
quality. This ontological framework can be used designers. Developers need to have a conceptual
to answer questions about common accessibility framework in which to situate disabled-related
standards, user abilities and disabilities, as well guidelines, which they often do not have due to
as about the technical capabilities and constraints lack of experience with people with disabilities
of appropriate assistive devices, thus forming the and assistive technologies.
context for semantic validation of Web acces- Consequently, even if people with disabilities
sibility. In This Chapter we present the general want to be independent and do things for them-
overview of the framework, and detail it in the selves by themselves, unfortunately, most Web
context of existing Web accessibility standards, sites and Web applications are not fully accessible
in order to facilitate accessibility assessment of today. There are a number of reasons for this as
Web sites across different audiences. explained below:

• The Web has evolved over the latest years,


Background and the importance of accessibility has only
begun to be appreciated and encouraged in
Designing for people with disabilities is becoming recent years. Older solutions are unlikely
an increasingly important topic for a variety of to be fully accessible (or accessible at all).
reasons, but especially due to recent legislation Making an existing Web site accessible is
in many countries that aims at promoting and often very difficult and expensive, in much
enforcing the rights of people with disabilities. A the same way as making an existing build-
number of philosophies and methodologies have ing wheelchair-friendly can be very dif-
been developed to support this process. ficult, as well as un-aesthetic. Although
Firstly there has been the development of the efforts should be made to improve acces-
universal design, design for all, and universal us- sibility, it will be typically easier to do so
ability philosophies, as detailed by Shneiderman during a major refurbishment.
(2000). Many developers worry that they will be • Many developers and, more surprisingly,
expected to produce a system that will be usable designers, are not aware of the importance
by every user, regardless of their abilities, and or need for accessibility. Consequently,
that they might have to seriously compromise new developments are being built in bliss-
their overall design to achieve this aim. Clearly ful ignorance, as many of them do not
this would be in no-one’s interest. With the in- have the necessary knowledge or skills for
creasing ability to personalize interfaces to meet building accessible Web sites.
the requirements of different users, this is not • Some market stakeholders believe that cre-
necessary. ating accessible solutions will have prohib-
Secondly, there have been numerous sets of ited costs and, at the same time, make them
guidelines to help developers produce systems boring and less attractive to the majority of
that are accessible and usable by people with users (read: the non-impaired).
disabilities. These range from very general • Existing design and development tools
guidelines to the very specific guidelines for give little out-of-the-box assistance in

524
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

most cases or, at worst, make it impossible software houses worldwide are making efforts
to develop accessible solutions. to address its mandate, which puts it as a central
destination for Web accessibility verification
Accessibility Standardization practices.
Major standards bodies such as the US Human
Up to now, there are several initiatives concerning Factors and Ergonomics Society (HFES, n.d.) are
guidelines, tools and technologies for Web acces- engaged in furthering the accessibility drive. Their
sibility. The major steering body for accessibility efforts extend to features and functions of the
is the World Wide Web Consortium and its Web operating systems, drivers, application services,
Accessibility Initiative (WAI, n.d.). WAI has other software layers upon which the application
three main tracks: the Web Content Accessibility depends and applications that increase accessibil-
Guidelines (Chisholm et al., 1999), the Author- ity with a general aim of reducing the need for
ing Tool Accessibility Guidelines (Treviranus, add-on assistive technologies. The International
Richards, Jacobs, & McCarthieNevile, 1999), and Standards Organisation ISO/TS 16071:2003 (ISO
the User Agent Accessibility Guidelines (Jacobs, 16071, 2003), Ergonomics of Human–system In-
Gunderson, & Hansen, 2002). The activities of teraction also provides guidance on accessibility
W3C and WAI are the result of collaboration of for HCI interfaces. The guidelines were designed
groups and organizations from different countries, to complement general design for usability cov-
like the TRACE Research and Development ered by related standards on usability.
Centre (TRACE, n.d.), which is responsible for As thoroughly discussed in this Section,
compiling and publishing the original set of Web there exist several initiatives and standardiza-
accessibility guidelines that provided the backbone tion bodies concerning guidelines, standards and
of WAI guidelines. methodologies for accessibility assessment that
Apart from the guidelines, there are also leg- can be effectively applied in the context of Web
islative and standards initiatives for accessibility. technologies. It is also a fact that the existing
Strong governmental support in the Unites States standards and best practices concerning accessi-
has led to initiatives such as the Americans with bility are in most cases confusing and incomplete
Disabilities Act (ADA, 1990). The UK equivalent (Lopes & Carriço, 2008). Therefore, developers
is the Disability Discrimination Act (DDA, 1995) need to have a conceptual framework in which
amended in 1999, and now extended to the Special to situate Web accessibility-related guidelines,
Educational Needs and Disability Act (SENDA, which they often do not have, due to lack of
2001). The rulings of ADA are also extended to experience on technologies for the disabled. The
Section 508 of the US Rehabilitation Act (Section fundamental aspect of pushing forward accessibil-
508, 1998). This legislation defines processes and ity on Web site design and development practices
the monitoring role of the US federal government is to provide concrete and objective rules and
in the procurement of electronic and information standardized guidelines that homogenize acces-
technology. Regarding accessibility, it states that sibility assessment and quality control procedures.
regardless of medium, government must ensure Consequently, existing software that aims at
that disabled federal employees and members of assessing accessibility based on such guidelines
the public have the same accessibility as non- (a thorough list of such software packages can
disabled members. Where accessibility is not pres- be found at http://www.w3.org/WAI/ER/tools/
ent, government is directed to provide alternative complete) will provide incomplete and overly
means. Although, Section 508 is intended for the generalized answers to whether a given Web page
US federal government, many organizations and or Web site is accessible.

525
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

While it is clear that determining what truly the end users, including the person’s disabilities
represents accessibility in the customer’s view and individual preferences.
can be elusive, it is equally clear that the number A central reference for classifying disabilities
and frequency of problems and defects associ- concerns the World Health Organization’s Inter-
ated with a Web site are inversely proportional to national Classification of Functioning, Disability
its accessibility. Software problems and defects and Health (ICF, n.d.), particularly tailored to
regarding accessibility are among the few direct impairment qualification on medical diagnosis
measurements of software processes and products. tasks. Consequently, it stresses just on profound
Such measurements allow us to quantitatively disabilities, leaving out several impairments such
describe trends in defect or problem discovery, as color blindness. Obrenovic, Abascal, & Star-
repairs, process and product imperfections, and cevic (2007) have leveraged ICF concepts into
responsiveness to customers. Problem and defect an accessibility description framework to help
measurements also are the basis for quantifying designers and developers discuss and describe
several significant software accessibility attri- multimodal interaction issues.
butes, factors, user characteristics and criteria. Gruber (1993) proposes an ontology archi-
Although the advantages of measurement in tecture that tries to cover comprehensively the
the Web site design and development process situation of persons with special needs for the
are indisputable, the popularity of measurement purpose to utilize this information for customi-
methods, within accessibility terms, in practice zation of their home environment’s services is
is rather limited (McGarry, 2002; Varkoi, 1999; proposed. This approach tries to combine con-
SEI, 2006). Very often difficulties arise when textual information like personal aspects (e.g.,
trying to focus the measurement. In many cases disabilities, preferences), technical aspects (e.g.,
it is unclear what should be measured and also equipment, services, network) and natural aspects
how the measurement data obtained should be (e.g., location, time) in a way that the smart home
interpreted (Habra, Abran, & Lopez, 2004; Kulik, environment’s services can adapt to the end user
2000). Choosing the correct measurement enti- more or less automatically while keeping the user
ties and ranking the importance of measurement in control.
accessibility indicators is still a challenging task Several pre-existed ontologies for support-
(Neely, 1998). Despite the difficulties, metrics ing context-aware smart environments, like
such as those defined by Vigo et al. (2007) CoOL (Strang, Linnhoff-Popien, & Fank, 2003),
provide insightful cues on how to approach the COBRA-ONT (Chen, Finin, & Joshi, 2003),
problem of measuring the accessibility of Web CONON (Wang, Zhang, Gu, & Pung, 2004),
sites, based on WCAG standards. This will help SOUPA (Chen, Perich, Finin, & Joshi, 2004),
designers and developers to have a better (and and UbisWorld (Heckmann, 2005). All these on-
measurable) understanding of accessibility on tologies share common concepts and structures.
Web technologies. From these, SOUPA incorporates most concepts
of previously defined ontologies and seems to be
Ontologies for Disability the most elaborated one of the listed ontologies.
and Accessibility However, all of them still lack a specific support
for persons with special needs towards a compre-
There are several efforts towards the direction of hensive specification.
the definition of ontological concepts and taxono- A major contribution to the field of ontologies
mies for people with disabilities. These efforts try for disabilities was made from EU’s FP6 ASK-IT
to cover adequately the personal requirements of project (ASK-IT, n.d.). Within ASK-IT, ontology

526
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

modeling and mapping produced a collection of technology and subsequent competencies and
shared sub-ontologies, which reflect mobility demands on interface design and interaction styles.
impaired people user needs, and relationally map Abecker et al. (2001) propose “AccessOnto” as
available services to them. These needs were an accessibility requirements theoretic ontologi-
initially specified and afterwards, the ontology cal framework consisting of four components: a
authoring procedure was based on content models requirements elicitation subsystem, an inference
derived from these specifications. It also defines engine, a requirements explanation subsystem and
the interrelationships that may rationally hold an accessibility knowledge base. It has the inten-
between user groups of people with disabilities tion of extrapolating a requirements specification
and various user information needs of different based on rules extracted from the accessibility
content types, including multi-modal content. knowledge database based on end user traits data
The potential for applying ontologies in end elicited by the end user.
user diverse environments and their potential for The aforementioned ontological frameworks
promoting a unified methodology is exemplified emphasize the fact that there is little coupling
by the ontology devised by Uschold, King, Mo- between ontologies regarding accessibility and
ralee, & Zorgios (1998). This ontology includes disabilities, and Web accessibility assessment
lexical and relational terms based on the idea of practices (as they tend just to frame different ac-
the activity (anything that involves doing) linked cessibility scenarios). Our proposed ontological
to the doer or operative unit which may be a per- framework will be based on existing ontological
son, organizational-unit or machine said to have models, as well as in best practices for ontology
capability and on occasion possessing roles in engineering, affording the design of a multi-layer
respect of an activity such as activity-owner. knowledge base for accessibility and disability
Wooldridge, Jennings, & Kinny (2000) also requirements mapping into Web accessibility
adopted a role-oriented analysis as a natural step verification procedures that can provide sup-
in the Gaia methodology. Another example, the port for the requirements and needs of different
Framework for Distributed organizational Memo- accessibility-centric user groups.
ries (Abecker et al., 2001), describes the various
actors in domain ontologies according to their
goals, knowledge and competencies. Van Heijst, SEMANTIC WEB ACCESSIBILITY
Schreiber, & Wielinga (2000) also capture the role FRAMEWORK
of an ontology in the accessibility requirements
specification process, where they illustrate how For many people, in particular for groups at risk of
a methodology can extract semantics from an exclusion, the complexity and lack of accessibil-
ontology at different levels of depth to produce ity and usability of Web sites is a major barrier to
conceptual models. information access. We respond to this challenge
As stated by Masuwa-Morgana & Burrell by proposing a tailored accessibility assessment
(2004), an ontology for accessibility require- ontological framework, the Semantic Web Ac-
ments could be centered in a similar fashion, on cessibility Framework (SWAF), which affords
an activity such as a use case in which there are the specification of user characteristics and their
doers (people and access technologies). The only requirements, and associate them to specific ac-
difference is that in that ontology for accessibility cessibility assessment procedures.
requirements there would be a need to reliably The main goal of SWAF is to provide sup-
furnish (with clear identities and essence) descrip- port for the formal and unambiguous definitions
tions of doers-people and patterns of doer-access of accessibility domains, as well as the possible

527
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

semantic interactions between them. We have sions: Web Accessibility Descriptions, and Web
specified SWAF to be integrated into accessibil- Accessibility Mapping, as depicted in Figure 1.
ity verification environments (e.g., authoring Each dimension is further explained in the fol-
tools, Web accessibility evaluators, integrated lowing Sections.
development environment - IDE). This will
establish a common vocabulary for exchanging Web Accessibility Descriptions
and describing the complex information that is
related to accessibility assessment of Web sites. The first dimension provides constructs to describe
The framework aims to formalize conceptual different Web accessibility concepts (WAD). To
information about: explore the differences and synergies between Web
accessibility fields, and to support the inclusion of
• The characteristics of users with dis- external concepts from other domains, the WAD
abilities, devices, applications, and other dimension cuts the concept space into a Generic
aspects that should be taken into account ontology and a set of Domain Specific ontologies,
when describing an audience with disabili- as detailed next.
ties and developing tailored Web sites.
• Web accessibility standards and associated Generic Ontology
checkpoints.
• Semantic verification rules to help describ- The Generic ontology forms the core ontology
ing requirements and constraints of audi- and describes top-level entities and concepts
ences, and associating them to accessibil- that are critical for the semantic validation of
ity checkpoints. Web accessibility. Thus this ontology provides
more abstract and generic knowledge such as
In order to cope with these goals, the framework general characteristics and disabilities of users,
must comply with the following requirements: devices, Web accessibility standards, and other
main aspects that constitute the basis for applying
• To be as formal as possible, thus providing accessibility-based approaches into the accessibil-
all the necessary definitions in a concise, ity validation field.
unambiguous, and unified form; Domains are specified in classes and subclasses
• Provide information that can be eas- providing a hierarchical model representing all
ily processed by software applications the knowledge fields that are necessary for the
and integrated into accessibility validation accessibility validation. There are also a number
processes; of properties denoting the relationship between
• Easily implemented by software develop- classes. A part of the Generic Ontology is de-
ers and other users involved in the software picted in Figure 2. This partial snapshot of the
development process of Web accessibility ontology consists of the main classes Disabilities,
tools. WAI_WCAG and Devices. The Disabilities class
contains three subclasses: HearingImpairment,
One of the main issues in designing and de- SpeechImpairment, and VisualImpairment. There
veloping the proposed framework was to make it is also a property of the type hasIncludedDomains
maintainable and extensible, while assuring model denoting that the classes WAI_WCAG and Devices
consistency within the framework. Therefore, we include disabilities.
have separated SWAF into two distinct dimen-

528
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

Figure 1. Semantic Web accessibility framework

Domain Specific Ontologies ogy from the domain specific ones is to facilitate
the extension of SWAF to different application
To better illustrate how the Generic Ontology can domains (e.g., outside the scope of Web related
cope with real scenarios within the Web accessibil- accessibility guidelines and applications).
ity domain, we defined several domain specific Each DSO uses the basic entities of the Ge-
ontologies and integrated them into SWAF. These neric Ontology to describe the specific concepts
domain specific ontologies (DSO) cope with the and structures that are needed for the semantic
key aspects that are part of the integration of validation of Web accessibility. This ensures that
Web accessibility into Web design and develop- all terms and their relationships utilized by each
ment processes. This way, ontologies are able accessibility approaches separately are included
to represent a more detailed description of their in the generic ontology scheme. To cover every
corresponding domain, fruitful for extensibility spectrum of applicability of accessibility assess-
scenarios (e.g., using Web accessibility validation ment procedures, there should exist a correspond-
ontologies in Mobile Web tailoring scenarios). ing domain specific ontology. Next, some of these
The purpose of distinguishing a generic ontol- ontologies are described.

Figure 2. Excerpt of the generic ontology

529
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

General Characteristics and taxonomy to afford the classification of Charac-


Disabilities Ontology teristic instances. This taxonomy distinguishes
Ability, Disability, and Preferences, as well as
As discussed previously, validating accessibility more specific concepts (e.g., SensorialAbility or
is a process that must cope with user’s disabilities, LearningDisability). Accordingly, the hasChar-
as well as with each individual’s preferences. acteristic property has been refined to cope with
Thus it is of great importance to consider the us- the three main domains.
ers’ personal capabilities determined by her/his Lastly, to afford a semantic extensibility and
impairments. Consequently, different categories proper categorization of user characterization
of disabilities (based on the ICF categorization) concepts, we defined an extendsCharacteristic
are incorporated within this ontology, such as: property that maps between Characteristic in-
stances in a taxonomical way.
• Visual impairments. Disorders in the func- All of these concepts provided by the ontol-
tions of the eye ranging form reduced ca- ogy strive for a strong and expressive tool for
pability of sight, color-blindness to total Web designers and developers to describe and
disability to see (e.g., cataracts or retinal characterize their target users in a clean, thorough
detachment). way, along the line of the descriptive ontologies
• Hearing impairments. Disorders in per- devised by Obrenovic et al. (2007). Furthermore,
ceiving audio, ranging from problems in by affording an extensible way of organizing user
understanding normal conversations to characterization concepts independently from us-
complete deafness (e.g., high or low tone ers/audiences, Web designers and developers can
hearing loss). build their own taxonomies with respect to their
• Specific learning impairments. Disorders particular needs without being tied to a particular
manifested by significant difficulties in the way of thinking and organizing information typi-
acquisition and use of listening, speaking, cal of stricter solutions.
writing, reading, reasoning, or mathemati-
cal abilities. Web Accessibility Standards
and Guidelines Ontology
The degree of the users’ disabilities determines
the extent of Web accessibility concepts and guide- This domain ontology covers the main evalua-
lines that must be followed by Web sites (e.g., tion guidelines for Web accessibility assessment
enlarging font sizes) to suit to users’ computing devised in the Web Accessibility Initiative, such
environment and usage context. as WCAG. These guidelines are divided into
To afford the specification of such concepts, checkpoints and arranged based on their impact
the General Characteristics and Disabilities On- and priority. The combination of these factors is
tology provides a set of supportive constructs at given in levels (none, A, AA, or AAA), depend-
a meta-level (e.g., generalizations). The main ing on their evaluation outcomes. For instance,
concepts are User and Characteristic. A hasChar- to claim conformance on level A, all the priority
acteristic property maps characteristics to users, one checkpoints must be satisfied.
thus affording the description of users. We have The table presented in the Appendix of this
further detailed several meta-concepts under the chapter reproduces the most fundamental Prior-
Characteristic umbrella. Since this ontology is ity 1 checkpoints, which have been incorporated
tailored to accessibility scenarios (in the broad into this ontology. It is important to notice that
sense of ability to access), we introduced a small some of these checkpoints may be irrelevant in

530
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

different situations. Certain Web site instances accessibility is assessed for Web pages must also
might not have markup that can trigger acces- cope with these differences.
sibility problems. Furthermore, they might also This ontology provides a simple set of meta-
be irrelevant based in the particular constraints level concepts to describe devices ecology: a
and preferences of a user (from an accessibility Device hasFeature DeviceFeature. Like in the
point-of-view). description of users and accessibility scenarios,
In particular, the checkpoints 1.2 and 9.1 apply this ontology affords out-of-the-box instances for
(only) if image maps are used, the checkpoints 5.1 common cases of devices and device characteris-
and 5.2 apply if tables are used, the checkpoint tics without closing the door to extensibility and
12.1 applies if frames are used, the checkpoint 6.3 odd-case scenarios.
applies if applets or scripts are used, and finally The description of devices can be used at two
the checkpoints 1.3 and 1.4 apply if multimedia is different fronts within development processes:
used. Most of these checkpoints are just relevant it can help triggering semantic validation rules
for those with visual impairments. on user/device mismatches (e.g., user with total
On the meta-level, this ontology introduces the blindness and computer screen), and tying Web
Guideline and Checkpoint concepts, which can be accessibility guidelines and checkpoints to par-
mapped through an includesCheckpoint property. ticular devices and/or device features.
While Web designers and developers can leverage
verification processes with the already supplied Web Applications Ontology
instances for WCAG, the extensibility provide
by this meta-level affords the addition of new Web pages are not the only end for Web tech-
guidelines and checkpoints to their development nologies. Nowadays, an increasingly number of
processes in an effective way. Furthermore, this applications is being ported from “traditional”
will allow them to leverage out-of-the-box all desktop environments into Rich Internet Applica-
domain independent verification rules provided tions (RIA), by taking advantage of the richness
in SWAF’s Web Accessibility Mapping Dimen- of Web browsers, thus becoming easily available
sion. to any kind of users. However, since these tech-
nologies cannot cope with specific semantics of
Devices Ontology desktop applications widgets, accessibility issues
may arise. The proposed domain ontology supports
Owing to the rapid development of electronic the development of accessible Web applications
technologies, it tends to be common to access by affording the inspection of ARIA keywords
Web sites outside the traditional field of a desktop (Accessible RIA) within Web pages, according
PC and a computer screen (e.g., PDAs, mobile to application requirements.
phones, assisting devices, etc.). This has brought Figure 3 represents a partial snapshot of the
more specific assistive technologies to improve ontology describing the knowledge domain of
interactivity for users with disabilities, as well HearingImpairment. One of the main classes is the
as broad personal preferences. This includes the CompleteDeafness that consists of four subclasses
ability of coping with diverse input/output mo- Title, Parameters, Keywords and AssignedARIA.
dalities combination within interactive scenarios. Each of these subclasses is characterized by a
Since the diversity of these technologies varies set of properties that can be either a simple data
along different axes (e.g., display resolution, im- property or an object property that denotes the
ages coloring, multimedia process, etc.), the way relationship between two classes.

531
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

Figure 3. Example of domain description ontology

This ontology supports both HTML-specific Mapping Ontology


concepts (such as key HTML terms that have
influence on accessibility issues), besides Web The WAM dimension provides a mapping ontol-
application domain concepts. This way, Web pages ogy comprising of a set of lexical and notational
are perceived as a subset of Web applications from synonyms to express the semantics of the rela-
the point-of-view of key concepts, thus affording tionships between concepts within the General
reuse scenarios of the ontology. Ontology and Domain Specific Ontologies. This
For example, the Web Applications Ontology mapping is necessary, since each ontology domain
provides the GUICharacteristic and GUITechnol- represents the semantics of different knowledge
ogy abstract concepts (coupled with more fine- domains. It is important to notice that these
grained concepts such as HTML), to support the mapping concepts can be use to tie terms from
description of the Web Applications domain. the General Ontology to any Domain Specific
Ontology, as well as between different Domain
Web Accessibility Mapping Specific Ontologies. This way, the Semantic Web
Accessibility Framework can support different
Finally, SWAF is completed with the Web Acces- interdependent relationships between DSOs,
sibility Mapping (WAM) dimension. This dimen- thus affording richer Web accessibility validation
sion aims to cover the establishment of mapping scenarios.
relationships between the ontologies of the first Since the proposed integration needs require
dimension, and to validate the semantics of these that information be passed seamlessly among the
relationships. These relationships can be used, different layers, generic and domain ontology
e.g., for efficient navigation and searching inside mapping is absolutely necessary. This semantic
the ontologies, as well as to afford the creation information stems from the semantic metadata
of semantic rules-based accessibility verification. description of the content and has to be mapped
Two ontology layers are provided in this dimen- to the corresponding classes and properties of the
sion (as detailed next): Mapping Ontology, and relevant domain description ontology. Therefore,
Rules Ontology. each domain specific ontology will provide a set
of concepts to the mapping ontology to support
this type of mappings.

532
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

For instance, the Web applications ontology While some rules can be specified with General
provides properties to map Checkpoint instances Ontology concepts, its use is fairly limited, as they
(inherent from the Web Accessibility Standards are not tied to particular application domains. By
and Guidelines Ontology) to Application instances using terms originated from Domain Specific
(described with concepts from the Web Applica- Ontologies, and by combining them according
tions Ontology). This mapping property allows to the semantics of existing validation processes,
the specification of which checkpoints are valid Web design and development processes can be
to the particular set of technologies available for augmented with more interesting verification rules
the design and development of Web applications. that are triggered according to specific application/
Another set of mapping concepts provide support audience requirements. It is worth mentioning the
for bridging Checkpoint instances with Device fact that this ontology serves as an entry point for
instances and User instances, thus closing the semantic validation processes. We devised it as a
loop between characterization of devices and placeholder upon which the SWAF ontology can
audiences and tailored Web accessibility assess- (and, in fact, should) be enriched with application-
ment processes. specific and technology-specific semantic Web
accessibility validation rules.
Rules Ontology As a simple example, we present the description
of a set of rules for users that have been character-
The last piece in the Web Accessibility Mapping ized as having some sort of visual disability, and
dimension of the Semantic Web Accessibility how to cope with content presentation. This is one
Framework concerns the specification of semantic of the critical rule types that are to be supported
validation rules for Web accessibility. This ontol- within Web accessibility validation scenarios.
ogy will provide the required set of rules that go User rules are defined as the set UR = {U1, U2,
beyond the syntactic analysis of Web accessibility U3}, where each one of the rules represents a
processes, such as the description of checkpoints, single semantic validation according to a specific
users, etc. The role of this ontology is, therefore, user audience:
to bridge the semantic verification gap between
the Web Accessibility Description domain and U1: if user is color blind then content of black
the Mapping Ontologies. and white images and black text are
We have devised this ontology as a set of rules preferred.
based on SWRL (Horrocks, Patel-Schneider, U2: if user is partially sighted then content of
Boley, Tabet, Grosof, & Dean, 2004), a rules audio and appropriate image is preferred.
language that affords the specification Horn-like U3: if user is totally blind then pure audio con-
rules with OWL predicates. The Rules Ontology tent is preferred.
can be used to reason which concepts from other
ontologies (both at the instance and meta levels) The same approach can be used in other do-
and which combinations of them are satisfied by mains, e.g., for devices. When verifying if Web
accessibility validation procedures. By setting up sites can cope with device capabilities, one can
these rules within an inference engine, relevant check if content can be appropriately fit into the
accessibility rules will be reasoned out according constraints imposed by devices. Such rules do-
to the information residing within the ontologies main, e.g., DR = {D1, D2, D3} can be defined
of the Semantic Web Accessibility Framework’s as:
Web Accessibility Domain dimension.

533
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

D1: if device is a mobile phone then image applications (?APP) encompassing HTML tech-
depth must be black and white. nologies (wao:HTML) must provide alternative
D2: if device is a PDA image color contrast non colored information. It is worth mentioning
must be high. that the constructs provided by the Web Ac-
D3: if device is a PC then image depth can be cessibility Ontology (wao:hasGUITechnology,
of any size. wao:HTML, wao:hasGUICharacteristics,
wao:characteristicName, and
When defining such semantic rules, the con- wao:characteristicValue) form the core validation
structs available on the different ontologies can rule, since they bind more general concepts (color
be used to express the concrete situations that blindness, WCAG checkpoint) to concrete concepts
help Web designers and developers in the tailored inherent of the application domain. All of these
Web accessibility verification processes in an concepts would have to be substituted, if target-
effective, unambiguous way. As an example, we ing the specific WCAG rule to other application
detail how to express the Checkpoint 2.1 from technology (e.g., outside the scope of Web acces-
the Web Accessibility Content Guidelines 1.0 (as sibility). Likewise, this type of rule can be easily
shown in the Appendix), which state: “Ensure adapted to device constraints such as the device
that all information conveyed with color is also domain rules described above, as follows:
available without color, for example from con-
text or mark-up”. When targeting to audiences
composed by individuals with color blindness, wao:hasGUITechnology(?APP, ?x1)
this checkpoint can be expressed in SWRL as &
(compact syntax): wao:HTML(?x1) &
wao:hasGUICharacteristics(?x1,
?x2) &
wao:hasGUITechnology(?APP, ?x1) wao:characteristicName(?x2, “al-
& ternativeNonColoredInformation”)
wao:HTML(?x1) & &
wao:hasGUICharacteristics(?x1, wao:characteristicValue(?x2,
?x2) & “true”) &
wao:characteristicName(?x2, “al- do:hasFeature(?DEV,
ternativeNonColoredInformation”) Individual(“colorDepth1bit”))
& =>
wao:characteristicValue(?x2, wasgo:isDefiningValidApplica
“true”) & tion(Individual(“WAI_check-
gdco:hasDisability(?USER, point2.1”), ?APP)
Individual(“colorBlindness”))
=> Here we attach Checkpoint 2.1 just to those
wasgo:isDefiningValidApplica devices (?DEV) that are severely constrained by
tion(Individual(“WAI_check- color depth, e.g., just black and white displays
point2.1”), ?APP) (“colorDepth1bit”). All other atomic rules can
be retained, thus showing that Web accessibility
Here we are ensuring that conforming to verification semantics are similar between differ-
Checkpoint 2.1, only for users (?USER) char- ent characterization and verification domains (e.g.,
acterized by color blindness (colorBlindness), color blind vs. color depth). In both rules, the test

534
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

for alternativeNonColoredInformation provides Future Trends


the bridge towards an actual accessibility check.
In this case, a software component attached to a As described earlier, Web accessibility is gaining
SRWL rules processor is triggered and analyses traction as time goes by, and as Web technologies
an HTML document accordingly. mature. Web designers and developers are becom-
More complex rules can be built on simpler ing more receptive to accessibility and inclusive
rules, thus affording sharing semantics between design practices, to broaden the spectrum of users
verification scenarios. This will afford a richer and that can be targeted by a Web site or Web applica-
more complete approach to the implementation tion. Still, they tend to follow blindly guidelines
and integration of SWAF into Web accessibil- such as WCAG, thus lacking the perception that
ity aware design and development tools. In the these do not cope with a high range of accessibility
next example we rework the two rules presented situations that are not taking into account.
above by refactoring the common set of rules in Furthermore, the lack of integration of acces-
a modular way: sibility tools within the design and development
process of Web sites and Web applications tend
to leave accessibility assessment procedures to
wao:hasGUITechnology(?APP, ?x1) quality assurance tasks or, at best, usability testing
& tasks. This fact sets Web accessibility assessment
wao:HTML(?x1) & as a patch to existing development processes,
wao:hasGUICharacteristics(?x1, which has consequences on the adequacy of Web
?x2) & sites and applications.
wao:characteristicName(?x2, “al- We believe that the entry point for disrupting
ternativeNonColoredInformation”) how Web accessibility is perceived nowadays must
& come from proper tool support, e.g., by means of
wao:characteristicValue(?x2, open and free Web accessibility tools that can be
“true”) plugged into existing Integrated Development
=> Environments (e.g., NetBeans, Visual Studio,
ex:verifyHTMLColor(?APP) etc.) and design tools (e.g., Dreamweaver). This
ex:verifyHTMLColor(?APP) & way, Web designers and developers will have an
gdco:hasDisability(?USER, acute sensibility for Web accessibility issues dur-
Individual(“colorBlindness”)) ing the design and development processes they
=> are working in.
wasgo:isDefiningValidApplica As an extra point, the Semantic Web Ac-
tion(Individual(“WAI_check- cessibility Framework detailed in this Chapter
point2.1”), ?APP) provides a fine-grained control of audiences and
ex:verifyHTMLColor(?APP) & their characteristics, and how these can cope with
do:hasFeature(?DEV, existing Web accessibility guidelines. We expect
Individual(“colorDepth1bit”)) that by providing this feature out-of-the-box,
=> Web design and development teams will bring
wasgo:isDefiningValidApplica audience-modeling procedures to their develop-
tion(Individual(“WAI_check- ment processes. This will give them more control
point2.1”), ?APP) on implementing Web sites and Web applications
that are accessible and verifiable during develop-
ment stages.

535
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

Figure 4. Architecture for accessibility validation services

Lastly, while Web accessibility is an important particular application and technology-dependent


issue to take into account, it is just the starting ontologies (e.g., Web sites and HTML). We envi-
point for providing digital services to end-users sion three extra components in this architecture
that are totally accessible and universally usable. that complement SWAF: (1) an ontology-oriented
Other domains, such as mobile phones, desktop search engine, where developers can search for
applications, or even embedded services must also information residing in SWAF-based knowledge
be targeted by accessibility assessment procedures bases, (2) an ontology integration tool, to afford
during early design and development stages. the specification of new domain-specific ontolo-
The SWAF framework described in this Chapter gies, and (3) a set of reporting tools centered on
will be extended in the future to cope with these providing concise information about accessibility
scenarios in a very effective way, since extensibil- assessment procedures.
ity is one of the core concerns inherent to it. On A supportive Integrated Development Envi-
Figure 4 we present our vision of the application ronment will tie these technologies to already
of SWAF in the context of application develop- existing features as a complementary facet of
ment (not just to Web sites and Web applications), development (e.g., similar to a debugging/helper
and how it can be tied to Integrated Development feature). We believe that enriching IDEs with
Environments. such features, as well as supportive accessibility
This architecture builds on the core technolo- simulation and reporting facilities will bring Web
gies and concepts defined in SWAF in different accessibility and general accessibility assessment
fronts. First, different application domains are procedures to a wide range of designers and de-
supported through a plug-in fashion (e.g., Web velopers, thus lowering the burden of providing
accessibility would be one of the plug-ins), based accessible applications to all users without any
on semantic technologies such as the Mapping and kind of barriers.
Rules ontologies, or the General Characteristics
and Disabilities ontology. An inference engine
would provide context-reasoning features tied to

536
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

Conclusion ADA. (1990). Americans with disabilities act.


Retrieved May 7, 2008, from http://www.usdoj.
This chapter presented SWAF, the Semantic Web gov/crt/ada
Accessibility Framework, as the foundation for
ASK-IT. (n.d.). ASK-IT ontological framework
the semantic description of Web accessibility
(Public Deliverable D1.7.1). ASK-IT FP6 Inte-
audiences, concepts and verification rules. This
grated Project.
framework provides the basic constructs for the
creation of Web accessibility verification engines Chen, H., Finin, T., & Joshi, A. (2003). An on-
that are capable of performing assessments tailored tology for context-aware pervasive computing
to specific user audiences and interaction devices. environments. In Proceedings of the IJCAI-03
We have divided SWAF into two dimensions, Workshop on Ontologies and Distributed Systems
Web Accessibility Descriptions (which includes (ODS 2003), Acapulco, Mexico.
general and domain specific concepts) and Web
Chen, H., Perich, F., Finin, T., & Joshi, A. (2004).
Accessibility Mapping (which affords semantic
SOUPA: Standard ontology for ubiquitous and
mapping and rules between concepts from the
pervasive applications. In Proceedings of the First
first dimension), in order to afford the extension
Annual International Conference on Mobile and
of SWAF into different domains in the scope of
Ubiquitous Systems: Networking and Services
Web accessibility.
(Mobiquitous 2004), Boston, MA, USA.
Ongoing work is currently being done in several
fronts in the SWAF realm, including: (1) building Chisholm, W., Vanderheiden, G., & Jacobs, I.
a comprehensive set of concept instances for user (1999). Web content accessibility guidelines 1.0.
and device feature characterization, (2) providing Retrieved May 6, 2008, from http://www.w3.org/
support for guidelines and standards other than TR/WAI-WEBCONTENT
WCAG, (3) improving the Mapping Ontology
DDA. (1995). Disability discrimination act. Re-
to cover more situations, (4) implement a robust
trieved May 7, 2008, from http://www.disability.
inference engine supporting SWAF concepts,
gov.uk/dda/index.html
(5) integrate this inference engine into existing
Integrated Development Environments and other Gruber, T. R. (1993). A translation approach to
Web site development tools, and (6) extend the portable ontology specifications. Source, Knowl-
SWAF ontology to cover other application do- edge, Acquisition, 5(2).
mains outside the scope of the Web.
Habra, N., Abran, A., & Lopez, M. (2004). Towards
a framework for measurement lifecycle (Tech Rep
TR37/04). University of Namur.
References
Heckmann, D. (2005). Ubiquitous user modeling.
Abecker, A., Bernardi, A., Van Elst, L., Lauer, Unpublished doctoral dissertation, Department of
A., Maus, H., Schwarz, S., & Sintek, M. (2001). Computer Science, Saarland University.
FRODO: A framework for distributed organiza-
tional memories (DFKI Document D-01-01, GFKI HFES. (n.d.). US human factors and ergonom-
GmbH). Milestone M1: Requirements Analysis ics society. Retrieved May 7, 2008, from http://
and System Architecture. Retrieved May 7, 2008, www.hfes.org
from www.dfki.uni-kl.de/frodo/ontology

537
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

Horrocks, I., Patel-Schneider, P. F., Boley, H., Neely, A. (1998). Measuring business perfor-
Tabet, S., Grosof, B., & Dean, M. (2004). SWRL: mance. Why, what and how? London: Profile
A Semantic Web rule language combining OWL Books Ltd.
and RuleML. Retrieved June 16, 2008, from http://
Obrenovic, Z., Abascal, J., & Starcevic, D. (2007).
www.w3.org/Submission/SWRL/
Universal accessibility as a multimodal design is-
ICF. (n.d.) International classification of function- sue. Communications of the ACM, 50(5), 83–88.
ing, disability and health. World Health Organi- doi:10.1145/1230819.1241668
zation. Retrieved May 7, 2008, from http://www.
Section 508. (1998). Civil rights division home
who.int/classifications/icf/site/index.cfm
page. Retrieved May 7, 2008, from http://www.
ISO 16071. (2003). ISO/TS 16071: 2003 ergonom- usdoj.gov/crt/508/508home.html
ics of human–system interaction. International
SENDA. (2001). Special educational needs and
Standards Organisation.
disability act. Retrieved May 7, 2008, from http://
Jacobs, I., Gunderson, J., & Hansen, E. (2002). www.dcs-gb.net/part4dda.html
User agent accessibility guidelines 1.0. Retrieved
Shneiderman, B. (2000). Universal usability.
May 6, 2008, from http://www.w3.org/TR/WAI-
Communications of the ACM, 43(5), 84–91.
USERAGENT/
doi:10.1145/332833.332843
Kasunic, M. (2006). The state of software measure-
Strang, T., Linnhoff-Popien, C., & Fank, K.
ment practice: Results of 2006 survey (Tech. Rep.
(2003). CoOL: A context ontology language to
CMU/SEI-2006-TR-009, ESC-TR-2006-009).
enable contextual interoperability. In Proceed-
Software Engineering Institute.
ings of the 4th IFIP International Conference on
Kulik, P. (2000). Software metrics: State of the Distributed Applications and Interoperabile Sys-
art. Retrieved May 7, 2008, from http://www. tems (DAIS2003), Paris, France. TRACE. (n.d.).
klci.com Designing more usable websites. Retrieved May
6, 2008, from http://trace.wisc.edu/world/web
Lopes, R., & Carriço, L. (2008). The impact of
accessibility assessment in macro scale universal Treviranus, J., Richards, J., Jacobs, I., & McCa-
usability studies of the Web. In Proceedings of the thieNevile, C. (1999). Authoring tool accessibility
5th International Cross-Disciplinary Conference guidelines 1.0. Retrieved May 6, 2008, from http://
on Web Accessibility, Beijing, China. www.w3.org/wai/au/wai-autools-19990621/wai-
autools.html
Masuwa-Morgana, K. R., & Burrell, P. (2004).
Justification of the need for an ontology for ac- Uschold, M., King, M., Moralee, S., & Zorgios, Y.
cessibility requirements (theoretic framework). (1998). The enterprise ontology. The Knowledge
Interacting with Computers, 16(3), 523–555. Engineering Review, Special Issue on Putting
doi:10.1016/j.intcom.2004.04.001 Ontologies to Use, 13.
McGarry, J., Card, D., Jones, C., Layman, B., Van Heijst, G., Schreiber, A. T., & Wielinga, B.
Clark, E., Dean, J., & Hall, F. (2002). Practical J. (2000). Understanding, building, and using
software measurement. Objective information for ontologies: A commentary to using explicit on-
decision makers. New York: Addison-Wesley. tologies in KBS development. LADSEB-CNR,
National Research Council.

538
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

Varkoi, T. (1999). Development of measurement Accessibility Guidelines: A set of best prac-


programs to support process improvement in tices that must be followed by designers and
small software companies. In Proceedings of the developers when implementing software solu-
FESMA’99 Conference, Amsterdam, The Neth- tions (e.g., Web site) that will help on providing
erlands (pp. 141-149). accessible information. By being guidelines, it
should not be assumed that content is accessible
Vigo, M., Kobsa, A., Arrue, M., & Abascal, J.
just by following them.
(2007). Quantitative metrics for measuring Web
Checkpoint: A concrete verification task that
accessibility. In Proceedings of the 4th Interna-
materializes a (part of a) guideline. Checkpoints
tional Cross-Disciplinary Conference on Web
can be fully automated if application technology
Accessibility, Banff, Canada. W3C. (n.d.). World
provides corresponding support (e.g., verifying if
Wide Web Consortium – Web standards. Retrieved
all images have associated textual captions).
May 6, 2008, from http://www.w3.org
Integrated Development Environment: A
WAI. (n.d.). Web accessibility initiative. Retrieved computer application used by developers that
May 6, 2008, from http://www.w3.org/WAI provides several features to ease the task of devel-
oping applications, such as text editor, compiler,
Wang, H. X., Zhang, Q. D., Gu, T., & Pung, H.
automation features, etc.
K. (2004). Ontology based context modeling
Universal Usability: a research field that stud-
and reasoning using OWL. In Proceedings of
ies the adequacy of user interfaces and information
2nd IEEE Conference on Pervasive Computing
to all users, regardless of their characteristics,
and Communications Workshops (PerCom 2004
knowledge, or mean of interaction (Shneider-
Workshops), Orlando, FL, USA.
man, 2000).
Wooldridge, M., Jennings, N., & Kinny, D. Usability: A research field that studies how
(2000). The Gaia methodology for agent-oriented adequate user interfaces are to users, how easily
analysis and design. Journal of Autonomous can they learn to perform tasks, and what are
Agents and Multi-Agent Systems, 3(3), 285–312. their levels of satisfaction when interacting with
doi:10.1023/A:1010071910869 user interfaces.
User Interface: The “visible” side of an ap-
plication, where users can acquire and interact
with information.
Key terms and Definitions
Web Accessibility: The subfield of accessibil-
ity that is targeted to the specific technologies and
Accessibility: The ability to access. Often
architecture that compose the World Wide Web.
tied to people with disabilities (e.g., total blind-
This includes technologies such as HTML, CSS
ness), accessibility thrives to break the barriers
and JavaScript, as well as the HTTP protocol.
to information access. We follow the strict sense
of accessibility by embracing any situation where
the ability to access information can be disrupted
by device or even surrounding environment
constraints.

539
A Service Oriented Ontological Framework for the Semantic Validation of Web Accessibility

APPENDIX

Table 1. WAI Web Content Accessibility Priority 1 Checkpoints

Cp Description
Cp1.1 Provide a text equivalent for every non-text element (e.g., via “alt”, “longdesc”, or in element content).
Cp2.1 Ensure that all information conveyed with colour is also available without colour, for example from context
or mark-up.
Cp4.1 Clearly identify changes in the natural language of a document’s text and any text equivalents (e.g., captions).
Cp6.1 Organize documents so they may be read without style sheets.
Cp6.2 Ensure that equivalents for dynamic content are updated when the dynamic con-tent changes.
Cp7.1 Until user agents allow users to control flickering, avoid causing the screen to flicker.
Cp14.1 Use the clearest and simplest language appropriate for a site’s content.
Cp1.2 Provide redundant text links for each active region of a server-side image map.
Cp9.1 Provide client-side image maps instead of server-side image maps except where the regions cannot be defined
with an available geometric shape.
Cp5.1 For data tables, identify row and column headers.
Cp5.2 For data tables that have two or more logical levels of row or column headers, use mark-up to associate data
cells and header cells.
Cp12.1 Title each frame to facilitate frame identification and navigation.
Cp6.3 Ensure that pages are usable when scripts, applets, or other programmatic objects are turned off or not supported.
If this is not possible, provide equivalent information on an alternative accessible page.
Cp1.3 Until user agents can automatically read aloud the text equivalent of a visual track, provide an auditory descrip-
tion of the important information of the visual track of a multimedia presentation.
Cp1.4 For any time-based multimedia presentation (e.g., a movie or animation), synchronize equivalent alternatives
(e.g., captions or auditory descriptions of the visual track) with the presentation.
Cp11.4 If, after best efforts, you cannot create an accessible page, provide a link to an alternative page that uses W3C
technologies, is accessible, has equivalent in-formation (or functionality), and is updated as often as the inac-
cessible (original) page.

This work was previously published in Handbook of Research on Social Dimensions of Semantic Technologies and Web Ser-
vices, edited by M.M. Cruz-Cunha, E.F. Oliveira, A.J. Tavares, and L.G. Ferreira, pp. 49-67, copyright 2009 by Information
Science Reference (an imprint of IGI Global).

540
541

Chapter 2.16
Building Semantic Web
Portals with a Model-
Driven Design Approach
Marco Brambilla
Politecnico di Milano, Italy

Federico M. Facca
Leopold-Franzens-Universität Innsbruck, Austria

Abstract The authors generalize the development process to


cover Semantic Web needs and devise a set of new
This chapter presents an extension to Web appli- primitives for ontology importing and querying. The
cation conceptual models toward Semantic Web. chapter also presents a comparison of the proposed
Conceptual models and model-driven methodolo- approach with the most relevant existing proposals
gies are widely applied to the development of Web and positioned with respect to the background and
applications because of the advantages they grant adopted technologies.
in terms of productivity and quality of the outcome.
Although some of these approaches are meant to
address Semantic Web applications too, they do not Introduction and Motivation
fully exploit the whole potential deriving from in-
teraction with ontological data sources and from se- Modern Web applications comprise distributed data
mantic annotations. The authors claim that Semantic integration, remote service interaction, and manage-
Web applications represent an emerging category of ment of workflow activities, possibly spawned on
software artifacts, with peculiar characteristics and different peers. In this scenario, a wider attention to
software structures, and hence need some specific the semantics of data and applications is mandatory
methods and primitives for achieving good design to allow effective design and evolution of complex
results. In particular the contribution presented in systems. Indeed, if semantics of data and applica-
this chapter is an extension of the WebML modeling tions is known, their integration becomes more
framework that fulfils most of the design require- feasible. Moreover, explicit semantic annotation of
ments emerging in the new area of Semantic Web. Web applications can facilitate content search and

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Building Semantic Web Portals

access, and foster a future generation of Web clients We claim that conceptual modeling and model-
that exploit the semantic information to provide driven development can increase dramatically the
better browsing capabilities to customers. efficiency and efficacy of the design and imple-
The Semantic Web aims at bringing formal mentation of such applications, by offering tools
“semantics” to the human-readable information and methodologies to the designer for specifying
so as to make it machine-readable and allow better semantically-rich Web applications.
and easier automatic integration between different The model-driven approach to software
Web applications. To address this challenge, many development has been proven valid in several
semantic description languages arose, like RDF, application fields and is currently one of the best
OWL and WSML; some of them are currently practices of the software engineering discipline.
W3C Recommendations. All these languages Developing a Semantic Web application, as with
allow to formally model knowledge by means of any other kind of software system, is a complex
ontologies: the resulting formal models are the achievement that requires the ability to master
starting point to enable easy information exchange a broad spectrum of tasks, jointly performed by
and integration between machines. a number of persons with different skills for a
These languages are suitable for reasoning and long timeline. Software engineering and Web
inference, i.e., to deduct more information from engineering (Ceri et al., 2002) demonstrated that
the model by applying logical expressions. This following a well organized development process,
makes the modeling task easier since not all the centered on the appropriate modeling concepts,
knowledge has to be modeled. These languages is essential to overcome the complexity inherent
are supported by a wide range of tools and APIs, to such kind of developments.
that cover design of knowledge (e.g., Protégé (Noy This chapter aims at demonstrating how
et al., 2001) and OntoEdit (Sure et al., 2002)), model-driven design can impact on specifica-
provide storing facilities (e.g., Sesame (Aduna, tion, design, and implementation of Semantic
2007) and Jena (HP, 2007)), and offer reasoning Web portals as well. In the proposed approach,
on the data (e.g., Racer (Racer Systems, 2007) and we leverage a conceptual modeling approach for
Pellet (Sirin et al.,2007)). Based on these modeling visually designing the Web application domain
languages, a set of querying languages have been model and hypertext model. Conceptual mod-
devised too; among them, we can mention TRIPLE eling works at higher abstraction levels with
(Sintek & Decker, 2002) and SPARQL (W3C, respect to direct implementation design, thus
2007), a W3C candidate recommendation. allowing to specify the application design with a
Unfortunately, although the theoretical bases top-down philosophy. The first design steps aim
and some technological solutions are already in at describing the platform-independent domain
place for Semantic Web support, the techniques and hypertext models, disregarding the low-level
and methodologies for Semantic Web applica- details. Further refinements can take more and
tion design are still rather rough. This leads to more into account such details, finally leading
high costs of implementation for Semantic Web to the expected outcome of the work. These ap-
features, even if embedded within traditional Web proaches take great advantage from the adoption of
applications. These extra costs are related not only C.A.S.E. (Computer Aided Software Engineering)
to the design of the architecture and deployment tools, which provide facilities for designing and
of the Semantic platforms, but also to the repeti- implementing applications according to specific
tive and continuous task of semantic annotation methods and notations. C.A.S.E. tools sometimes
of contents and application pages. include partially or completely automatic code

542
Building Semantic Web Portals

generation too. on). The Section Requirements for Semantic Web


As a concrete example of modeling language, engineering presents a set of requirements that Web
we pick WebML (Web Modeling Language) (Ceri Engineering approaches for modeling Semantic
et al., 2002), a well known methodology and set Web applications must comply with, according
of metamodels in the Web engineering environ- to our analysis; The Section Modeling Semantic
ment, and its companion C.A.S.E. tool WebRatio Web application with WebML proposes our exten-
(WebModels s.r.l., 2007). sion of a metamodel for the design of Semantic
To discuss our approach and related back- Web portals (i.e., new primitives and metamodels
ground technologies, a running example will be within the WebML framework); in Section Case
used throughout the chapter. We discuss a realistic Study: A Music Semantic Portal we present a
scenario based on the reuse of existing ontolo- realistic Semantic Web portal scenario designed
gies available on the Internet that can be easily and implemented with the proposed methodology.
exploited to create new Semantic Web Portals. The Section Implementation Experience discusses
In particular, we will consider two ontologies for the architectural and implementation aspects of
the musical domain to build a Web application the proposed approach. Finally Section Conclu-
offering access to music contents, considering sion and Future Research Directions propose a
also users profile information. We combine the summary of the lesson learned from the research
MusicBrainz ontology (http://www.musicbrainz. presented and possible research direction starting
org) for the music domain information; the Mu- from the work reported in the chapter.
sicMoz (http://www.musicmoz.org/) hierarchy
to classify music genres; the RDF Site Summary
(RSS-DEV Working Group, 2000) for music news; Background
and the Friend Of A Friend (Foaf) ontology (Miller
& Brickley, 2000) to describe for user’s profiles This section presents an overview of the back-
and relationships among them. The case study ground technologies and relevant work in the
application is similar to other existing Semantic field of engineering Semantic Web applications.
Web applications (e.g., http://Foafing-the-music. In particular we first introduce an overview of the
iua.upf.edu/, http://squiggle.cefriel.it/music/), most relevant languages adopted for defining and
that provide personalized access to the contents querying ontologies; then we present a review
exploiting distributed semantic information. The of relevant Web engineering methodologies that
presented application, although rather simple cover Semantic Web technologies; a subsequent
because of space reasons, can be considered a full- paragraph is dedicated to introduce WebML, the
fledged Semantic Web Portal since it aggregates Web application modelling language that we
different sources of information spanned across extend for supporting Semantic Web features
the Internet and presents them in a structured and (extensions are discussed in Section Modeling
user-friendly manner. Semantic Web application with WebML).

Structure of the Chapter Languages for the Semantic Web

This chapter is organized as follows. The Section Several ontology languages have been developed
Background contains a description of the back- during the last few years, and they will surely be
ground technologies, metamodels and method- exploited in the context of the Semantic Web.
ologies for Web engineering and Semantic Web Some of them are based on XML syntax, such
applications (i.e., WebML, OWL, RDF, and so as Ontology Exchange Language (XOL), SHOE

543
Building Semantic Web Portals

(which was previously based on HTML), and Uniform Resource Identifier (URI). In the case
Ontology Markup Language (OML), whereas of a Web page, the URI can be the Unified Re-
Resource Description Framework (RDF) (W3C, source Locator (URL) of the page. The URI does
2004b) and RDF Schema (W3C, 2004c) are lan- not necessarily enable the access via the Web to
guages created by World Wide Web Consortium the resource; it simply unambiguously identifies
(W3C) working groups. Later, two additional the resource.
languages have been built on top of RDF(S) - the The subject of a statement is the resource we
union of RDF and RDF Schema - to improve its want to make a statement about. For example if
features: Ontology Inference Layer (OIL) and we want to make a statement about a music art-
DAML+OIL. DAML+OIL was submitted to ist, we need to identify it with a URI (e.g. http://
W3C as starting proposal for a new ontology musicbrainz.org/mm/mm-2.1#artist_id1234). The
language: OWL (W3C, 2004a). Recently, a new predicate defines the kind of information we want
emerging initiative, the Web Service Modeling to express about the subject. For example if we
Ontology (WSMO), introduced a new ontology want to make a statement about the name of the
language called WSML (Fensel et al., 2006). A Artist we can use the URI http://musicbrainz.org/
more detailed discussion can be found in Chapter mm/mm-2.1#name that references the property
Ontology Languages for the Semantic Web: An defined in the ontology. The object defines the
Overview. The ontologies adopted in our case study value of the predicate, for example in our case
are a mix of RDF(S) and OWL, that are shortly we want to state that the name of artist is “Bono”.
presented in the next paragraphs. The object can be a literal, like in this example,
or another resource represented by the object’s
Resource Description URI.The statement presented can be written in
Framework and RDF Schema triple notation asa:<http://musicbrainz.org/mm/
mm-2.1#artist_id1234> <http://musicbrainz.org/
RDF (W3C, 2004b), developed by the W3C for mm/mm-2.1#name> “Bono”.RDF supports also
describing Web resources, allows the specifica- an XML syntax for writing and exchanging RDF
tion of the semantics of data based on XML in a graphs, called RDF/XML. The RDF/XML repre-
standardized, interoperable manner. It also pro- sentation of the previous statement is:<rdf:RDF
vides mechanisms to explicitly represent services, xmlns:mm=”http://musicbrainz.org/mm/mm-
processes, and business models, while allowing 2.1#” xmlns:rdf=”http://www.w3.org/1999/02/22-
recognition of non explicit information. The RDF rdf-syntax-ns#”><rdf:Description
data-model is based on subject, predicate, object rdf:about=”http://musicbrainz.org/mm/mm-
triples, so called RDF statements, to formalize 2.1#artist_id1234”><mm:name>Bono</
meta-data. RDF is domain independent in that no mm:name></rdf:Description></rdf:RDF>
assumptions about a particular domain of discourse A statement can be represented as a graph too,
are made. It is up to the users to define their own as depicted in Figure 1.
ontologies for the user’s domain in an ontology The RDF data model does not provide mecha-
definition language such as RDF Schema (RDFS) nisms for defining the relationships between prop-
(W3C, 2004c). erties (attributes) and resources - this is the role
RDF statements are based on the concept of of RDFS (W3C, 2004c). RDFS offers primitives
resources, which can be used in the different roles for defining knowledge models that are closer to
of the statement. A resource can be everything; frame-based approachesb.
a book, a person, a Web page, a CD, a track on a In particular RDFS introduces “meta”-classes
CD, and so on. Every resource is identified by a that are used to define resource types: rdfs:Class

544
Building Semantic Web Portals

Figure 1. The graphical representation of a RDF statement.

to define RDF classes (concepts); rdfs:Literal to • Cardinality restrictions. RDFS does not
define literal values such as strings and integers; provide any means to restrict the number
rdfs:Property to define properties. The full RDFS of distinct values a property may or must
vocabulary and its description can be found in take.
(W3C, 2004c). The following fragment defines the • Disjoint classes. RDFS does not provide
concept artist as previously introduced:<rdfs:Class primitives to declare two classes to be
rdf:about=”http://musicbrainz.org/mm/mm- disjoint.
2.1#Artist”rdfs:label=”Artist”><rdfs:subClass • Set combination of classes. RDFS does not
Ofrdf:resource=”http://www.w3.org/2000/01/ allow to define new classes by building
rdf-schema#Resource”/></rdfs:Class><rdf:Pr the union, intersection, or complement of
opertyrdf:about=”http://musicbrainz.org/mm/ other classes.
mm-2.1#name” rdfs:label=”name”><rdfs:domain • Special characteristics of properties. In
rdf:resource=”http://musicbrainz.org/mm/mm- RDFS is not possible to define that a prop-
2.1#Artist”/><rdfs:range rdf:resource=”http:// erty is transitive (e.g. greater than), unique
www.w3.org/2000/01/rdf-schema#Literal”/></ (e.g. is mother of), or the inverse of another
rdf:Property> property.
RDF and RDFS are not rich enough to com-
pletely describe an ontology as they still miss The following OWL fragment represents
some important concepts within their Descrip- a OWL properties and its inverse. The in-
tion Logic. verseOf element specifies that playedBy is
the inverse of plays.<!-- Artist plays a Track-
Ontology Web Language -><owl:ObjectProperty rdf:ID=”http://mu-
sicbrainz.org/mm/mm-2.1#plays”><rdfs:domain
OWL (W3C, 2004a) is designed for applications rdf:resource=”http://musicbrainz.org/mm/mm-
that need to process the content of information 2.1#Artist”/><rdfs:range rdf:resource=”http://
instead of just presenting information to humans. musicbrainz.org/mm/mm-2.1#Track”/></
Furthermore OWL facilitates greater machine owl:ObjectProperty><!-- Track played
interpretability of Web content than that sup- by an Artist --><owl:ObjectProperty
ported by XML, RDF, and RDF Schema (RDFS) rdf:ID=”http://musicbrainz.org/mm/
by providing additional vocabulary along with a mm-2.1#playedBy”><rdfs:domain
formal semantics. rdf:resource=”http://musicbrainz.org/
OWL provides a number of additional model- mm/mm-2.1#Track”/><rdfs:range
ling primitives that increase the expressiveness rdf:resource=”http://musicbrainz.org/
compared to RDFS and solve some shortcomings mm/mm-2.1#Artist”/><owl:inverseOf
of RDFS: rdf:resource=”http://musicbrainz.org/mm/mm-

545
Building Semantic Web Portals

2.1#plays” /></owl:ObjectProperty>OWL has Furche et al., 2006); while a broader discussion


three increasingly-expressive sublanguages: OWL of Web and Semantic Web query languages is
Lite, OWL DL, and OWL Full. presented in (Bailey et al., 2005). In the next
paragraph we give some more details on SPARQL
• OWL Lite supports those users primarily since it is probably the most adopted query lan-
needing a classification hierarchy and sim- guage for RDF and it is the language adopted by
ple constraint features. our solution.
• OWL DL includes all OWL language con-
structs with restrictions such as type sepa-
ration (a class can not also be an individual SPARQL
or property; a property can not also be an
individual or class). OWL DL is so named SPARQL (W3C, 2007) is a query language that has
due to its correspondence with description already reached candidate recommendation status
logics, a field of research that has studied a at the W3C, and is on a good way to become the
particular decidable fragment of first order W3C recommendation for RDF querying. Que-
logic. rying RDF data with languages in the SPARQL
• OWL Full gives the maximum expressive- family amounts to matching graph patterns that
ness and the syntactic freedom of RDF are given as sets of triples of subjects, predicates
with no computational guarantees. For ex- and objects. The triple syntax adopted by SPARQL
ample, in OWL Full a class can be treated is based on Turtle (Beckett, 2004). Solutions to
simultaneously as a collection of individu- SPARQL queries are given in the form of result
als and as an individual in its own right. It sets: each result set contains a set of mappings
is unlikely that any reasoning software will from the variables occurring within the query to
be able to support every feature of OWL nodes of the queried data. For instance, the query
Full. For instance, general inference in that extracts the name and the nationality of all
OWL Full is clearly undecidable as OWL the Artists from the ontology previously presented
Full does not include restrictions on the use can be written as:BASE <http://musicbrainz.org/
of transitive properties which are required mm/mm-2.1#>PREFIX mm: <http://musicbrainz.
in order to maintain decidability. org/mm/mm-2.1#>PREFIX rdf: <http://www.
w3.org/1999/02/22-rdf-syntax-ns#>SELECT
Query Languages for ?name ?nationalityWHERE{ ?id rdf:type
the Semantic Web mm:Artist .?id mm:name ?name .OPTIONAL{
?id mm:nationality ?nationality .}}
With the diffusion of RDF data sources, the The WHERE clause specifies the graph pattern
problem of querying them in a easy way caused to selected data to be mapped to the variables;
the flourishing of many different query languages variables are identified by either ? or $ prefix.
for RDF based on different principles: some are The OPTIONAL clause specifies that a certain
inspired by SQL, like SPARQL (W3C, 2007) and triple is optional: it is not compulsory that all the
RQL (Karvounarakis, 2004), others are based returned results contains the triples stated has
on different principles like graph patterns - e.g. optional. If the OPTIONAL clause is removed,
TRIPLE (Sintek & Decker, 2002) - and reactive the results will include only the triples that contain
rules - e.g. Algae (Prud’hommeaux, 2004). In the whole graphical pattern requested. The FROM
depth comparison of the most relevant RDF query clause can be used to specifies the URL (or some
languages can be found in (Haase et al., 2004; other identifier) of the data to be queried.

546
Building Semantic Web Portals

SPARQL includes also other three types of phases: conceptual design, navigation design,
query: the CONSTRUCT query clauses, that create abstract interface design, and implementation.
new RDF graphs with data from the RDF graph The conceptual design builds the conceptual
queried; the DESCRIBE query that return RDF class schema for the application domain. This
“descriptions” of the resources matching the query schema is described in UML extended with a few
part (e.g., the query DESCRIBE mm:u2 returns a new characteristics like the ability to specialize
RDF graph representing the mm:u2 resource); and relations. The UML diagram is mapped to an OWL
the ASK query that return true or false according model according to some heuristics rules.
to the fact that the specified pattern has a solu- The navigation design defines the navigational
tion or not (e.g., the query ASK mm:u2 rdf:type class schema and the navigational context schema.
mm:Artist returns true if the resource mm:u2 is The main navigational primitives are navigational
an instance of the class mm:Artist). classes (nodes), navigational contexts, and access
structures. In the same way as for the conceptual
Methodologies to Design class schema, one can specialize navigational
Semantic Web Applications relations. The mappings between the conceptual
schema and navigational class schema are defined
While design methodologies for traditional Web using RQL. The navigation context allows the
applications offer rather mature and established description of sets of navigational objects.
solutions methodologies for developing Semantic The abstract interface design defines the visual
Web applications are still in an early development aspect of the deployed application by means of
phase. Realizing the benefits of the Semantic Web the abstract widget ontology and concrete widget
platform (e.g., interoperability, inference capabili- ontology. The implementation phase produces a
ties, increased reuse of the design artifacts, etc.) Semantic Web application based on the previous
traditional design methodologies are now focusing SHDM specifications.
on designing Semantic Web applications: e.g., Figure 2 presents the Navigational Class
OOHDM (Schwabe & Rossi 1998) evolved in Schema (top part) and the Navigational Context
SHDM (Lima & Schwabe, 2003). New meth- Schema (bottom part) of a simple Web applica-
odologies like XWMF (Klapsing et al., 2001), tion based on the musicbrainz and musicmoz
OntoWebber (Jin et al., 2001) and Hera (Vdovjak ontology.
et al., 2003) were specifically designed by con-
sidering the Semantic Web peculiarities. Among
them, the most complete are Hera and SHDM that HERA
are shortly described in the next paragraph.
The Hera Methodology (Vdovjak et al., 2003) is
The Semantic Hypermedia a model driven methodology for designing and
Design Method developing Web applications using Semantic Web
technologies. It is developed from a database per-
SHDM (Lima & Schwabe, 2003) is an ontology- spective; the conceptual modelling (similar to the
based design methodology. It extends the expres- ER modelling) and the querying of the different
sive power of OOHDM (Schwabe & Rossi, 1998) Hera models are important issues in the proposed
by defining ontologies for each of the OOHDM methodology.
models. These ontologies are specified in OWL, a The Hera methodology has three main layers:
more expressive language than RDFS. In the same the conceptual model layer (CM), the application
way as OOHDM, SHDM identifies four different model layer (AM), and the presentation model

547
Building Semantic Web Portals

Figure 2. SHDM navigational design for a Web application based on the case study ontologies: Navi-
gational Class Schema (top part) and Navigational Context Schema (bottom part).

layer (PM). The first layer describes data content By distinguishing these different layers, Hera
used for generation of hypermedia presentations differentiates at design level between the semantic
and makes available data coming from different, aspects, the navigational aspects, and the interface
possibly heterogeneous data sources; the integra- aspects of a Semantic Web application.
tion of different data sources is possible thanks to The core components of the application model
a data integration model (a sub-layer that allows layer are called slices. They are associated with
decoupling between the conceptual model and a concept from the CM model and may contain
the data sources). The second layer describes the properties of the concept or other slices. Different
navigation structure and functionality; finally the not nested slices may be interconnected by slice
PM layer describes spatial layout and rendering relationships, that can be classified in aggregation
of hypermedia presentations. An orthogonal layer, relationships (e.g., index, tour, indexed guided
called adaptation layer, captures adaptation issues tour, etc.) or reference relationships (i.e., links
in all the above layers. with an anchor specified). Slices represent views

548
Building Semantic Web Portals

Figure 3. An Hera application model based on the case study ontologies

over the conceptual model and are translated to the extension we introduced to model Semantic
RQL queries. Web applications is presented later in this chapter.
All the HERA models are, as matter of fact, WebML uses conceptual modeling techniques for
ontologies, and hence a Web application model describing web applications. The WebML design
is a collection of instances of the different model methodology comprises three main phases: data
ontologies. Therefore, it is possible to define a design, hypertext design, and implementation. It
fragment of an application model as subclass of provides a visual notation and a XML serialization
another fragment (this applies also the other two for the proposed models.
models). For specifying the data underlying the applica-
Figure 3 presents a fragment of an application tion, WebML exploits an extended version of the
model based on the musicbrainz and musicmoz Entity-Relationship model, which consists of enti-
ontology: the first slice on the left represents a ties (classes of data elements), and relationships
page that contains a genre and the set of artists (semantic connections between entities).
linked to that genre; each artist is linked with The hypertext design defines the navigational
a reference link to the artist slice that collects structure of the application. WebML also allows
information about the artist an the set of tracks designers to describe hypertexts, called site views,
played by the artist. for publishing and managing content. A site view
is a piece of hypertext, which can be browsed by
a particular class of users. Multiple site views can
WebML be defined for the same application. Site views
are then composed of areas and pages. Areas and
Our approach to Semantic Portals specification is pages can be nested in areas. Finally pages are
based on WebML (Ceri et al., 2002). In this para- the containers of elementary pieces of content,
graph we introduce the basic features of WebML; called content units, typically publishing data

549
Building Semantic Web Portals

Figure 4. An example Web page modeled in WebML (left) and its underlying data model (right)

retrieved from the database, whose schema is pertext model and its underlying E-R model. It
expressed through the data model. In particular, shows a single page, which allows the user to see
WebML primitives for content publishing denote his reviews about music albums. In the My Home
alternative ways for displaying one or more entity Page a Get unit retrieves the identifier of the cur-
instances: e.g., the data unit publishes a single rent user from the session parameter CurrentUser
instance of data, the index unit a list of data. Unit and provides it in input to the following Data unit,
specification requires the definition of a source which publishes the user’s personal profile data.
and a selector: the source is the name of the entity The data to be published are restricted by means
from which the unit’s content is extracted; the of a selector condition, specified below the unit.
selector is a condition, used for retrieving the The user identifier is further propagated to the
actual objects of the source entity that contribute Index unit My Reviews by means of a transport
to the unit’s content. link. The index unit shows the list of reviews
Between units/pages one can define hyper- related to the received user identifier.
textual links as oriented connections. WebML WebML-based development is supported by
distinguishes several types of links: navigational, the WebRatio CASE tool (WebModels s.r.l., 2007),
automatic, and transport links. These links can which offers a visual environment for designing
carry information from the source to the des- the WebML conceptual schemas, storing them in
tination. The information is stored in the link XML format, and automatically generates the run-
parameters. The navigational links require user ning code (through XSLT model transformations),
intervention, while both automatic and transport which is deployed as pure J2EE code.
links are traversed without user intervention: for
automatic links once the source is presented also
the associated destination is shown; the transport Requirements for Semantic
links do not define navigation and are solely used Web Engineering
to transport information.
WebML also supports the specification of While Web Engineering managed to bring software
content management operations. They allow engineering practices to the Web development
creating, deleting or modifying an instance of an area, no discipline has addressed the peculiar
entity (respectively through the create, delete, and needs of Semantic Web application design. A
modify units), or adding or dropping a relationship good methodology for the design of Semantic
between two entity instances (respectively through Web applications must provide additional facili-
the connect and disconnect units). ties to the developer, specifically addressing the
Figure 4 presents an example of WebML hy- new needs. To understand them, we define the

550
Building Semantic Web Portals

features of these applications, and then infer the ontology; (2) to seamlessly integrate new
characteristics of the methodologies and models ontologies, not fitting the default ontol-
that must be provided to the designer. To collect ogy; and (3) to reuse existing and shared
the requirements that a Semantic Web application ontologies.
should comply with, we analyzed some existing
Semantic Web Portals (e.g., http://www.mind- From the previous set of characteristics, we
swap.org, http://Foafing-the-music.iua.upf.edu, derived the following requirements for the con-
and http://ontoworld.org) and we extracted the ceptual models pursuing the design of Semantic
following set of needs: Web applications:

• Support of semantic languages: Semantic • Semantic Web application models should be


Web applications should be aware of and aware of and support semantic languages.
support (i.e., be able to query and manage) • The models themselves should be “seman-
different Semantic Languages and meta- tic”, i.e., grant self-annotation and explicit
models (RDFS, OWL, WSML, …). semantic extraction.
• Semantic application models: Semantic • The models should allow flexible inte-
Web applications should be designed and gration of heterogeneous sources and
specified by means conceptual models that applications.
include and support semantic descriptions. • The models should allow transformations
• Flexible integration: Semantic Web ap- towards a query language able to capture
plications should embrace the philosophy all the aspects of ontologies, including in-
of flexibility and heterogeneity integration ference, verification, query on instances,
of Semantic Web. and query on classes.
• Classes and instances access and queries: • The models should easily allow: to specify
Both domain ontology classes and instanc- semantic data sources as underlying level
es should be easily and seamlessly accessi- of the application; to exploit these sources
ble by Semantic Web applications, through for populating Web pages, and for (auto-
appropriate querying primitives, including matically) annotating such Web pages.
data instance and structure queries. • The models shall be able to import and
• Inference and verification: Ontology- reference distributed data and ontologies,
based web applications should exploit aiming at the reuse and sharing of the
available inferring systems on ontological knowledge.
data, both for semantic queries and verifi-
cation of data.
• Semantic data sources: A Semantic Web Modeling Semantic Web
application relies on semantic data (e.g., Applications with WebML
ontologies) that offer a machine under-
standable data description that may be In this section we give an overview of extensions
used to populate and generate Web pages to the WebML methodology and models that we
and also to provide semantic annotations. developed for complying with the requirements
• Importing and reuse of ontologies: of Semantic Web applications (see previous sec-
Semantic Web applications shall allow tion). WebML showed its flexibility and ease of
to: (1) import new (possibly distributed) extension in many other contexts (e.g., Web Ser-
data conforming to the Web application vices, Processes, Adaptive Web applications, Rich

551
Building Semantic Web Portals

Figure 5. The extended development process for Semantic Web applications, new and modified steps
have yellow background

Internet applications): this allowed to consolidate Extending the WebML


and standardize the extension process. Basically, Development Process
extending WebML for a new design domain (like
Semantic Web portals) requires to introduce The injection of semantics within Web applica-
changes to the overall development process and tions requires the extension of the methodology
to the related metamodels (and if required to adopted in the development of “traditional” Web
introduce new metamodels in the development applications with additional tasks that formal-
process). In particular, to deal with Semantic Web ize the new design steps. Figure 5 depicts the
applications, we analyzed the current aspects of extended version of the development process for
the WebML methodology and provided the proper Web applications (yellow blocks represent the
extensions for each of them: the development pro- new tasks we introduced to fulfill Semantic Web
cess is enriched with steps that allow to describe application requirements). The original version
the tasks related to the design of ontologies and was proposed in (Ceri et al., 2002) and is adopted
semantics of the web applications/services; the with slight variations by most of the existing Web
data model is extended to support semantic data Engineering approaches. We want to stress that
sources (i.e., ontologies); the hypertext model is this development process is of general purpose
enriched with new primitives that support ontology and can be used for developing Semantic Web
querying, with particular attention to advanced and applications regardless of the use of any model-
inferring queries. Finally the presentation model driven technique (actually it is valid also in case
is extended to allow for semantic annotations of of traditional development techniques).
the applications. During Requirements Specification, a software
analyst derives information about the applica-
tion domain and functional requirements for the
application from the business requirements and

552
Building Semantic Web Portals

produces a detailed and formal specification for the implementation. After testing, the application
application designers. Already at this step some is released and undergoes ordinary maintenance
requirements related to semantic aspects can and evolution activities. Notice that the waterfall
be highlighted and properly formalized. Based representation may be adjusted for some design
on the resulting specification, the designer can experiences, considering that in some cases some
select existing domain ontologies to be imported steps are not needed at all (e.g., if only imported
(Ontology Import) and integrated in the Web ap- ontologies are necessary, the data design step can
plication. Both ontology schemas and ontology be skipped). We did not depict all the variants for
instances maybe imported at this point. Never the sake of clarity.
less, for the Web application design, only ontology In the following paragraphs we discuss in more
schemas are relevant. The imported ontologies can detail the changes introduced to the development
be possibly modified or merged to better suite the process (yellow blocks in Figure 5).
Web application purposes with specialized tools
like Protégé (Noy et al., 2001). During the Data Extending the WebML Data Model
Design phase, the database structure and new on-
tologies can be created. Then, during the Hypertext Current existing model-driven methodologies
Design the actual Web application structure is for Semantic Web applications either evolved
designed; in this phase, new primitives allows to from existing ones by extending their data source
specify how to query ontologies. Notice that the coverage to ontologies, or have born with native
design of interactions with relational data remains support of semantic data sources.
unchanged with respect to the standard WebML Although ontology support is obviously nec-
design techniques. At this point the designer takes essary for Semantic Web applications design, we
decisions about the Presentation Design of the Web think that relational data sources can still provide
application (e.g. design of graphical mock-ups and great added value to Web applications. Relational
resources) and about the semantic annotations he databases are still a valuable option for modeling
wants to include in the rendered pages. The An- portion of domain data where features offered
notation Design enriches the Hypertext Design by ontologies are not needed (e.g., hierarchies,
and relies on the Presentation Design for decid- polymorphism, reasoning support, …) and where
ing the actual position, formatting, and display performances are a key issue (e.g., transactional
style of annotations. Architecture Design mainly data).
concerns the definition of hardware and software Therefore, allowing seamless interaction be-
components as well as the design of the required tween ontologies and databases is a desideratum
network infrastructure. Once an overall applica- of current Semantic Web applications. Notice
tion design has been specified, the Implementation that we do not aim at extending the data model
phase can be carried out. It produces in output of WebML so as to model ontologies (see Noy
the software components (databases, ontologies, et al. 2001), but at allowing Web applications to
HTML templates, business logics, etc.) that build query imported semantic knowledge together
up the actual Web application, running on the se- with relational sources. By adopting a conceptual
lected architecture. During the following Testing model, like WebML, interaction between ontol-
and Evaluation phase, several tests (regarding e.g. ogy instances and database instances can be quite
functionality, usability, and performance) validate straightforward.
the application’s conformance with respect to the The designer is in charge of carefully deciding
initial business requirements, and could lead to what is going to be part of an ontological data
novel derived requirements or modifications of source and what is going to part of a relational data

553
Building Semantic Web Portals

source. As data design (both in case of ontologies As stressed above, these basic WebML primi-
and databases) in general is an already consoli- tives remain valid, nevertheless they require some
dated discipline, we do not provide further details small extensions to support challenges posed by
in this chapter. The only critical aspect is that the ontological data sources. Ontologies allow for
designer needs also to foresee if there is any need queries with a wider expressive power and require
of references between the two data sources (e.g., some different modeling rules for the information
the URI identifier of a Foaf user profile may be within respect to relational data. This reflects into
referenced in the relational schema to connect changes in the notations that the primitives must
some relational attributes). use for defining the conditions and the selection
of the data.
Extending the WebML Some of the challenges posed by ontologies
Hypertext Model are:

WebML comes with a basic set of primitives for • There is no distinction between relation-
data access (e.g., Index unit, Multidata unit, the ships and attributes within the set of prop-
Data unit) that have a general purpose meaning erties of a class. E-R style relationships
(see Section “Background”) and are perfectly fit- might be considered as ontological proper-
ting in the role of query and navigation primitives ties having an URI as value, and attributes
for both relational and ontology sources. Indeed, to ontological properties having a literal as
this is a general feature of conceptual models: if value;
the used abstractions are generic enough, they do • Several Semantic Web framework (e.g.,
not need to be changed when their data source OWL, RDF) assume that any instance of a
grounding is changed. They can be easily extended class may have an arbitrary number (zero or
for supporting the additional expressive power more) of values for a particular property;
and the different data model of the ontological • Properties specification may include cardi-
sources. nality constraints and classes as range (and
For example, in WebML the Index unit is of course domain). In this case, it is pos-
extended so that, besides extracting lists of sible to publish as values also structured
relational instances, allows to produce lists of objects and not only atomic attributes.
instances of a particular class within an ontology
model. In previous sections we highlighted some The revised units allow to model in a visual
requirements related to this kind of query: (1) the and simple way queries over ontologies. The
possibility to show only direct instances or also data integration between relational data sources
inferred instances; (2) the need for querying both and different ontologies can be tackled directly
instances and classes, thus mixing instances and in the Hypertext Model, by exploiting the data
(sub)classes in the results too. The same principles flow mechanism provided by WebML. One of
can be applied to Multidata unit and Data unit. the main advantages of WebML is the ability to
The Hierarchical Index unit, already defined specify business logic of applications by intercon-
in WebML, fits perfectly with the ontological data necting smaller business logic components (units)
sources since it can be extended to browse and in chains and passing parameters between them.
publish a portion of ontology in a hierarchical This allows to define complex business logics
tree representation: for instance, given a class, it composed by units that may query different on-
allows to publish the hierarchical tree underlying tologies or relational data sources and then can
it, comprising subclasses and instances. exchange information along the chain of transport

554
Building Semantic Web Portals

links in a very easy way. Thus, parameters on the properties, values; to check existence of specific
links become the actual contact point between concepts; and to verify whether a relationship
traditional and semantic data and provide the holds between two objects. These units have
mechanism for orthogonalizing data issues and a polymorphic behaviour: their business logic
hypertext issues. and the results returned change according to the
configuration of their input parameters. The com-
Advanced Data Access Primitives plete summary of the behaviour of these units is
presented in Table 1.
Many possible queries, using semantic data Some other capabilities are needed within
sources and semantic query languages, cannot be Semantic Web applications. For instance, we may
expressed using the basic data access primitives need to import at runtime ontologies instances
presented in the previous section. By carefully compliant with ontological schemas defined (or
analyzing the semantic query languages presented imported) at design time; to extract semantic
at the beginning of the chapter, we introduce a descriptions of ontology portions; or to merge
new set of operational primitives that cover ad- and compose ontologies. To this purpose, we
vanced queries over ontological data available in introduce three new units, visually represented in
these languages. In particular these new units are Figure 7. The Set Composition operation unit is
largely influenced by two languages presented in able to perform classic set operations (i.e., union,
the background section, namely, SPARQL (W3C, intersection, difference) over two input sets of
2007) and RDF Schema syntax (W3C, 2004c). URIs, considering the hierarchy of the URIs in-
SPARQL was also selected among the various volved. E.g. suppose we have two set of classes:
semantic query languages as the one adopted in the A = {ProgressiveRock, Jazz, Metal}B = {Rock,
implementation of the runtime components (see JazzFusion}.In this case, the set operation will give
Section Implementation Experience) because of its the following results:A ∩ B = {ProgressiveRock,
large software support and because it is probably Metal, JazzFusion}A ∪ B = {Rock, Jazz}
going to become a W3C Recommendation. since Rock is superclass of ProgressiveRock
The core set of new units aims to fill the gap and Metal, and Jazz is superclass of JazzFusion.
between the basic WebML querying primitives and The Import Ontology unit allows to import at
the increased expressive power of semantic lan- run time ontological data sources consistent with
guages. The basic primitives can not be combined one or more of the ontology models imported at
to design any query that heavily exploits reasoning design time. This unit validates imported data
and mixing between schemas and instances. Thus, against existing schemas and according to the
the basic WebML primitives miss two of the main designer choice, allows to store only the url of the
assets of semantic languages. For example, they newly imported ontology (i.e., it will be remotely
do not allow to infer all the classes an instance queried every time, so modification on the remote
belongs to; they do not allow to infer the classes source will be propagated to the application) or
a particular property belongs to (in ontological to import the ontology in the local OWL/RDF
models, a property can be associated to more than repository (i.e., it will be accessed locally, but
one class, since it is modeled independently from modifications to the original data will not be
the class itself). The new units (i.e., SubClassOf, propagated to the application). Notice however that
InstanceOf, HasProperty, HasPropertyValue, in our methodology the navigational model of the
SubPropertyOf), described in Figure 6, aim at Web application cannot be changed dynamically at
providing explicit support to advanced ontologi- runtime, thus if the imported ontology contains a
cal queries and allow to extract classes, instances, new class unrelated with already existing classes

555
Building Semantic Web Portals

queried in the hypertext, these new class will not that allows to retrieve artists or albums that are
be reachable in the navigation of the hypertext. correlated to artist searched by the user. The value
The Describe unit returns the RDF descrip- submitted in the form is passed to an index unit
tion of an URI, thus enabling data exporting and that, by means of a filter, extracts all the Artists
semantic annotation of pages. whose name contains in the string submitted by
The different advanced querying units are de- the user. The URI of the Artist picked from the
signed such that they can be combined in chains index is passed to the HasPropertyValue unit.
to compose complex business logic pattern that This unit extracts a set of URIs (instances of the
class Album or Artist) that have the passed URI as

Figure 6. The new WebML units for advanced queries on ontologies

enables reasoning over ontological data. For in-


stance, Figure 8 reports a fragment of the portal value of the relatedWith property. The set of URIs

556
Building Semantic Web Portals

Table 1. Design time and runtime descriptors for two semantic units

Design Time Descriptor Runtime descriptor


Semantic <SWINDEXUNIT class=”mf:Track” id=”swinu1” <descriptor service=”org.webml.onto.
Index name=”Tracks” ontology=”onto1”> SWIndexUnitService”>
unit <DisplayedProperties <onto>onto1</onto>
property=”mf:title”/> ...
<DisplayedProperties <input-params>
property=”mf:descriptor”/> <input-param type=”mm:Artist”
<SortProperties order=”ascending” name=”swdau2.rdf:ID” />
property=”mf:title”/> </input-params>
<Filter boolean=”or”> ...
<FilterCondition id=”fselector1” <query type=”SELECT”>
property=”mf:playedBy” SELECT DISTINCT ?instance ?p1 ?p2
predicate=”eq” name=”Artist”/> WHERE {?instance rdf:type mm:Track .
</Filter> ?instance mm:title ?p1 .
</SWINDEXUNIT> ?instance mm:descriptor ?p2 .
?instance mm:playedBy ?fs1 .
FILTER (?fs1 = $swdau2.rdf:ID$)}
ORDER BY DESC(?p1)
</query>
</descriptor>
Subclas- <SUBCLASSOFUNIT id=”iof1” <descriptor service=”org.webml.onto.
sOf unit name=”SubClassOf” ontology=”onto1”> SubClassOfUnitService”>
</SUBCLASSOFUNIT> <onto>onto2</onto> ...
<input-params>
<input-param type=”mz:Genre”
name=”swinu3.rdf:ID” />
</input-params>
<query type=”SELECT”>
SELECT ?id WHERE
{?id rdfs:subClassOf $swinu3.rdf:ID$}
</query>
</descriptor>

is then passed to the InstanceOf unit that checks if Extending the Presentation Model
they are instances of the class Artist. In this case, to Support Semantic Annotations
the URIs are passed over through the OK link to
an index unit showing a list of Artist, otherwise the Each WebML semantic unit has been designed as
URIs are passed on the KO link to publish a list of able to automatically extract a RDF description
Album (not shown in the figure). of its contents. The designer has to specify in the
HTML templates how he wants to use the RDF
annotations. This RDF fragments can be used to

Figure 7. Symbols of the new WebML semantics management units

557
Building Semantic Web Portals

Figure 8. A portion of a Semantic Web application described by the new WebML units

annotate a single portion of the page (the one where proper reference to its property URI (setting the
the unit is rendered) or they can be aggregated and property attribute in the surrounding HTML tag),
published together in a single predefined location, and reference resource (setting the about attribute
as a global semantic annotation of the whole page in the surrounding HTML tag). Once the designer
itself. Annotations can be made visible in the page has defined the generic template, correct annota-
for user reference or kept hidden in the HTML tions are extracted in an automated way at runtime
code if they are meant to be used only by machine according to the underlying ontological model.
readers. The WebML presentation model (that al- Figure 9 presents an example of data unit with
lows to place objects in a page grid, according to the relative generated RDFa annotation.
appropriate styles) can be exploited for defining
how and where semantic annotations are rendered
within the generated pages. If advanced annotation Case Study: a Music
is needed, the solution is the Describe unit, which Semantic Portal
allows to query the ontologies to extract complex
RDF fragments to be used as annotations. In this In this section we show a simplified example of
way, it is possible to extract semantic fragments a Semantic Web portal modelled with WebML
also of instances or classes not published in any leveraging on the extensions we introduced in the
content unit of the displayed page. Different tem- previous section. For the design of the application
plates may be applied to different portions of the we adopt the ontologies cited in the Introduction,
application according to the requirements of the namely the MusicBrainz ontology, the ontolo-
developed Web application. gization of the MusicMoz hierarchy; the RDF
For example, the designer may decide to adopt Site Summary for music news; and the Friend
a template that generates RDFa (Adida & Birbeck, Of A Friend (Foaf) ontology. The main ontology
2007) annotations. RDFa is a syntax that expresses adopted in the case study application is the Mu-
the structure underlying the published data us- sicBrainz ontology; Figure 10 visually represents
ing a set of elements and attributes that embed a fragment of it. The application includes also a
RDF in HTML. An important goal of RDFa is to small relational schema that contains registration
achieve this RDF embedding without repeating information for the users, described by the User
existing HTML content when that content is the entity with its login data and the URI of its cor-
structured data. Within our framework RDFa is responding Foaf profile, if he has imported one
particularly easy to adopt since published data in the Web application.
reflects an ontology schema; each displayed Figure 11 reports a fragment of the WebML
property value can be simply published with the model for the proposed application: content

558
Building Semantic Web Portals

Figure 9. An example of data unit with the relative generated RDFa annotation (highlighted in black).

units with the RDF symbol use ontological data of Artist corresponding to his preferences. From
sources (e.g., Artists index unit), while the other here (and from any other page presenting a visual
units publish data from the a relational database query over the Artist class) the user can browse
(e.g., User Data data unit). The user starts his the Artist details page, where detailed informa-
navigation from the User Home Page, where the tion about the selected Artist and his Album are
Foaf Profile data unit is published; the user also presented. The user can ask for the exporting of
import a profile if it is not available yet. This part the RDF description of the Artist he is currently
of the application actually shows how integration browsing. From the Albums index unit is possible
between ontological data sources and relational to reach the Album Details page that reports infor-
data can be achieved using parameters transported mation on the navigated instance of Album and
over links: when the user imports the ontology that its Track instances. Then accessing the landmark
represents his Foaf profile, he actually stores the page Search by genre (a landmark page is a page
URI of the profile in the User relational entity; accessible from any point in the hypertext also
this URI is later used to publish his Foaf profile without an explicit link), the user can navigate
from the ontology repository according to the a hierarchical representation of the class Genre,
database schema. and then accesses all the Artist instances that are
The user can navigate from Foaf Profile data related to the selected genre. The SubClassOf unit
unit to the Suggestion page that presents an index extracts indirect sub-genres of the chosen one, thus

559
Building Semantic Web Portals

Figure 10. A fragment of the MusicBrainz ontology representing Artist, Album, and Track, with the
respective relationships

allowing to display associated artists. Finally the choice is made by the designer from the CASE
News page reports an index of RSS Item and, by tool). The runtime environment offered by We-
selecting an item from the list, the user can display bratio has been extended exploiting the plug-in
the Artist associated with it (thanks to the filter that mechanism of the toolsuite:
allows to retrieve Artist instances whose name or
alias is included in the title or in the description • We devised a general purpose ontology
property of the item). New sources of RSS Item data access layer to be exploited by every
may be imported adopting the Import Ontology unit;
unit, such as in the case of the Foaf profile. • Then, we developed a runtime Java com-
ponent and an XML descriptor for each
unit.
Implementation Experience
Some extensions were needed on the design
In this section we discuss the architecture design time interface too, in order to provide the proper
we adopted to implement the presented exten- management of ontological sources and units. Fig-
sions to the WebML metamodel. These design ure 12 reports a screenshot of the extended design
choices are discussed according to the reference environment showing the use case Semantic Web
implementation of WebML, the Webratio toolsuite application presented in the previous section.
(WebModels s.r.l., 2007). As API to integrate
ontologies both in the design environment and Implementing the Ontological Units
in the runtime framework we adopted the Jena
framework (HP, 2007), while the reasoning sup- In the WebRatio framework, each unit is imple-
port is obtained by means of the Jena integrated mented by means of a generic class representing
reasoner, or by means of the integration of Pellet the runtime component that is executed for every
(Sirin et al., 2007) with the Jena framework (the instance of that kind of unit. Then, for each new

560
Building Semantic Web Portals

Figure 11. A portion of a WebML diagram for a Semantic Music Portal

unit (including the revisited traditional units that behave according to the logics specified in the
access ontologies) we developed an XML descrip- runtime descriptors, defined for each instance
tor specifying its parameters, its properties, the of the unit.
binding to the implementation classes, and so on.
To better clarify the structure of the descriptor, Ontological Data Source Layer
we show two examples of an ontological unit
descriptors (see left column of Table 1). By means We defined a new data access layer to allow the
of an associated XSLT transformation, design time interaction with ontologies, comprising a set of
descriptors are translated to runtime descriptors general purpose Java classes to be reused by all the
that include automatically generated template new units for querying the ontology repositories.
of SPARQL queries (right column of Table 1). These classes provide facilities to import ontolo-
Units are implemented by Java components that gies and to select OWL/RDF classes, properties,

561
Building Semantic Web Portals

Figure 12. The Semantic Web Portal for the Music domain modelled with the WebRatio CASE tool

and instances (possibly filtered by one or more services use or implement these general purpose
conditions). The main aspects of the class structure services.
are represented in Figure 13. The OntologyMod-
elService enables connections to local and remote
ontologies specified at design time or imported Conclusion
at runtime by means of the Import Ontological
Source unit. Three abstract classes offer the query In this chapter we presented our research on the de-
services corresponding to the query methods sign of Semantic Web applications; we introduced
offered by SPARQL on the ontology contents: briefly all the used background technologies and
the AbstractSelectQueryService class perform languages that are at the core of our proposal or
selection over data (SPARQL SELECT query); that are adopted by other well-known frameworks.
the AbstractDescribeQueryService retrieves the To ease the comparison of our work with other
RDF describing a given URI (DESCRIBE query), solutions, we described in detail Hera and SHDM,
the AbstractAskQueryService verifies simple the most complete existing solutions.
predicates (ASK query). The AbstractAskQuery- The core of this chapter is the presentation
Service is extended by the AskQueryService that of our extension to the WebML methodology
is used by some of the advanced querying units to and models for supporting the design and the
verify predicates (e.g., to check whether a class is specification of Semantic Web applications. The
subclass of another). In general, ontological unit described solution provides a full coverage of the

562
Building Semantic Web Portals

Figure 13. UML class diagram of the class hierarchy of the new implemented units

development process: it allows the designer to techniques to import other data sources within
specify basic and advanced queries on ontological ontologies).
data sources, to import existing sources, and to SHDM does not allow to import ontologies but
annotate Web pages with semantic descriptions only to create them from UML diagrams. Then, it
of the contents and of the models. Our approach offers a tricky way to link these ontologies to the
provides substantial added value with respect external ones. Only a specific WSDM extension
to the existing frameworks for Semantic Web (Casteleyn et al., 2006) provides an approach to
application design, although some of them are annotate pages so as to make them machine read-
more advanced on some aspects (e.g., seamless able. Most of the new methodologies offer runtime
integration of different ontologies). frameworks that include or allow integration of
Table 2 reports a summary that compares the reasoners, while some of them do not clarify if the
features of the previously cited models for Se- reasoning is supported also at design time.
mantic Web Portals and the WebML extensions An important factor to assure the success of
presented in this chapter. All the methodologies, a design methodology is the existence of CASE
except for XWMF, have a complete development tool support, since a powerful methodology that
methodology that covers all the needed aspects to is not accompanied by adequate tools will make
create a Semantic Web application. They also offer the designer tasks very difficult to fulfil. While
a wide support for ontology languages: basically most of the traditional design methodologies have
all the models support both RDF and OWL (ex- powerful CASE tools, no established tool support
cept for XWMF). However, our extension is the is provided for Semantic Web design, although all
only one that leverages on Semantic Web query the cited methodologies offer some basic tools.
languages to offer advanced query primitives that We support our proposal with a prototype imple-
allow both query on schema and instances, together mentation within the CASE tool WebRatio.
with simple reasoning patterns over data. Other
models (e.g., Hera) offer query on data schema
and instances. Hera and OntoWebber offer direct Future Research Directions
support to data integration by means of an integra-
tion model that can be used to query different data Future research will span on several directions:
schemas using the same query. WebML offers the one of the main aspect we want to address is to
chance to integrate relational, XML and ontology generalize the way for extending the querying
data sources, while other methodologies seem expressive power of the WebML units. Indeed,
to support explicitly only ontologies (of course, the provided extensions address some specific
this issue can be solved by adopting extraction additional query power, but if further query ex-

563
Building Semantic Web Portals

Table 3. Comparison of methodologies for modeling Semantic Web portals

Requirement XWMF OntoWebber SHDM Hera WebML+Sem.


Methodology Partial Yes Yes Yes Yes
Semantic Model Description Yes Yes Yes Yes Partial
Advanced query support No Partial Partial Partial Yes
Flexible integration No Partial Yes Yes Partial
Heterogeneous data sources No No Partial Partial Yes
Distributed data sources No No No Yes Yes
Reuse of ontologies Yes Yes Partial Yes Yes
(Automatic) Annotation No No No No Yes
Reasoning Support No No Yes Yes Yes

pressive power is needed, researchers will need • Extensive testing of the new framework,
to devise new appropriate units. With a general including the application of the approach
representation framework, we plan to avoid this to real industrial scenarios;
burden and to provide a quicker and more compact • Integration of existing Eclipse based so-
way for obtaining the results. lutions for ontology editing with in the
Another interesting aspect that was high- CASE tool.
lighted by this extension experience is related to
the traditional way in which WebML provides
querying facilities: standard WebML units (index
unit, multi-data unit, data unit, and so on) provide References
both query service and publishing features in one
single component (the unit specified within the W3C (2004a, February). OWL Web ontology
page). Our extensions instead provide components language overview: W3C recommendation. Re-
that offer only the query services, without deal- trieved June 16, 2007, from http://www.w3.org/
ing with publishing, which is left to traditional TR/owl-features/.
WebML units. This separation of concerns could W3C (2004b, February). RDF Primer: W3C
probably be applied to traditional querying too, recommendation. Retrieved June 16, 2007, from
so that the two aspects are always kept separate. http://www.w3.org/TR/owl-features/
This would mean a deep rethinking of the WebML
hypertext models. We plan to consider this hint W3C (2004c, February). RDF Vocabulary De-
and to study possible solutions to the problem in scription Language 1.0: RDF Schema W3C
the immediate future. Recommendation. Retrieved June 16, 2007, from
Besides pure research topics, we plan to extend http://www.w3.org/TR/rdf-primer/
and refine our implementation. Future implemen-
W3C (2007, June). SPARQL Query Language
tation work includes:
for RDF: W3C Candidate Recommendation. Re-
trieved June 26, 2007, from http://www.w3.org/
• Providing a content integration layer to al-
TR/rdf-sparql-query.
low for seamless integration of different
ontologies;

564
Building Semantic Web Portals

Adida, B., & Birbeck, M. (2007). Rdfa primer 1.0: Furche, T., Linse, B., Bry, F., Plexousakis, D.,
Embedding RDF in XHTML, W3C Working Draft & Gottlob, G. (2006). RDF querying: Language
12 March 2007. Retrieved June 16, 2007, from constructs and evaluation methods compared. In
http://www.w3.org/TR/xhtml-rdfa-primer/ Proceedings of Summer School Reasoning Web
2006, Lisbon (pp. 1–52). Springer.
Aduna (2007). Sesame. Retrieved June 26, 2007,
http://www.openrdf.org/. Haase, P., Broekstra, J., Eberhart, A., & Volz, R.
(2004, November 7-11). A comparison of RDF
Bailey, J., Bry, F., Furche, T., & Schaffert, S.
query languages. In Proceedings of the third In-
(2005). Web and Semantic Web query anguages:
ternational Semantic Web Conference, Hiroshima,
A survey. In J. Maluszinsky and N. Eisinger
Japan, (pp. 502–517). Springer.
(Eds.), Reasoning Web Summer School 2005, (pp.
35–133). Springer-Verlag. HP. (2007). Jena a Semantic Web framework
for Java. Retrieved June 26, 2007, http://jena.
Beckett, D. (2004). Turtle - Terse RDF triple
sourceforge.net.
language. Retrieved June 16, 2007, from http://
www.dajobe.org/2004/01/turtle/. Jin, Y., Decker, S., & Wiederhold, G. (2001,
July 30-August 1). OntoWebber: Model-Driven
Brambilla, M., Celino, I., Ceri, S., Cerizza, D.,
ontology-based Web site management. In I.F. Cruz,
Della Valle, E., & Facca, F. M. (2006, November
S. Decker, J. Euzenat, & D.L. McGuinness (Eds.),
5-9). A software engineering approach to design
Proceedings of SWWS’01, The First Semantic
and development of Semantic Web service ap-
Web Working Symposium, Stanford University,
plications. In The Semantic Web - ISWC 2006,
California, USA. (pp. 529–547)
5th International Semantic Web Conference,
ISWC 2006, Athens, GA, USA, Proceedings (pp. Karvounarakis, G., Magkanaraki, A., Alexaki,
172-186). Springer. S., Christophides, V., Plexousakis, D., Scholl,
M., & Tolle, K. (2004). RQL: A functional query
Casteleyn, S. Plessers, & P., Troyer, O.D. (2006,
language for RDF. In Gray, P., King, P., & Pou-
July 11-14). Generating semantic annotations dur-
lovassilis, A. (Eds.), The Functional Approach
ing the Web design process. In Proceedings of the
to Data Management (pp. 435–465). Springer-
6th international Conference on Web Engineering
Verlag.
(Palo Alto, California, USA). ICWE ‘06 (pp. 91-
92). ACM Press. Klapsing, R., Neumann, G., & Conen, W. (2001,
November). Semantics in Web engineering:
Ceri, S., Fraternali, P., Bongio, A., Brambilla, M.,
Applying the resource description framework.
Comai, S., & Matera, M. (2002). Designing data-
[IEEE Computer Society.]. IEEE MultiMedia,
intensive Web applications. Morgan Kauffmann
8(2), 62–68. doi:10.1109/93.917972
Fensel, D., Lausen, H., Polleres, A., de Bruijn, J.,
Lima, F., & Schwabe, D. (2003). Application mod-
Stollberg, M., Roman, D., & Domingue, J. (2006).
eling for the Semantic Web. In 1st Latin American
Enabling Semantic Web services: The Web service
Web Congress (LA-WEB 2003), Empowering Our
modeling ontology. New York: Springer-Verlag.
Web, Sanitago, Chile, Proceedings (pp. 93–102).
IEEE Computer Society.
Miller, L., & Brickley, D. (2000). Foaf project.
Retrieved June 16, 2007, from http://www.Foaf-
project.org.

565
Building Semantic Web Portals

Noy, N. F., Sintek, M., Decker, S., Crubezy, M., Additional Reading
Fergerson, R. W., & Musen, M. A. (2001). Creating
Semantic Web contents with Protègè 2000. [IEEE Berners-Lee, T. (2003). Web services - Semantic
Computer Society.]. IEEE Intelligent Systems, Web talk. Retrieved June 16, 2007, from http://
16(2), 60–71. doi:10.1109/5254.920601 www.w3.org/2003/Talks/08-mitre-tbl/

Prud’hommeaux, E. (2004). Algae RDF query Berners-Lee, T., Hendler, J., & Lassila, O. (2001).
language. Retrieved June 16, 2007, from http:// The Semantic Web. Scientific American, 5(284),
www.w3.org/2004/05/06-Algae/. 34–43.

Racer Systems. (2007). RacerPro. Retrieved June Brambilla, M., Ceri, S., Fraternali, P., & Ma-
16, 2007, from http://www.racer-systems.com/. nolescu, I. (2006c). Process modeling in Web
applications. [ACM Press.]. ACM Transactions
RSS-DEV Working Group. (2000). RDF Site on Software Engineering and Methodology, 15(4),
Summary (RSS) 1.0. Retrieved June 16, 2007, 360–409. doi:10.1145/1178625.1178627
from http://web.resource.org/rss/1.0/spec.
Broekstra, J., Kampman, A., & van Harmelen, F.
Schwabe, D., & Rossi, G. (1998, October). An ob- (2002, June 9-12). Sesame: An architecture for
ject-oriented approach to Web-based application storing and querying RDF and RDF schema. In
design. In Theory and Practice of Object Systems Proceedings of the First International Semantic
(TAPOS), 207–225. John Wiley & Sons Web Conference (ISWC 2002), Sardinia, Italy,
Sintek, M., & Decker, S. (2002). TRIPLE - A (pp. 54-68). Springer.
Query, inference, and transformation language for De Troyer, O., Casteleyn, S., & Plessers, P. (2007).
the Semantic Web. In 1st International Semantic WSDM: Web semantics design method. In Rossi,
Web Conference, ISWC 2002, Chia, Sardinai, Italy, G., Pastor, O., Schwabe, D., & Olsina, L. (Eds.),
Proceedings (pp. 364 - 378). Springer. Web Engineering: Modelling and Implementing
Sirin, E., Parsia, B., Grau, B. C., Kalyanpur, A., & Web Applications, Human-Computer Interaction
Katz, Y. (2007). Pellet: A Practical OWL-DL Rea- Series, 12, 303-351. Springer.
soner. [Elsevier]. Journal of Web Semantics, 5(2), Fernandez, M. F., Florescu, D., Levy, A. Y., &
51–53. doi:10.1016/j.websem.2007.03.004 Suciu, D. (2000). Declarative specification of
Sure, Y., Erdmann, M., Angele, J., Staab, S., Studer, Web sites with Strudel. The VLDB Journal, 9(1),
R., & Wenke, D. (2002). OntoEdit: collaborative 38–55. doi:10.1007/s007780050082
ontology development for the Semantic Web. Fons, J., Pelechano, V., Albert, M., & Pastor, O.
In 1st International Semantic Web Conference, (2003, October 13-16). Development of Web
ISWC 2002, Chia, Sardinai, Italy, Proceedings applications from Web-enhanced conceptual
(pp. 221 - 235). Springer. schemas. In Conceptual Modeling - ER 2003,
Vdovjak, R., Frasincar, F., Houben, G. J., & Barna, 22nd International Conference on Conceptual
P. (2003). Engineering Semantic Web information Modeling, Chicago, IL, USA, Proceedings (pp.
systems in Hera. [Rinton Press]. Journal of Web 232–245). Springer.
Engineering, 2(1-2), 3–26.
WebModels s.r.l. (2007). Webratio tool. Retrieved
June 16, 2007, from http://www.webratio.com

566
Building Semantic Web Portals

Karvounarakis, G., Christophides, V., Plexousakis, Staab, S., & Maedche, A. (2001). Knowledge Por-
D., & Alexaki, S. (2001, October 29-November 2). tals - Ontologies at Work. [AAAI.]. AI Magazine,
Querying RDF descriptions for community Web 21(2), 63–75.
portals. In 17iemes Journees Bases de Donnees
Tzitzikas, Y., Spyratos, N., Constantopoulos, P., &
Avancees (BDA’01) Agadir, Maroc Proceedings
Analyti, A. (2002). Extended Faceted Taxonomies
(pp. 133–144).
for Web Catalogs. In Third International Confer-
Kifer, M., Lausen, G., & Wu, J. (1995). Logical ence on Web Information Systems Engineering,
foundations of object-oriented and frame-based WISE 2002, Singapore, December, 2002 (pp.
languages. [ACM Press.]. Journal of the ACM, 192–204). IEEE Computer Society
42, 741–843. doi:10.1145/210332.210335
van der Sluijs, K., Houben, G. J., Broekstra, J., &
Maedche, A., Staab, S., Stojanovic, N., Studer, Casteleyn, S. (2006). Hera-S - Web Design Using
R., & Sure, Y. (2003). Semantic portal - The seal Sesame. In Proceedings of the 6th international
approach. In Fensel, D., Hendler, J., & Lieber- Conference on Web Engineering (Palo Alto, Cali-
man, H., (Eds.), Spinning the Semantic Web (pp. fornia, USA, July 11 - 14, 2006). ICWE ‘06 (pp.
317-359). MIT Press, Cambridge, MA. 337–344). ACM Press.
Murugesan, S., Deshpande, Y., Hansen, S., & van Harmelen, F., Horrocks, I., & Patel-Sch-
Ginige, A. (2001). Web engineering: A new dis- neider, P. (2001). Reference Description of the
cipline for development of Web-based systems. DAML+OIL (March 2001) Ontology Markup
[Lecture Notes in Computer Science. Springer.]. Language. Retrieved June 16, 2007, from http://
Web Engineering, 2016, 3–13. doi:10.1007/3- www.daml.org/2001/03/reference.html
540-45144-7_2
Vdovjak, R., Barna, P., & Houben, G. J. (2003).
OMG. (2003). Model driven architecture guide EROS: Explorer for RDFS-based Ontologies.
v1.0.1. Tech. rep., Object Management Group. In ACM International Conference on Intelligent
Retrieved June 16, 2007, from http://www.omg. User Interfaces, IUI 2003, Proceedings (pp.
org/docs/omg/03-06-01.pdf. 330). ACM.
OMG. (2006). Ontology definition metamodel.
tech. rep., object management group. Retrieved
June 16, 2007, from http://www.omg.org/cgi-bin/ Endnotes
doc?ad/06-05-01.pdf
1
The serialization reported is the TURTLE
Rossi, G., Schwabe, D., & Lyardet, F. (1999, No- (Beckett 2004) RDF serialization format.
vember). Web application models are more than 2
Frame based languages are metalanguages
conceptual models. In 18th International Confer- that apply the frame concept to the structur-
ence on Conceptual Modeling (ER 99), Paris, ing of language properties. They are rather
France, Proceedings (pp. 239–252). Springer. focused on the recognition and description
Staab, S., Angele, J., Decker, S., Erdmann, M., of objects and classes, and relations and
Hotho, A., Maedche, A., Schnurr, H.-P., Studer, interactions are considered as “secondary”.
R., & Sure, Y. (2000). Semantic Community They are widely adopted for ontology model-
Web Portals. Computer Networks, 33(1-6) (pp. ing since inference based on them it’s easier
473–491). Elsevier. than using other paradigms.

567
Building Semantic Web Portals

Appendix: Questions for Discussion

Beginner

Q1: Are model-driven techniques influenced by technology changes?


A: Model-driven techniques aim at abstracting the modeling layer of applications from the actual
implementation layer. I.e., model-driven approaches should capture abstract characteristics of the ap-
plication field so that their model is resilient to technology changes and to the different implementation
platforms that can be chosen. For example, the WebML methodology proposed in this chapter showed
how abstraction introduced to handle relational data are still valid for semantic data sources, proving
that the abstraction level of the adopted primitives doesn’t depend on the underlying implementation
and the underlying technologies (SQL or SPARQL). The same discussion applies to SHDM that evolved
from OOHDM.
Q2: Why the import and reuse of ontologies is a key requirement for Semantic Web applications?
A: Reuse is one of the key goals of Software Engineering and is usually widely applied to software
components and libraries. This aspect is even more emphasized at the level of semantic content speci-
fication and usage. Thanks to the introduction of Semantic Web languages, like OWL, also data model
can be widely reused and shared. Indeed, one of the key factors for the success of the Semantic Web
initiative is the spread of a set of domain ontologies with a wide consensus on their definition. This will
enable different applications to be transparently integrated thanks to the fact that they are sharing the
same data model.

Intermediate

Q3: Why the adoption of a standard query language is important in the definition of metamodels for
Semantic Web applications?
A: The use of a standard query language contributes to abstract furthermore the modeling layer
from the actual implementation. Indeed, if the implementation of the conceptual primitives relies on a
specific API, this implies the change of the whole querying mechanism (comprising the generation of
queries) if a slightly different implementation is chosen. On the other hand, the use of a query language
(like SPARQL) enables to change the adopted implementation at the lower level of the adopted query
engine. This is particularly evident with relational database technologies, where the SQL language (and
possibly some standardized APIs like JDBC/ODBC) allows for (theoretical) total independence from
the actual database engine adopted. Semantic Web technologies are still evolving very quickly, there-
fore a similar decoupling is still missing, but probably the wide adoption of the SPARQL language will
provide a similar benefit.

Advance

Q4: The WebML extension for Semantic Web achieves different source integration thanks to parameter
passing on the links between WebML components. Is there any other valuable option that still maintains
valid the primitives proposed and offer the opportunities to integrated different data sources?
A: This issue can be easily solved by adopting one of the currently available semantic bridges that
allows to use a single domain ontology to which heterogeneous data sources can be mapped (e.g.,

568
Building Semantic Web Portals

relational databases, ontologies, web services, xml documents). The mapped data sources can then
be accessed through queries (e.g. SPARQL queries) to the domain ontology. The adoption of such a
technology will affect only the implementation layer, maintaining valid the design primitives proposed
and the generated SPARQL queries from the ontology model. The main difference stands on the level
where integration occurs: instead of integrating the sources at the application modelling level, sources
are integrated at the data design level.

Practical Exercises

Exercise 1. Define a navigation model for allowing users to navigate the popular Wine ontology using
the WebML notation (http://www.w3.org/TR/owl-guide/wine.rdf)
Solution:
The simplest solution is to provide access to the wines directly through the wine categorization provided
in the ontology or through flavors. A hierarchical index can be used for browsing the wine categories
and selecting a wine. The Wine Details page shows the information of the selected wine.
Two simple index units can be used to display the flavors and hence to show the wines of the selected
flavour. Once a wine is chosen, the user is sent to the Wine Details page again.
Exercise 2. Define a navigation model using the SHDM model-driven methodology for the Wine
ontology that displays a result similar to the navigation model created in WebML.
Solution:
The content model shows the fragment of ontology that has been used. The navigation model repre-
sents a similar navigation to the one shown for WebML
Exercise 3. Model a WebML diagram that allows to select a Grape instance and then retrieves all the
classes of Wine that use the selected Grape.
Solution:
The solution consists of an index unit that allows to select instances of class WineGrape, and of a
chain of semantic query units that select all the objects made from the chosen grape (the HasProperty-
Value unit) and then, among the returned objects, extract all the classes of the elements of the returned
objects (InstanceOf unit). Finally, the list of classes is shown in an index. Notice that the parameter
[URI] represents the set of URIs resulting from the respective queries.

569
Building Semantic Web Portals

Figure 14.

This work was previously published in Semantic Web Engineering in the Knowledge Society, edited by J. Cardoso and M.
Lytras, pp. 46-106, copyright 2009 by Information Science Reference (an imprint of IGI Global).

570
571

Chapter 2.17
Enabling Distributed Cognitive
Collaborations on the Semantic
Web
Amna Basharat
National University of Computer and Emerging Sciences, Pakistan

Gabriella Spinelli
Brunel University, UK

Abstract work called Cognitive Modelling of Multi-Agent


Action (COMMAA) for modeling agents’ actions
To date research on improving the state of in an attempt to provide an architecture that im-
multi-agent collaboration has only focused on proves the flexibility of Multi-agent interaction
the provision of grounding tools, technologies, by promoting cognitive awareness. The results
protocols, standards and infrastructures that drive of the evaluation show an improved flexibility,
the Semantic Web and agent architectures. The interoperability and reusability of agents’ collec-
basic cognitive and interactional requirements of tive behaviours and goals.
agents have been neglected leading to the current
state-of-the-art development of the Semantic
Web whereby its full potential is constrained by Introduction
the rigid state of multi-agent collaboration. This
chapter illustrates and discusses an alternative ap- Agents may be autonomous and intelligent entities
proach to the development of the agent mediated which typically operate in distributed collabora-
Semantic Web. The fundamental premise of our tive environments called Multi-Agent Systems
approach is that enhancing agents cognitive and (MAS) which allows multiple heterogenous agents
interactional abilities is the key to make the digital to collaborate by engaging in flexible, high-level
world of agents more flexible and adaptive in its interactions (Wooldridge, 2002; Jennings 2000).
role to facilitate distributed collaboration. The Presently, the usability of agent-based applica-
novelty of this research is that it adapts cognitive tions in a Semantic Web environment is limited
models from HCI to develop a heuristic frame- due to lack of flexibility in agent’s collaboration

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Enabling Distributed Cognitive Collaborations on the Semantic Web

with multiple agents including humans. This Based on the rationale and concerns described
imposes constraints on the interoperability and above, the objectives and a brief outline of the
reusability of agents’ behaviour that operate in chapter presented in the next section.
MAS environment. In addition, the inflexibility
of the agents’ behaviour does not provide direct
mapping to the end user since the end user cannot Objectives of the Chapter
predict how the agent will behave, thus generating
cognitive overload on humans. To date, research The following chapter will serve the following
on improving the state of multi-agent collaboration aims and objectives:
has only focused on the provision of grounding
tools, technologies, protocols, standards and • Delineate upon the current limitations in
infrastructures that drive the Semantic Web and the state of multi-agent collaboration in or-
agent architectures. Neglect of basic cognitive and der to elaborate the rationale, need and the
interactional requirements are discovered to be synergistic role of cognitive dimension to
the basic reasons for the rigid state of multi-agent the Semantic Web with particular regard to
collaboration constraining its full potential. distributed collaborations amongst agents
This research presented in this chapter adapts • Describe the conceptual constituents of
a distributed cognitive view of the agent medi- a theoretical framework called Cognitive
ated Semantic Web and argues that enhancing Model of Multi-Agent Action (COMMAA)
cognition is the key to make the digital world derived from cognitive models in HCI to
of agents more flexible and adaptive in its role improve the state of multi-agent collabora-
to facilitate distributed collaboration. To this tion
end, work on imparting cognition to improve • Detail upon the Design and Implementation
interaction between multiple agents has been of Semantic Representational and Ontologi-
limited. The novelty of this research is that it cal Models based on the theoretical prin-
adapts cognitive models from HCI to develop ciples of COMMAA that allow cognitive
a heuristic modelling framework for COgnitive processing of an agents action using state
Modelling of Multi-Agent Actions (COMMAA) of the art Semantic Web technologies
in an attempt to provide an architecture that im- • Describe heuristic reasoning mechanisms
proves the flexibility of Multi-agent interaction that can be derived from cognitive models
by promoting cognitive awareness. The highlight to enhance the cognition of Semantic Web
of the framework is that it identifies architectural agents
and knowledge-based requirements for agents to • Analyze and discuss the impact of using
structure ontological models for cognitive pro- COMMAA to model multi-agent collabora-
filing in order to increase cognitive awareness tive applications on the Semantic Web
between themselves, which in turn promotes
flexibility, reusability and predictability of agent
behaviour. The ultimate aim is towards applica- Background
tions which advocate user-centeredness such that
as little cognitive overload is incurred on humans. The Semantic Web vision of Berners Lee (2001)
The Semantic Web is used as an action mediating has enabled the Web applications to move from a
space, where shared knowledge base in the form purely human user community towards a mixed
of ontological models provides affordances for user community consisting of humans as well as of
improving cognitive awareness. software agents. This imposes certain challenges

572
Enabling Distributed Cognitive Collaborations on the Semantic Web

and brings new requirements towards models for can only interact with agents they have been de-
modelling Semantic Web-based systems (Scott et signed to interact with. One reason is that agents
al. 2005; Klein et al. 2004; Neuhold 2003). The are implemented using mechanisms such that
foremost issue is that of adaptive collaborative they conform to only a limited set of interaction
coordination and cooperation for utilizing services protocols generally resulting in inflexible or rigid
and Web information and imposes challenges agents. It is reported in research and learnt from
from the perspective of interaction as well as previous experience (Ahmad et al. 2005, Shafiq et
interoperability amongst both agents and humans al. 2005, Tariq et al. 2005a, Tariq et al. 2005b) that
(Arai & Ishida 2004). As software agents become the agents show unyielding behaviour to messages
more capable and more prevalent, they must be not specified by the protocol. Efforts such as Louis
able to interact with a heterogenous collection of and Martinez (2005a;2005b) have attempted to
both humans and software agents, which can play address the issues but the focus has been largely
diverse roles in a system, with varying degrees of to provide semantic handling of messages.
autonomy, initiative, and authority across differ- It is therefore believed that there is still a long
ent tasks (Schreckenghost et al. 2002). However, way to go before true homogenisation of agent
research supporting such interaction with these communities can be achieved. This is because the
types of agents has received relatively little at- variance of agent communication and functional
tention (Martin et al. 2003a; 2003b). pragmatics introduces a certain level of mismatch
and the need of flexible and adaptive interactions
Limitations in the State of that promote interoperation becomes imperative.
Multi-Agent Interaction This is particularly essential when the agents from
diverse platforms intend to co-exist and cooperate
While there has been a significant proliferation of in mutual. In addition, most agent-based applica-
agent architectures and applications in the Seman- tions assume pre-defined knowledge of agents’
tic Web domain, there is significant separation of capabilities and/or neglect basic cognitive and
concerns from the principles that ensure flexibility interactional requirements in multi-agent collabo-
of distributed interaction between heterogenous ration. Thus the research community is faced with
agents. Recently, much effort has been expended the challenges of improving the limited visibility
on making agents interoperate in the emerging of agent’s processing ability and behaviours that
open environments and standards. The Founda- may be a result of possible mismatches between
tion for Intelligent Physical Agents (FIPA), an the agents’ mental and implementation models. In
IEEE Computer Society standards organization, addition, it is claimed that inadequate adjustability
has attempted to facilitate the interoperation and of the agent’s autonomy and a basic lack of com-
inter-working between agents across multiple, het- patibility between the required capabilities and
erogenous agent systems (FIPA, 2007). A variety those provided by an agent impose further research
of FIPA-Compliant platforms have emerged (Luck challenges (Martin et al. 2003a; 2003b).
et. al (2005) provide a review). Despite this effort,
this goal has still not yet been achieved as Louis Potential of Cognitive Models to
and Martinez (2005a) point out. In addition, the Improve the State of Multi-Agent
Agentcities European project (Willmott, 2003) Interaction
which resulted in the deployment of a worldwide
open testbed environment, underlined the lack of Some interdisciplinary research has stressed the
‘spontaneous’ exchanges between agents running potential of cognitive models studied in cognitive
in this environment. In almost all cases, agents science as substantial means of better probing

573
Enabling Distributed Cognitive Collaborations on the Semantic Web

multi-agent issues, by taking into account es- users and a part of the world, often by manipulat-
sential characteristics of cognitive agents and ing machine representations of the world (Clark,
their various capacities (Sun 2001). The term 2001). At the basis of this research is the idea to
cognitive models has traditionally been associ- view the Semantic Web as a distributed cogni-
ated with humans as cognition is essentially a tive system, a basic unit of analysis, composed
human characteristic. Cognition can be thought of human and machine agents in a work domain
of as a modelling process which creates a model that is delineated by roles, work and communi-
from which deductions can be drawn (Meredith cation norms, artefacts, and procedures (Zhang
1970). Humans are known to continually create et al, 2002).
and access internal representations of their current According to (Lu, Dong & Fotouhi 2002)
situation - referred to as their cognitive model (Saja the Semantic Web uses ontologies to describe
1985). It is said that the state of interaction with a various Web resources, hence, knowledge on
system can be greatly improved by design activi- the Web is represented in a structured, logical,
ties that account for and support the emergence and semantic way allowing agents to navigate,
of a user’s cognitive model. These models are harvest and utilise this information (Payne, et al.,
referred to as mental models in human-computer 2002a;2002b). Agents can also read and reason
interaction (Norman 1986; 1988). about published knowledge with the guidance of
Inspired by the potential of cognitive models, ontologies. Also the collection of Web-services
the research presented in this chapter investigates described by ontologies like OWL-S (Ankolekar
the possibility of meeting the above mentioned et al. 2001; Martin et al. 2005;2007) will facili-
challenges by bringing cognitive models and theo- tate dynamic matchmaking among heterogenous
ries in the Semantic Web to achieve more robust agents: service provider agents can advertise their
and effective architectures for agents that facilitate capabilities to middle agents; middle agents store
distributed collaboration– be it amongst agents these advertisements; a service requester agent
or between agents and humans or vice versa. The can ask a middle agent whether it knows of some
premise of the research is based on the hypothesis provider agents with desired capabilities; and
that if artificial (or software) agents were to be the middle agent matches the request against the
designed to emulate the interactional and cogni- stored advertisements and returns the result, a
tive properties of humans in a complementary subset of the stored advertisements (Sycara et
way, such that they interact with each other and al. 2002;1999).
their environment in the manner that humans Based on the analysis of literature and state
do, it would increase their functional capabil- of the art, agent-mediated distributed computing
ity to serve humans. Additionally it would also paradigm for the Semantic Web is viewed as a
reduce the cognitive load that humans require in layered abstract architecture shown in Figure 1
distributed collaborations while viewing it from -a lens through which multi-agent collaboration
a distributed cognitive perspective. can be viewed. Applying the distributed cognitive
view of Semantic Web, the abstract layered model
revises the traditional view of the Semantic Web
Research Motivation and by adding cognitively modelled interactions.
context: Distributed The above abstract model of the Semantic Web
Cognitive view of the agent and agent characteristics requires multi-agent
mediated Semantic Web interaction which can consist of three interaction
levels: human-human, human-agent, and agent-
The view adapted for Semantic Web is that of a agent. In each interaction level, both interaction
world-mediating system as it mediates between design and interoperability are necessary for

574
Enabling Distributed Cognitive Collaborations on the Semantic Web

Figure 1. Layered Abstract Model for illustrating various levels of Cognitive Collaborations in Agent
Mediated Semantic Web

mutual accessibility and understanding among interactional gaps between heterogenous agents on
them as has also been highlighted by Arai & Ishida the Semantic Web. The resulting contribution lies
(2004). The chapter extends the vision for the need in the adoption of the AC to develop a heuristic
and requirements for modelling these interactions modelling framework called COgnitive Model of
from a distributed cognitive perspective (Basharat Multi-Agent Actions (COMMAA).
and Spinelli, 2008a). By applying the Distributed The framework is proposed to model multi-
Cognition perspective Chandrasekharan (2004) agent actions in a collaborative MAS environment
this model considers the importance of studying from a cognitive perspective. The framework
interaction and interoperation amongst multi- is intended to aid the designer in modelling the
agents not in isolation but within the environment agent behaviour and action through a cogni-
agents inhabit. The combination of the Semantic tive cycle. Using the principles of the proposed
Web inspired by cognitive model can generate framework designers can define both functional
a framework where agents and application can and non functional aspects of the design of an
better cooperate. agent’s interactive role in a collaborative scenario,
After a review of the available cognitive ap- especially focusing on the concepts of semantic
proaches that model human activities in interaction and articulatory distances, as derived from AC,
with any system (artefacts and the environment), as mismatch between agents’ goals and its func-
the Action Cycle (AC) (Norman 1986; 1992) has tional capabilities.
been selected as the most promising approach for Not only this research proposes the theoretical
this research. The aim is to unfold the potential guidelines for cognitive modelling of agents, it also
of the AC in an attempt to identify the design and provides design illustration for the architectural

575
Enabling Distributed Cognitive Collaborations on the Semantic Web

elements necessary for realizing these principles Actions (COMMAA) is shown in the Figure 2. The
in the Semantic Web context. Thus the framework primary aim of the framework is to cognitively
identifies architectural and knowledge-based model the agent’s action in a collaborative MAS
requirements for agents to structure ontologi- environment situated in the Semantic Web such
cal models for cognitive profiling in order to that the limitations in the state of Multi-Agent
increase cognitive awareness amongst agents. Interaction can be overcome.
By cognitive awareness, a term coined within The conceptual constituents of the framework
this research, it is intended the ability of the Web of COMMAA include:
agents to diagnose their processing limitations
and to establish interactions with the external • Cognitive agent Action Cycle (CogAC):
environment (in the form of other agents including COMMAA is based on Cognitive agent
humans and software agents) using the principles Action Cycle (CogAC) which serves as the
derived from the framework for COMMAA. This fundamental core of the framework, and
is with the aim to support users’ goals in a more is designed to aid the designer elaborate
direct manner by providing agents that can share, the agent’s functional behaviour using two
discover and access each other’s capabilities in stages namely Execution and Evaluation,
a collaborative manner and are able to function each with its respective steps. The CogAC
dynamically and adaptively without continuous views agent as the primary entity that inter-
human intervention. acts and functions in a MAS environment.
This brings about a more effective MAS en- The stages of an agent interacting with a
vironment; where agents may delegate each other MAS environment are described such that
tasks and goals based on each other’s awareness in order to accomplish a goal, which is in
of abilities, behaviours and affordances. The ulti- turn delegated to it by a human user, the
mate aim is towards applications which advocate following steps are traversed by an agent:
user-centeredness such that as little cognitive Goal Formation, Intention formation, Ac-
overload is incurred on humans. The strength tion specification, Execution, Perception,
of this framework lies in its robust theoretical Interpretation and Evaluation.
foundation that has found validation in a devel- • Cognitive Distance Model (CogDM): The
opmental infrastructure that helps realise the further elaboration of the steps of CogAC
theoretical principles. leads to the formulation of agent’s semantic
and articulatory disposition in each stage
of execution and evaluation, described and
theoretical framework and illustrated by the Cognitive Agent Distance
conceptual constituents for Model CogADM. These dispositions also
COgnitive Model of help to identify the agent’s Gulf of Execution
Multi-Agent Actions (COMMAA) and Evaluation. These two are discussed in
the subsections to follow.
The framework developed in this work and • Cognitive Agent Design Principles: The
presented below is based on these fundamen- principles are derived using the Action
tal principles of Agents’ action, borrowed and Cycle mapped for agents. The principles
modified from the Human Action Cycle (Norman may serve as heuristic for evaluating the
1986) and its elaboration for Direct Manipula- design of agent-based applications. They
tion Interfaces by Hutchins et al. (1986). The help to identify the significant mismatches,
framework of COgnitive Model of Multi-Agent constraints and affordances. The framework

576
Enabling Distributed Cognitive Collaborations on the Semantic Web

Figure 2. Conceptual design of Theoretical Framework / Conceptual Constructs of COMMAA – Cogni-


tive Model of Multi-Agent Action

defines CogADPs only at abstract level. identified, it helps in identifying the archi-
These serve as blueprints which may be tectural needs and components required for
specialised to a domain specific context successful completion of agent action cycle.
by the designer to derive a context specific As these are identified by the designer, they
cognitive profile for an agent. These may be may be cross checked against the environ-
specialised according to the domain knowl- ment that the agent in being built using. In
edge and the specific contexts of application other words, it may also be said that provision
the agent may be operating in. The design of these capabilities ensure that the agent’s
principles aid in Cognitive Mismatch (Dis- Cognitive Design Principles will be met to
tance) Analysis, that allow at design time to some extent. The availability of these ele-
be made known the possible stages where ments would ensure that the agent is Cog-
distance of execution or evaluation may nitively Directed, or Cognitive Directness
occur. The designer by analyzing whether is exhibited in agent’s action with respect
provision for these principles in made in the to its interaction with the environment. The
agent’s infrastructure can help develop the Architectural elements identified ensure the
cognitive profile of the agents, alternatively minimal design requirements that must be
agents may have the dynamic capability of met in order to bridge the agent’s gulfs of
identifying these distances at runtime and execution and evaluation.
may change, update their profile dynami- • Agent Knowledge Elements: The knowl-
cally. edge requirements are identified by the
• Agent Architectural Elements: As each designer such that at each stage of agent’s
sub-stage of agent’s execution and evaluation execution and evaluation, these are the ele-
stages are elaborated, and as distances are ments that the agent must possess or must

577
Enabling Distributed Cognitive Collaborations on the Semantic Web

be provided with in order to achieve the suc- the two stages namely Execution and Evaluation
cessful realisation of the respective stage. in order to successfully achieve the goals of the
interaction respectively.
Cognitive Agent Action Cycle The stages of an agent interacting with a MAS
(CogAC) environment are described such that in order to
accomplish a goal, which is in turn delegated to it
Agents are designed to continually act upon and by a human user, the following steps are traversed
monitoring the MAS Environment, interacting by an agent: Goal Formation, Intention formation,
with it and collaborating with other agents and Action specification, Execution, Perception, Inter-
entities, evaluating its state, and executing actions. pretation and Evaluation (as shown in Figure 3.).
The system is a closed loop: when agents act, it is The essential concepts are the Gulfs of Evaluation
usually done so in response to some prior evalua- and Execution, each arising as a result of semantic
tion of its perceptions or as a result of some goal and articulatory distances (cognitive distances in
delegated to it by a human. After an agent acts, general). The relevance of these concepts to the
it evaluates the impact of the executed act, often agent domain is formally described in the next
modifying the action as it carries it out. In MAS sections in the form of Cognitive Agent Distance
environment the fundamental unit of agent’s social Model, which are detailed further.
ability is its interaction with other agents using The inter-relationship between the CogAC, its
messages in Agent Communication Language stages of execution and evaluation, the semantic
(ACL). Agent’s Interaction is modelled to have and articulatory dispositions and other elements
two stages as shown in Figure 3. The Interaction of COMMAA is schematically shown in Figure 3.
will have some Goal i.e. the objective that needs In Table 1 and Table 2 these stages are elaborated.
to be achieved using the interaction. In order to In addition, a generic view of the corresponding
achieve this goal, agent will need to go through CADPs, Knowledge Requirements and archi-

Figure 3. High-level conceptual design of CogAC in relation to other constructs of COMMAA

578
Enabling Distributed Cognitive Collaborations on the Semantic Web

tectural needs are also identified. Together with agent needs and the manner in which this is rep-
these main components, the three additional com- resented in its environment, as shown in Figure
ponents are identified to help the designer model 4. This manner of representation also includes
agents’ behaviour in a more robust manner. Cogni- the mechanisms with which the knowledge is
tive Agent Design Principles, Agent Architectural accessed and reasoned about.
Elements and Agent Knowledge Elements are all The prospective relevance of the cognitive
identified, as each stage of the agent’s execution distances is highly relevant to Communication,
and evaluation are elaborated upon. collaboration and interactions taking place be-
tween agent application residing on the Semantic
Cognitive Agent Distance Model Web - since the basis of interaction is the commu-
(CogADM) nication language and its vocabulary represented
as knowledge on the Semantic Web.
On the Semantic Web, Knowledge is invisible Cognitive Distance in this context is taken
and intangible. While meanings are essential to to be a measure of the gulfs of execution and
knowledge, they cannot get across to an agent evaluation
without some kind of representational form.
Knowledge representation has two aspects: the —the conceptual gap, or mismatch between the
meaning of the information, named semantics, and agent’s goals and intentions, and the way in
the physical form or appearance, named syntax. which they are in cohesion to, or represented
When an agent interacts with a knowledge rep- by, the multi-agent system environment. A large
resentation, it interacts with both the semantics distance is representative of a large gulf in the
and the syntax. execution or evaluation stages, signifying that a
There is often, however, a gap, known as lot of cognitive load is incurred in translating the
Cognitive Distance, between the knowledge an

Table 1. Steps of CogAC in Execution Stage (Agent’s Gulf of Execution) cross referenced with design
principles, Knowledge elements and Architectural requirements

579
Enabling Distributed Cognitive Collaborations on the Semantic Web

Table 2. Steps of CogAC in Evaluation Stage (Agent’s Gulf of Evaluation) cross referenced with design
principles, Knowledge elements and Architectural requirements

agent’s intentions into the system’s representations, to interpret or evaluate the system’s output or
or vice versa. response as a result of some interaction with it.

That is, a large distance of execution means it Directness of Agent’s Interaction with
is relatively difficult or not possible for the agents MAS Environment
to express their query or desires to the system, and
a large distance of evaluation indicates mismatch Using the directness of interactions as usability
of some form within the agent’s infrastructure measures would involve understanding the agent’s

Figure 4. Cognitive Knowledge Gap between agent and multi-agent system environment

580
Enabling Distributed Cognitive Collaborations on the Semantic Web

problem solving strategies, approaches and inten- gulfs of execution and evaluation must be detailed
tions as now well the MAS environment supports as below:
the agent’s functional needs. A design and evalua-
tion methodology that assesses directness requires Agent’s Gulfs of Execution: The gulf of execu-
cognitive basis because understanding the user’s tion arises as a result of cognitive distances of
mental processes is key to assessment. The con- execution between the agent and its interaction
cept of directness as suggested by (Hutchins et.al, with the environment (multi-agent, open). The
1986), and later by (Cuomo 1993) is adapted here gulf can be identified by elaborating the cognitive
to refer to the degree of capacity of an agent to distances of execution – which result due to the
bridge the Gulfs of Execution and Evaluation. difference or mismatch between the intentions
The concept of multi-agent interaction is virtu- and the allowable, available actions, capabilities
ally an unexplored area in terms of operationally and affordances in its environment. One indicator
defining and assessing the directness of interac- representing this gulf is to determine how well the
tions an agent engages in with its MAS environ- agent is able to do the intended actions directly,
ment to a degree that can be applied in practice without extra effort: is the agent able to fulfil its
and measured using qualitative or quantitative goals delegated to it by humans? Do the actions
tools. Directness of an Agent’s Interaction with its provided by the agent match those intended by the
environment may be used as a qualitative indicator person? Does the environment/infrastructure (in-
of the amount of cognitive processing needed to ternal infrastructure and/or external environment)
carry out a successful interaction. Directness is provide affordances that allow for the intentions
inversely proportional to the amount of cognitive of the agent? If there is a limitation, then there is a
processing it takes to manipulate and evaluate gulf of execution in the state of agent’s interaction
the results of agents’ interaction with its MAS which must be bridged, either with collaborative
environment. Moreover the cognitive processing effort with other agents in the environment or
required for an agent is a direct result of the gulfs eventually by the human, who would incur cogni-
of execution and evaluation the agent has to deal tive overload or processing. The ultimate aim is
with. The better the agent’s architecture, the less to design agents that are able to readily identify
cognitive processing needed and the more direct and bridge/overcome such a gulf to efficiently
the resulting interaction between agents. achieve the goals set up for them.
Thus Distances are complementary to Direct-
ness; the lower the distances, the greater the direct- Agent’s Gulfs of Evaluation: The gulf of evalu-
ness, and vice-versa. The two terms distance and ation arises as a result of cognitive distances of
directness may be alternatively used, depending evaluation between the agent and its interaction
on the nature of mismatch or the extent of gulf in with the environment (multi-agent, open); the
the stages of agent action. It would be subsequently possible mismatches between agents reasoning
shown in the later sections how these concepts capabilities and its representation mechanisms.
are encoded into the profile of agents and how The Gulf of Evaluation reflects the amount of effort
the concepts are important to derive heuristic that the agent must exert to interpret the state of
mechanisms for agents’ reasoning. its interaction with the system and to determine
how well the expectations and intentions have
Applying the Gulfs of Execution and been met. The gulf is smaller when the system
Evaluation in Multi-Agent Environment provides information about its state in a form that
is easy to get, is easy to interpret and matches the
In order to identify gaps that separate agents manner in which agents’ affordances allows these
mental states from execution ones, the agent’s to be perceived.

581
Enabling Distributed Cognitive Collaborations on the Semantic Web

Bridging the Gulfs and the Cognitive The term affordance is adapted for COMMAA
Distances to refer to the agent’s capacity of action. At each
stage of the agent action cycle the agent’s corre-
Directness is inversely proportional to the amount sponding affordance is determined and accounted
of cognitive effort/processing it takes to manipu- for. Any constraint in any of the agent’s capacities,
late and evaluate the state of interaction with MAS being the ability to achieve the goal, to formulate
Environment and, moreover, the cognitive effort the intention or to interpret the consequences of
is a direct result of the gulfs of execution and its action, may result in a cognitive distance being
evaluation. The better the architecture and the introduced. Usually the term affordance is linked
environment of the agent helps the agent bridge to a machine or an application (Norman 1992).
the gulfs, the less cognitive processing needed In the context of this research affordance is the
and the more directed the resulting interaction. ability of one agent (the sender) to behave in a way
For this, the architecture should facilitate some that the other agents in the MAS environment(the
means of identify, through some indicators, the receiver) can understand, such that they both have
possible distances and also identify the capabili- a shared mental model and can trigger, comple-
ties and affordances available and not-available ment or facilitate each other’s action/behaviour;
corresponding to these indicators. therefore these affordances need to be ensured.
The affordances are provided by the cognitive
Agents’ Capabilities and Affordances artefacts that form part of the agent’s internal
infrastructure/architecture and its external en-
The concept of affordance is here explored and vironment.
exploited. It is believed that an affordance inspired Affordances are the opposite concept to dis-
agent architecture and environment will help to tances, so they are complimentary. For instance,
bridge the gulfs of execution and evaluation. By an agent’s affordance for intention formation will
interfacing perception and action in terms of ca- ensure there is no semantic distance of execution
pabilities and affordances for agents, the aim is and a semantic distance of execution will mean
to provide a new way for reasoning about agent’s that agent has no affordance for forming an in-
capacities and bring about cognitive awareness tention. The psychologist Gibson was the first to
amongst the agents about each others capacities frame affordances as unified relations between the
and constraints, when interacting in a collaborative environment and an actor (Gibson, 1979, p. 127).
environment. In Cognitive Science, an affordance Affordances can be explained as action possibili-
is a resource or support that the environment ties that actors have in the environment. That is,
offers an agent for action, and that the agent can an affordance exists relative to (1) properties of the
directly perceive and employ (Gibson, 1979). environment and (2) the action capabilities of an
Although, this concept has only rarely been used actor (McGrenere and Ho, 2000). For example, a
in Semantic Web agent architectures, it offers chair has the affordance of ‘sitting’, because of its
an original perspective on coupling perception, shape, height and carrying capacity and because
action and reasoning, differing notably from of the humans’ ability to sit, the length of their
standard reactive and hybrid architectures. Taking legs, and their weight. The concept of affordances
it literally as a means or a metaphor for coupling is of particular interest in the field of HCI, which
perception and action directly, the potential that primarily concerned is studying how properties of
affordances offer for designing new powerful and computers (the environment) and humans (actors)
intuitive agent-based Semantic Web architectures influence their interaction with each other.
is obvious (Vugt et al. 2006).

582
Enabling Distributed Cognitive Collaborations on the Semantic Web

Extending the affordance concept to the Se- agents (including humans if the interaction
mantic Web, it is believed that by semantically is being carried out with a human) and, (2)
identifying and encoding the affordances of an can be used to easily accomplish the agent’s
agent will help achieve cognitive directness of goals
interactions amongst the agents. This will contrib- • Articulatory distance: The degree to which
ute to making agents more interoperable in open the form of communication between an agent
and heterogenous environments. Some consid- and its environment reflects the application
erations that must be taken into account include objects and tasks involved
the relationship of goals with respect to agents.
Goals are central in affordance evaluations. It is Articulatory distance concerns the actual
important to understand that an affordance does form that communication takes between say two
not change as the needs and goals of the person agents; for example, the choice of Message Encod-
change (McGrenere & Ho, 2000 interpreting Gib- ing, Message Content, Content Type, Interaction
son, 1979). Similarly, for an agent, an affordance Protocol used/employed for the communication.
must be identified irrespective of what the agent’s A small articulatory distance results when the
eventual goals are. E.g. ff an agent affords Mes- input techniques and output representations used
sage translation from FIPA-ACL into FIPA-SL, are well suited to conveying the required infor-
it is independent of whether another agent will mation. Articulatory distance is also decreased
eventually participate in such interaction where when the form of inputs and outputs relate to
this translation is required. However, agent’s ac- the semantic concepts used of the underlying
tions do depend on the goal context. Agents will conceptual model.
typically need to act within the environment (they Semantic distance involves the capability of
use an affordance) because of a goal they want to agents required to express desired actions within
achieve for example, performing a task (Vugt et the concepts of the system. It deals with the pos-
al. 2006, Kakazu & Hakura 1996). sibility to express the concepts of interest concisely
using the available capabilities. Semantic distance
Semantic and Articulatory Distances also involves a measure of how closely the agent’s
in Multi-Agent Interaction Model conception of the task domain matches that of the
environment. The two distances that compose the
Following is a detailed account of how may the Cognitive Distance in Agent’s Behaviour are the
components of cognitive distance namely: the Semantic Distance and Articulatory Distance.
Semantic and the Articulatory distances occur in These can be considered as subsets of each of
the execution and the evaluation cycles. It is also the gulfs, but in reference to input behaviour (ini-
important to illustrate how the cognitive distances tiator’s behaviour in communication/interaction)
are identified with the help of agents’ capabilities and output behaviour (Respondent’s behaviour in
and affordances. A conceptual high-level view is communication/Interaction). These are illustrated
shown in Figure 5. Identifying these distances, and discussed in depth below.
along with agents’ capabilities and affordances is
of primary importance in highlighting the possible An Illustration of Semantic Distance of
gulfs of evaluation and execution. Execution

• Semantic distance: The degree to which Semantic Distance relates to the relationship
the semantic concepts used by the agent between an agents intentions and the meaning
are (1) compatible with those of the other of expressions, required to convey the agents’

583
Enabling Distributed Cognitive Collaborations on the Semantic Web

Figure 5. Components of cognitive distances in multi-agent collaboration

intention such that its meaning is interpretable An Illustration of Articulatory Distance


by the intended recipients in the environment. of Execution
Semantic distance is related to the ‘nouns’ and
‘verbs’ or ‘objects’ and ‘actions’ provided by an Whereas semantic distance relates to relation-
agent’s infrastructure and its environment. For ships between Agents’ formulated intentions
execution, forming an intention is the activity that and the meanings of expressions/functional
spans semantic distance. The intention specifies capabilities available, articulatory distance in an
the meaning of the input expression that is to agent context is defined to relate to relationships
satisfy or reach the’ agents goal or sub-goal. between the meanings of expressions/capability
If semantic indirectness of execution existed, and the agent has the affordance to realise the
agents would not be able to express their inten- capability in action, or whether agent affords
tions directly, or at all in order to achieve the the actions that are essential in achieving the
goal delegated to them. Lack of capabilities or desired intentions. A mismatch of such a form
affordances would be indicators of this condition. will cause an articulatory distance of execution
Agent is programmed at lower level of functional- to exist. The articulatory distance of execution
ity then desired by the human user. The agent or is illustrated in more detail in Figure 7.
the human may need to carry out more actions
then would be expected to accomplish the same An Illustration of Semantic Distance of
goal or intention. E.g. Agent may need to col- Evaluation
laborate or request another agent to achieve the
goal on its behalf. This requires agent to have Semantic distance also occurs on the evaluation
desired affordances and capabilities defined. The side of the interaction cycle. Semantic distance
semantic distance of execution is illustrated in of evaluation is proportional to the amount of
more detail in Figure 6. processing required by the agent to determine
whether the goal has been achieved. The semantic

584
Enabling Distributed Cognitive Collaborations on the Semantic Web

Figure 6. An illustration of agent’s semantic distance of execution

Figure 7. An illustration of agent’s articulatory distance of execution

585
Enabling Distributed Cognitive Collaborations on the Semantic Web

Figure 8. An illustration of agent’s semantic distance of evaluation

distance of evaluation is illustrated in more important implications in the current Semantic


detail in Figure 8. Web architectures. Traditional Web-service agent
architectures only allow agents to discover about
An Illustration of Articulatory Distance each others services. However this research claims
of Evaluation that an architecture that is inspired from cogni-
tive models would allow the agents to develop a
Articulatory distance or indirectness of evalu- cognitive awareness about each other which could
ation would be indicated by errors in interpreta- bring to a more effective MAS environment. To
tion and having to take extra actions to correctly validate this claim, a lower-level classification of
interpret the state of communication or the result design goals which provide the basis upon which
of interaction. An agent’s inherent articulateness the Cognitive Profiling Architecture is devised
is closely tied to its level of technology. The upon includes: (a) Enhanced Negotiation and
articulatory distance of evaluation is illustrated Collaboration based on Cognitive Awareness (b)
in more detail in Figure 9. Flexibility and Reusability (c)Adaptive Interac-
tion and Interoperability (d)Discovery based
on Heuristic reasoning and (e) Minimization of
Design and Implementation of cognitive load on humans.
COMMAA Inspired Agent In an attempt to realise the above design goals,
Architecture which are direct implications of COMMAA,
some important considerations are taken into
Design Goals for Cognitive Profiling account. Firstly, some mechanism is needed that
of Agents enables the agents to discover and find out about
each others cognitive distances, semantic and
The theoretical foundations of Cognitive Mod- articulatory dispositions, capabilities, affordances
els such as the Action Cycle (AC) bring about and constraints. Secondly, dynamic and built in

586
Enabling Distributed Cognitive Collaborations on the Semantic Web

Figure 9. An illustration of agent’s articulatory distance of evaluation

mechanisms are needed for heuristic reason- existence is acknowledged by a particular theory
ing and invoking/developing learning, adaptive or system (Honderich, 1995 cited in Bell et al.
measures for these constraints and distances to 2007). Such ‘things’ include both types (such as
be bridged. These considerations are considered the class of Agents) and individual elements (such
rudimentary to enable agents to be aware of as the agent TravelAgent). The adoption of such a
these limitations and Gulfs that may limit their definition is important because, when compared
functionality or the extent of services they can with more computationally orientated defini-
provide. As minimal architectural consequences, tions of ontology (for example, (Gruber 1993);
the elements necessary in COMMAA inspired p.1) states that “an ontology is a specification of
semantic-Web agent architecture are implemented a conceptualisation”), there is an explicit refer-
in a cognitive profiling architecture described in ence to a system’s ontic commitment (i.e., things
Basharat and Spinelli, (2008b.). The Cognitive whose existence is acknowledged or recognised).
Profile of Agents is detailed next. This leads to representations that are more closely
mapped to real world objects.
A Semantic Representation Model The use of ontology is also prospective in the
for Cognitive Profile of Agents Semantic Web action space, since the emerging
standards enable reasoning to be carried out
A possible way of implementing the cognitive effectively on such models. An ontology based
profile is by associating an ontology with an agent reasoner could be invoked on the ontology to
i.e. agent is given knowledge about its constraints carryout reasoning using the heuristic reason-
and affordances, its semantic and articulatory ing rules defined; agent could then be made to
disposition. At the heart of the framework is the reason about its state of processing according to
adoption of ontology to drive the cognitive pro- CogAC. Architectures that facilitate the sharing
file of agents. From a philosophical perspective of ontologies would enable the agents to discover
ontology can be defined as a set of things whose each other cognitive dispositions thus improv-

587
Enabling Distributed Cognitive Collaborations on the Semantic Web

Figure 10. Conceptual Model of Ontological Knowledge Model for Cognitive Profile of Agents

ing the manner in which they interact. The issue Cognitive Profile of Agent
at stake is to be able to represent the cognitive
profile adequately and in a manner that can be The framework’s correct implementation calls for
shared among agents. The Cognitive profile is maintaining a cognitive profile of Semantic Web
therefore implemented as an Ontology in OWL, agent. Following are the elements to be maintained
with OWL-DL (Description Logics) as basis in the cognitive profile:
of representation of the profile parameters and
properties. Being an emergent standard, OWL- • Agents Cognitive Mental States
DL ensures that the model caters for a more open • Capabilities
community. Ontologies have been recognised by • Affordances
the research community as a – model of expressing • Semantic Disposition (Semantic Distances
the knowledge model for agents e.g. by the recent of Evaluation and Execution)
research of (Laclavík et al. 2006). • Articulatory Disposition (Articulatory Dis-
tances of Evaluation and Execution)
Conceptual Model for Agents’ Cognitive • Goals, Intentions, Perceptions, Evalua-
Profile Ontological Model tions

Figure 10 shows a generic conceptual model, Cognitive Mental State of an Agent


with objects and properties (shown by labels on
associations between concepts) of the ontologi- An agent modelled along the lines of the CogAC
cal model to be implemented in order to enable requires various cognitive mental states corre-
shared cognitive profiling of agents. sponding to the various steps in the execution and

588
Enabling Distributed Cognitive Collaborations on the Semantic Web

Figure 11. Conceptual Flow of Activities to develop an application according to the guidelines of COM-
MAA

evaluation stages of the action cycle. These mental Abstract Heuristic Reasoning
states of the agent facilitate the agent’s behaviour Mechanism for Agents
in both its execution and evaluation cycles. It
provides provision for state modelling, represen- To bridge the gulfs of execution and evaluation,
tation and tracking an agent’s state of execution. agent must have some heuristic reasoning mecha-
In addition representing these states agents may nism built into its architecture, such that given a
have the dynamic capability of identifying these shared cognitive profile is available, it should be
distances at runtime and may change and update able to reason on the knowledge present in it to aid
their profile dynamically. The conceptual model the agent’s processing and help bridge the gulfs
is developed using the Protégé Ontology editor of execution and evaluation through collaboration
to generate OWL Ontology. or other means. The important considerations
are with regards to the representation, discovery
Modelling Agents Using COMMAA and reasoning of the cognitive distances. Rules
generate advice by defining the combination of
In order to utilise COMMAA in practice to agent knowledge, action stages, distances, abili-
process and analyse the interactions and tasks ties/capabilities, and affordances, typically with
of an agent, the process shown in Figure 11 is the generic format shown in Figure 12.
used during the design and development process
of an agent-based application. The figure also Heuristic Reasoning Mechanism for
summarises the components and principles that Agents Execution Stage
contribute to the agents state modelling, based
on the principles of COMMAA. The process is The encoding of agents cognitive distances re-
described in detail in Table 3. quires a reasoning mechanism. The reasoning is
carried out based on the Semantic and Articulatory

589
Enabling Distributed Cognitive Collaborations on the Semantic Web

Table 3 . Description of Activities in processing agents’ action through COMMAA with input and output
artefacts
Activities Description Input Artefact Output Artefacts
Cognitive  Model Agent’s Action using the  Agents Functional  Knowledge
Modeling of CogAC Requirements Requirements
Agent Behaviour  The agent’s behaviour is  Agents Knowledge and
programmed according to the stages Representational
described in the CogAC Constraints
 Elaborate Agent’s Action with the  Cognitive Agent Design
help of Seven Stages of CogAC Principles. The CADPs
 Identify agent’s knowledge are used as heuristics to
requirements and elements evaluate the design at
 As the knowledge elements needed each stage
by the agent are identified, it is
ensured that the architecture of the
agent provides for these knowledge
elements to ensure a successful
completion of the CogAC
 Program agent’s behaviour
according to available architecture

Scoping the  Identify Agents capabilities  Knowledge Requirements  Individual Agent


Agent’s Cognitive capacities and affordances  Agents Functional Cognitive Profile
Profile  Identify agents cognitive distances Requirements (Ontic Commitment
 Identify known possible agent’s gulf  Agents Knowledge and Model)
of execution and evaluation Representational
 Each of the components of the Constraints
Agent Profile are interpreted such
that they represent its Ontic
Committment

Encoding of  Encode Agent’s Cognitive Profile  Profile Parameters  Cognitive Profile


Cognitive Profile parameters in the ontology  Agents Semantic  (Ontic Commitment
 The Disposition of Agent (Semantic Disposition Model)
and Articulatory) in both stages of  Agent’s Articulatory
execution and evaluation is encoded Disposition
at design time

Sharing the  Registering or Publishing the profile  Cognitive Profile  Shared Cognitive
Cognitive Profile Profile Repository
populated with
Cognitive Profile of
Agent

disposition of agents encoded in their Cognitive Heuristic Reasoning Mechanism for


Profile. A reasoning mechanism for the execu- Agent Evaluation Stage
tion stage in the form of pseudo-code is given in
Figure 13. This is generic given that the action A reasoning mechanism for the evaluation stage in
cycle is applied in a generic context. It may be the form of pseudo-code is given in Figure 14.
specialised according to the agent application
being developed.

590
Enabling Distributed Cognitive Collaborations on the Semantic Web

Figure 12. Generic Format for Rule-based reasoning of agents’ cognitive mental states

Figure 13. Abstract Mechanism for Heuristic Reasoning of the Cognitive Profile Model (Execution
Stages)

Demonstration of Agents’ framework. The purpose of this distributed col-


Enhanced laborative multi-agent application is as follows:
cognitive Capabilities
• To show how the framework is applied to
In order to give a flavour of how the framework design of MAS based applications operable
presented above enhances the agents’ cognitive on the Semantic Web
capabilities by imparting improved cognition, the • To show how application of COMMAA helps
framework was applied to a simulated multi-agent build the cognitive profile of the Agent
based distributed collaborative application with • Show how sharing the cognitive Profile im-
the aim of testing, improving and evaluating the proves the collaboration between Agents

591
Enabling Distributed Cognitive Collaborations on the Semantic Web

Figure 14. Abstract Mechanism for Heuristic Reasoning of the Cognitive Profile Model (Evaluation-
Sub Stages)

Figure 15. Cognitively Modelled Agent-based Travel Planning Scenario Modelled according to COM-
MAA

592
Enabling Distributed Cognitive Collaborations on the Semantic Web

High-Level Architecture of Cognitive Profiles of Agents in TPA


Distributed Collaborative
Multi-Agent Application The cognitive profile ontology is central to repre-
senting the knowledge for agents. The elements of
Travel Planning Scenario the cognitive profile ontology serve to represent the
shared knowledge base of agents through which
The high-level architecture of the travel planning agents’ cognitive awareness will be enhanced.
scenario developed to demonstrate the enhanced The cognitive profiles are designed to simulate
cognitive abilities of agents is shown in Figure an environment such that some agents are limited
15) is a customised adaptation from the vision in certain capabilities, while others are equipped
of travel planning agents presented by Hendler with them in a complementary manner to facilitate
(1999). interoperability, resuse and adaptive collaboration
The top level functional goals of the demonstra- based on enhanced cognitive awareness. The
tion application are as follows: The Multi-Agent cognitive profiles of the agents in the prototype
based application is aimed to use cognitively application are shown in Table 5.
modelled agents to solve travel problems given
by a user. The user can propose to the user Agent Reuse Mechanisms Employed
his desired travel, and it will obtain a complete
plan that inclu des information about transport, The Behaviour API and Interaction Protocol API
lodging, etc. The agents will Extract, filter and of JADE (Bellifemine et. Al, 2001; 1999; JADE
store information automatically from the Semantic 2004) are used to model the Agent’s Actions, or
Web using other agents. The system aims to use the an action Plan. ACL Messages are used as repre-
same information that the user could find if he wish sentations for Intentions and perception. An agent
planning the travel himself. Cognitive Sharing of is simulated in such a way that it is given runtime
different kinds of abilities is to be demonstrated capability to change its behaviour, and dynami-
to gain efficiency in the problem solving task. cally change profile so as to demonstrate the power
The agents are simulated to reuse each others and potential of the Ontological Model.
capacities, behaviours and offer affordances to
each other. Agents closely work according to the Cognitive Modelling of Agent
user’s characteristics, and functions based on the Interactions and Communication
ultimate goals obtained from the user profile and Scenarios Through CogAC
adapt their functional behaviour according to the
learned user preferences. Table 6 shows an example of how the agents col-
laboration is modelled through the CogAC.
Roles Defined for Agents
Illustration of Improved Cognitive
Table 4 shows the Roles and Responsibilities de- Awareness
fined for Agents involved in Travel Planning Col-
laboration Scenario. The Collaborative scenario The provision of cognitive profile as shared
aims integrate the abilities of a set of heterogenous knowledge base serves as means to increase the
agents. The system is made by a set of agents that cognitive awareness for agents since they can not
can communicate and cooperate among them to only reason about their own cognitive distances,
reach the problem solution. All the agents in the they can also access and query other agents’ cog-
application use FIPA Based Agent Communica-
tion Language for standardization purposes.

593
Enabling Distributed Cognitive Collaborations on the Semantic Web

Table 4. Roles and Responsibilities defined for Agents involved in Travel Planning Collaboration Sce-
nario

Table 5. Capabilities and affordances defined for agents in the Travel Planner Application

594
Enabling Distributed Cognitive Collaborations on the Semantic Web

Table 6. Cognitively Modelled Agent Communication Scenario for Travel Planner Agent

Figure 16. Agent Collaboration Scenario to Illustrate Improved Cognitive Awareness

595
Enabling Distributed Cognitive Collaborations on the Semantic Web

nitive profiles allowing them to adaptively refine environment. An attempt to assess the directness
their interactions in attempt to achieve their goals of engagement using the distances in the multi-
in a collaborative manner. An illustration of how agent interaction scenario offered useful results.
this proves so is shown in Figure 16. As shown by the demonstration that the Cognitive
Modelling of agents give a powerful boost and
Cognitive Profile as an Affordance shows great potential in the manner that agent
utilise each other’s capabilities to support each
The cognitive profile in the form of ontology serves other functional execution and in return facilitate
as an affordance for the agent to be cognitively and reduce cognitive overload of humans. The
aware of its environment and make adaptive deci- increased cognitive awareness promotes interop-
sions about it. Following is a scenario in Figure eration amongst agents and results in behaviour
17 that illustrates how this proves so. sharing and reuse. The high visibility into each
other’s mental models increases the mapping of
agent’s functional and execution models. The 7
Evaluation and Future stages of CogAC ensure that enough feedback
Trends is received to ensure successful fulfilment of
goals through multi-agent collaboration. The
The application of the COMMAA to the appli- Cognitive Profile Ontological Model serves as
cation was found very useful in structuring the an essential artefact for agents which, continually
agents’ interaction and collaborations in an MAS

Figure 17. Multi-Agent Collaboration Scenario to illustrate the use of Cognitive Profile as an affor-
dance

*Further implementation details are beyond the scope of this chapter

596
Enabling Distributed Cognitive Collaborations on the Semantic Web

provides for both sides of the CogAC: execution used for classifying user roles and capabilities
and evaluation. and for maintaining and sharing user profiles and
Based on the above evaluation of the applica- roles. The proof-of-concept application has also
tion, The CogAC can come to be considered as demonstrated the feasibility of implementing the
a fundamental part of the functioning of agent’s constructs of COMMAA using the combination
actions in interaction with its MAS environment. of Multi-Agent Platform and the Semantic Web
Thus agent modelled along lines of COMMAA middleware.
exhibit strong principles of user-centred design In addition, the COMMAA takes a holistic
and advocate ease of use, efficiency and reuse view of agent’s action and its processing in the
and interoperation. Although, the model is used stages of execution and evaluation. It increases
to model software agents, similar model can be the cognitive awareness amongst agents by elabo-

Table 7. Improvements achieved as a result of application of COMMAA


Design Goals Evaluation

Cognitive Awareness:  Agents profile has shown to be encoded using OWL ontology and published. Thus Agents
are able to cognitively describe, publish and access each others capabilities, affordances and
distances/constraints by Quering the ontology using SPARQL based mechanism. More dynamic
and adaptive behaviour of agents has been shown. Thus the implementation shows cognitively
aware modelling agents interaction.
 Agents can carry out cognitively aware communication and collaboration with each other
 Through Cognitive awareness, agents help each other identify and bridge the gulfs of execution
and evaluation
Enhanced Negotiation  As shown by the collaboration scenarios, agents negotiation abilities are enhanced as a result of
and Collaboration improved cognitive awareness

Flexibility and  As shown, through cognitive profiling, JADE Behaviour API and Interaction protocols have
Reusability: been used with much more flexibility and their reuse is promoted by agents sharing cognitively
their goals and abilities.
 Agents are equipped with reasoning mechanism to dynamically reason about their semantic
and articulatory disposition. This allows them to adapt to the required interaction scenario, thus
providing intrinsic support for more flexible interaction.
Adaptive Interaction  Heterogenous agents in different roles including Transport Agents, Transport and Translator
and Interoperability: Brokers, Currency Convertors etc. have been shown to participate in adaptive collaborative
scenarios through dynamic sharing of their cognitive profiles in order to help each other bridge
cognitive distances of execution and evaluation. In doing so, they help each other achieve their
goals.
 Interoperability is promoted through the sharing of cognitive profiles, dynamic publish and
access mechanisms.
 Reasoning through SWRLJess based rules enhances cognitive awareness, thus improving
interoperability.
Discovery based on  Agents dynamically reason about their profiles using SWRL Rules and DIG reasoners that
Heuristic reasoning: classify ontology. This allows dynamic discovery of cognitive distances and other agents’
profiles improving agents cognitive awareness of itself and the environment.

Minimisation of  User has been shown to be relieved of much cognitive overload since agent through all of the
cognitive load: above functionalities is able to perform much more. It is able to meet all the preferences in the
user profile. If the case was otherwise, the distribution of tasks would shift from the agent to the
user. E.g. if the agent was unable to meet the goal of converting dollar to pound, given it was
not able to find any other agent to achieve the goal, the user would have the added cognitive
overload of meeting the desired goal on its own.
 Thus human information processing has shown to be reduced as a result of increased cognitive
awareness and improvement in the flexibility and adaptability of agents’ collaborative abilities.

597
Enabling Distributed Cognitive Collaborations on the Semantic Web

rating the action infrastructure, its limitations, Ontology Learning, Alignment and Mapping:
constraints (Distances) and its capabilities and A serious issue in making this model work on a
affordances. In this way it facilitates the archi- larger scale will be ensuring the standardization
tecture for the agents that are designed to work or for the interoperability. Issues of Ontology learn-
be programmed to work more adaptively on the ing (Maedche & Staab 2001), Ontology alignment
Semantic Web. This was made sufficiently evident and mapping (as highlighted in works by Laera
using the implementation of the Travel Planning et al. 2006, Mocan, Cimpian & Kerrigan 2006;
scenario. Similarly, agents can use the semantic Sampson & Lanzenberger 2006) also become
information, share it, yet can communicate using important for standardization and homogeniza-
ACL languages, programmed according to the tion purposes. Standardization of cognitive profile
Agent principles and do not need to rely on the for agents will be another issue foreseen if this
Web service interfaces and profiles. model were to work successfully, but with rich
The results of the application are validated semantic model of OWL and RDF will allow for
against the design goals presented earlier in standardization to be achieved. However it pro-
Table 7. vides substantial stimulus for future research.
Enhanced Learning and Reasoning Mecha-
Future Directions nisms: Furthermore, an idea that will add immense
value to the further development of agents’ cogni-
The abstract architecture opens new doors of tive model is enhanced Reasoning and learning
research. With the core framework in place the mechanisms. It would also be worth investigating
natural next step in its expansion is the specialised how the principles of COMMAA plays a useful
enhancement towards a more rigorous definition of role in Interface characteristics i.e. investigation
different levels and variations of cognitive profile into role of cognitive models applied to model in-
and its parameters. Identifying the best level of terface agents and their activities and management
detail for functionally decomposing each task or of user profiles. Another issue is with respect to the
intention and applying it consistently is difficult. extent to which agent’s knowledge model or cogni-
The application highlights that in order to make tive profile is to be shared. The notion of Public,
this model fully implementable or workable in Private profile could be considered. The extent
the real world, there needs to be taxonomy of of autonomy given to the agent moving towards
distances, directness measures, capabilities and Autonomous Semantic Web services (Paolucci &
affordances defined. This study initiates this Sycara 2003) is also highly relevant.
activity by identifying the rudimentary picture
of the basic Cognitive profile of agents e.g. at
present the capabilities were identified as high Conclusion
level constructs. They can be made much more
elaborate e.g. that of OWL-S. However to make The Semantic Web community has recognised the
it reach such a state of maturity where it could be advantages of an agent-based approach to build-
utilised in practice will take some more effort. A ing deployable solutions in a number of applica-
taxonomy could be appealing because it would tion domains comprising complex, distributed
allow a generic to specific discussion of cogni- systems. The chapter targeted some of the key
tive distances across different agent applications. challenges faced when developing autonomic and
Although some subset of cognitive distances will autonomous entities in the domain of Multi-Agent
always be generic, it can be suspected that a fairly Collaborations. By applying cognitive models to
large subset will need to be specialised across model agents’ behaviour on the Semantic Web,
limited applications.

598
Enabling Distributed Cognitive Collaborations on the Semantic Web

a reasonably successful attempt has been made Proceedings of the 12th International Confer-
at coupling cognitive science principles of user- ence on Informatics Research for Development
centred design with features of agent base systems of Knowledge Society Infrastructure (ICKS’04)
and architectures. (pp. 132-139).
This research took into account the emergent
Arai, S., Murakami, Y., Sugimoto, Y., & Ishida, T.
standards in both agents and Semantic Web in
(2003). Semantic Web service architecture using
order to render the framework principles pre-
multi-agent scenario description. In J. Lee & M.
sented. The principles and design of COMMAA
Barley (Eds.), Proceedings of the 6th Pacific Rim
have also been used to demonstrate agents’
International Workshop on Multi-Agents, PRIMA
improved capabilities through an evaluation of a
2003 (pp. 98-109).
multi-agent collaborative scenario to illustrate the
adaptive coordination of different agents acting as Basharat, A., & Spinelli, G. (2008a). Towards a
owners in heterogenous and dynamically chang- distributed cognitive view of the agent-mediated
ing environments. The results of the evaluation Semantic Web. In Proceedings of the IADIS Intel-
show an improved flexibility, interoperability ligent Systems and Agents (ISA 2008).
and reusability of agents’ collective behaviours
Basharat, A., & Spinelli, G. (2008b). Towards
and goals. Thus, it establishes COMMAA as a
engineering ontologies for cognitive profiling of
step forward in providing the next generation of
agents on the Semantic Web. In Proceedings of
Semantic Web, a successful framework of multi-
the IEEE International Workshop on Engineering
agent collaboration, which is inevitably required
Semantic Agent Systems (ESAS 2008).
for generating robustly engineered agents able to
carry out spontaneous and adaptive collaboration Berners-Lee,T., Hendler, J. & Lassila, O. (2001).
based on cognitive awareness of their environment The Semantic Web. Scientific American, 279(5),
and infrastructure. 34-43.
Bell, D., De Cesare, S., Iacovelli, N., Lycett, M.,
& Merico, A. (2007). A framework for deriving
References
Semantic Web services. Information Systems
Frontiers, 9(1), 69-84.
Ahmad, H. F., Suguri, H., Ali, A., Malik, S.,
Mugal, M., Shafiq, M. O., Tariq, A., & Basharat, Bellifemine, F., Poggi, A., & Rimassa, G. (2001).
A. (2005). Scalable fault tolerant agent grooming Developing multi-agent systems with a FIPA-
environment – SAGE. In F. Dignum, V. Dignum, compliant agent framework. Software - Practice
S. Koenig, et al. (Eds.), Proceedings of the 4th and Experience, 31(2), 103-128.
International Conference on Autonomous Agents
Clark, K. (2001). The politics of schemas: Part
and Multi agent Systems, AAMAS 05 (pp. 3-4).
1. Retrieved July 20, 2007, from http://www.xml.
Ankolekar, A., Burstein, M., Hobbs, J., Lassila, O., com/pub/a/2001/01/31/politics.html
Martin, D., McIlraith, S., Narayanan, S., Paolucci,
Chandrasekharan, S. (2004). Semantic Web:
M., Payne, T., Sycara, K., & Zeng, H. (2001).
Knowledge representation and affordance. In B.
DAML-S: Semantic markup for Web services. In
Gorayska & J. L. Mey (Eds.), Cognition and tech-
Proceedings of the International Semantic Web
nology: Co-existence, convergence, co-evolution.
Working Symposium (SWWS, (pp. 411-430).
Amsterdam: John Benjamins.
Arai, S., & Ishida, T. (2004). Learning for human-
Cuomo, D. (1993). Methodology and encoding
agent collaboration on the Semantic Web. In

599
Enabling Distributed Cognitive Collaborations on the Semantic Web

scheme for evaluating the usability of graphi- Laclavík, M., Balogh, Z., Babík, M., & Hluchý,
cal, direct manipulation style interfaces. In L. (2006). Agentowl: Semantic knowledge model
Proceedings of the Human Factors and Ergo- and agent architecture. Computing and Informat-
nomics Society, Inc., Santa Monica, CA (Vol. 2) ics, 25(5), 421-439.
(pp.1137-1141).
Laera, L., Tamma, V., Euzenat, J., Bench-Capon,
FIPA. (2007).  IEEE foundation for intelligent T., & Payne, T. (2006). Reaching agreement over
physical agents. (FIPA). Retrieved July 20, 2007, ontology alignments. In Proceedings of The Se-
from http://www.fipa.org, 2006 mantic Web – ISWC 2006 (LNCS 4273, pp. 371-
384. Berlin: Springer Berlin/Heidelberg.
Gibson, J. J. (1977). The theory of affordances.
In R. E. Shaw & J. Bransford (Eds.), Perceiving, Louis, V., & Martinez, T. (2005a). The jade
acting, and knowing. Hillsdale, NJ: Lawrence semantic agent: Towards agent communication
Erlbaum Associates. oriented middleware. AgentLink News Journal,
18, 16-18.
Gibson, J. J. (1979). The ecological approach to
visual perception. Boston: Houghton Mifflin. Louis, V., & Martinez, T. (2005b). An operational
model for the FIPA-ACL semantics. In Proceed-
Gruber, T. R. (1993). A translation approach to
ings of the AAMAS’05 Workshop on Agent Com-
portable ontology specifications. Knowledge
munication (AC2005), Utrech, The Netherlands.
Acquisition, 5(2), 199-220.
Lu, S., Dong, M., & Fotouhi, F. (2002). The Se-
Hendler, J. (2001). Agents and the Semantic Web.
mantic Web: Opportunities and challenges for
IEEE Intelligent Systems and Their Applications,
next-generation Web applications. Information
16(2), 30-37.
Research, 7(4). Retrieved from http://Informa-
Hendler, J. (1999). Is there an intelligent agent in tionR.net/ir/7-4/paper134.html
your future? Nature. Retrieved from www.nature.
Luck, M., McBurney, P., Shehory, O., & Willmott,
com/nature/Webmatters/agents/agents.html
S. (2005). Agent technology: Computing as inter-
Hutchins, E. L., Hollan, J. D., & Norman, D. A. action (a roadmap for agent based computing).
(1986). Direct manipulation interfaces. Human- AgentLink.
Computer Interaction, 1(4), 311-338.
Maedche, A., & Staab, S. (2001). Ontology learn-
JADE. (2004a). JADE (java agent development ing for the Semantic Web. IEEE Intelligent Systems
framework). Retrieved July 20, 2007, from http:// and Their Applications, 16(2), 72-79.
jade.cselt.it/
Maes, P. (1994). Agents that reduce work and
JADE. (2004b). JADE documentation. Retrieved information overload. Communications of the
July 20, 2007, from http://jade.cselt.it/ ACM, 37(7), 30-40.
Jennings, N. R. (2000). On agent-based software Martin, C., Schreckenghost, D., Bonasso, P.,
engineering. Artificial Intelligence, 117, 277- Kortenkamp, D., Milam, T., & Thronesbery, C.
296. (2003a). Aiding collaboration among humans
and complex software agents. In Proceedings of
Klein, G., Woods, D. D., Bradshaw, J. M., Hoff-
the AAAI, 2003.
man, R. R., & Feltovich, P. J. (2004). Ten chal-
lenges for making automation a “team player” Martin, C., Schreckenghost, D., Bonasso, P.,
in joint human-agent activity. IEEE Intelligent Kortenkamp, D., Milam, T., & Thronesbery, C.
Systems, 19(6), 91-95.

600
Enabling Distributed Cognitive Collaborations on the Semantic Web

(2003b). An environment for distributed col- Norman, D. A. (1992). Design Principles for
laboration among humans and software agents. cognitive artefacts. Research in Engineering:
In Proceedings of the International Conference Design Theory Applications and Concurrent
on Autonomous Agents, 2(1), 1062-1063. Engineering, 4(1), 43-50.
Martin, D., Burstein, M., McDermott, D., McIl- Norman, D. A. (1988). The design of everyday
raith, S., Paolucci, M., Sycara, K., McGuinness, things. New York: Doubleday.
D. L., Sirin, E., & Srinivasan, N. (2007). Bringing
Norman, D. A. (1986). Cognitive engineering. In
semantics to Web services with OWL-S. World
D. A. Norman & S. Draper (Eds.), User centered
Wide Web, 10(3), 243-277.
system design: New perspectives on human-
Martin, D., Paolucci, M., McIlraith, S., Burstein, computer interaction. Mahwah, NJ: Lawrence
M., McDermott, D., McGuinness, D., Parsia, B., Erlbaum Associates
Payne, T., Sabou, M., Solanki, M., Srinivasan, N.,
Norman, D. A., & Draper, S. (1986). User centered
& Sycara, K. (2005). Bringing semantics to Web
system design: New perspectives on human-
services: The OWL-S approach. In Proceedings
computer interaction. Mahwah,NJ: Lawrence
of the SWSWPC (LNCS 3387, pp. 26-42).
Erlbaum Associates
Miller, C. A., & Parasuraman, R. (2007). Design-
Paolucci, M., & Sycara, K. (2003). Autonomous
ing for flexible interaction between humans and
Semantic Web services. IEEE Internet Comput-
automation: Delegation interfaces for supervisory
ing, 7(5), 34-41.
control. Human Factors, 49(1), 57-75.
Payne, T. R., Singh, R., & Sycara, K. (2002a).
McGrenere, J., & Ho, W. (2000). Affordances:
Calendar agents on the Semantic Web. IEEE In-
Clarifying and evolving a concept. In Proceedings
telligent Systems and Their Applications, 17(3),
of the Graphics Interface (pp. 179-186). Mahwah,
84-86.
NJ: Lawrence Erlbaum Associates.
Payne, T. R., Singh, R., & Sycara, K. (2002b).
Meredith, P. (1970). Developmental models of
RCal: A case study on Semantic Web agents. In
cognition. In P. L. Garvin (Ed.), Cognition: A
C. Castelfranchi & W. L. Johnson (Eds.), Pro-
multiple view. New York: Spartan Books.
ceedings of the 1st International Joint Conference
Mocan, A., Cimpian, E., & Kerrigan, M. (2006) on: Autonomous Agents adn Multiagent Systems
Formal model for ontology mapping creation. In (pp. 802-803).
Proceedings of The Semantic Web – ISWC 2006
Saja, A. D. (1985). The cognitive model: An ap-
(LNCS 4273, pp. 459-472).
proach to designing the human-computer inter-
Neuhold, E. (2003). Semantic Web application face. ACM SIGCHI Bulletin, 16(3), 36-40.
modelling. In Proceedings of the Conceptual
Sampson, J., & Lanzenberger, M. (2006). Visual
Modeling -ER 2003 (LNCS 2813, pp. 1-2).
ontology alignment for Semantic Web applica-
Nigay, L., & Coutaz, J. (1991). Building user inter- tions. In Advances in conceptual modeling – theory
faces: Organizing software agents. In Proceedings and practice (LNCS 4231, pp. 405-414).
ESPRIT’91 Conference (pp. 707-719).
Schreckenghost, D., Martin, C., Bonasso, P.,
Norman, D. A. (1994). How might people interact Kortenkamp, D., Milam, T., & Thronesbery,
with agents. Communications of the ACM, 37(7), C. (2002). Supporting group interaction among
68-71. humans and autonomous agents. Connection Sci-
ence, 14(4), 361-369.

601
Enabling Distributed Cognitive Collaborations on the Semantic Web

Scott, R., Roth, E. M., Deutsch, S. E., Malchiodi, Sycara, K., Pannu, A., Williamson, M., Zeng,
E., Kazmierczak, T. E., Eggleston, R. G., Kuper, D., & Decker, K. (1996). Distributed intelligent
S. R., & Whitaker, R. D. (2005). Work-centered agents. IEEE Expert-Intelligent Systems and their
support systems: A human-centered approach Applications, 11(6), 36-46.
to intelligent system design. IEEE Intelligent
Sycara, K., Klusch, M., Widoff, S., & Lu, J. (1999).
Systems, 20(2), 73-81.
Dynamic service matchmaking among agents in
Spiekermann, S., & Paraschiv, C. (2002). Moti- open information environments. SIGMOD Record
vating human- agent interaction: Transferring (ACM Special Interest Group on Management of
insights from behavioral marketing to interface Data), 28(1), 47-53.
design. Electronic Commerce Research, 2(3),
Tariq, A., Basharat, A., Ahmad, H. F., Suguri,
255-285.
H., & Ali, A. (2005a). A hybrid agent architec-
Shafiq, M. O., Ali, A., Ahmad, H. F., Suguri, ture for modeling autonomous agents in SAGE.
H., Tariq, A., Basharat, A., & Nazir, F. (2005). In M. Gallagher, J. Hogan, & F. Maire (Eds.),
A distributed services based conference planner Proceedings of the 6th International Conference
application using software agents, Grid services on Intelligent Data Engineering and Automated
and Web services. In Proceedings of the fourth Learning - IDEAL 2005 (pp. 478-485).
international joint conference on Autonomous
Tariq, A., Basharat, A., Ali, A., Ahmad, H. F., &
agents and multiagent systems, The Netherlands.
Suguri, H. (2005b). A collaborative multi-agent
New York: ACM.
based conference planner application using
Sun, R. (2001). Cognitive science meets multi- Web and Grid services. In Proceedings of the
agent systems: A prolegomenon. Philosophical 2005 International Conference on Collabora-
Psychology, 14(1), 1-28. tive Computing: Networking, Applications and
Worksharing.
Sutcliffe, A. (2000). On the effective use and
reuse of HCI knowledge. ACM Transactions on Vugt, H. C., Hoorn, J. F., Konijn, E. A., & de Bie
Computer-human Interaction, 7(2), 197-221. Dimitriadou, A. (2006). Affective affordances:
Improving interface character engagement
Sutcliffe, A., Ryan, M., Doubleday, A., & Spring-
through interaction. International Journal of
ett, M. (2000). Model mismatch analysis: Towards
Human Computer Studies, 64(9), 874-888.
a deeper explanation of users’ usability problems.
Behaviour and Information Technology, 19(1), Willmott, S. (Ed.). (2003). Technical input and
43-55. feedback to FIPA from Agentcities. RTD and the
Agentcities initiative (Agentcities Task Force
Sutcliffe, A. G., & Kaur, K. D. (2000). Evaluat-
Technical Note 00003). Retrieved from http://
ing the usability of virtual reality user interfaces.
www.agentcities.org/note/00003/
Behaviour and Information Technology, 19(6),
415-426. Wooldridge, M. J. (2002). An introduction to multi-
agent systems. New York: John Wiley & Sons.
Sycara, K. P. (1998). Multiagent systems. AI
Magazine, 19(2), 79-92. Zhang, J., Patel, V. L., Johnson, K. A., Smith, J.
W., & Malin, J. (2002). Designing human-centered
distributed information systems. IEEE Intelligent
Systems, 17(5), 42-47.

602
Enabling Distributed Cognitive Collaborations on the Semantic Web

Key Terms and Definitions Cognitive Model: Internal representations of


the current situation created by either a human
and agent to assess their state with respect to the
Affordance: An affordance is a resource or
environment.
support that the environment offers an agent for
action, and that the agent can directly perceive Cognitive Profile: It is a semantic represen-
and employ. tation model which includes information about
the cognitive states of an agent, its functional
Agent: Agents are defined as autonomous,
capacities and affordances.
problem-solving computational entities capable
of effective operation in dynamic and open en- COMMAA (Cognitive Model of Multi-
vironments. Agent Action): A framework for modeling agents’
actions and interactions in its environment in an
Capability: The functional ability possessed
attempt to provide an architecture that improves
by an agent to achieve some given goal or re-
the flexibility of Multi-agent interaction by pro-
quirement.
moting cognitive awareness .
Cognitive Awareness: It refers to the ability
Multi-Agent System: Multi-Agent System
of the Web agents to diagnose their processing
(MAS) is a distributed collaborative environment
limitations and to establish interactions with the
which allows a number of agents to cooperate and
external environment (in the form of other agents
interact with other agents (including both people
including humans and software agents).
and software) that have possibly conflicting aims,
in a complex environment.

This work was previously published in Handbook of Research on Social Dimensions of Semantic Technologies and Web Ser-
vices, edited by M. M. Cruz-Cunha; E. F. Oliveira; A. J. Tavares; L. G. Ferreira, pp. 610-642, copyright 2009 by Information
Science Reference (an imprint of IGI Global).

603
604

Chapter 2.18
Utilisation of Case-Based
Reasoning for Semantic Web
Services Composition
Taha Osman
Nottingham Trent University, UK

Dhavalkumar Thakker
Nottingham Trent University, UK

David Al-Dabass
Nottingham Trent University, UK

Abstract also utilizes OWL semantic descriptions exten-


sively for implementing both the components of
With the rapid proliferation of Web services as the CBR engine and the matchmaking profile of
the medium of choice to securely publish applica- the Web services.
tion services beyond the firewall, the importance
of accurate, yet flexible matchmaking of similar
services gains importance both for the human Introduction
user and for dynamic composition engines. In this
article, we present a novel approach that utilizes the The Internet has become the market-place for a
case based reasoning methodology for modelling colossal variety of information, recreational and
dynamic Web service discovery and matchmak- business services. Web services are increasingly
ing, and investigate the use of case adaptation for becoming the implementation platform of choice
service composition. Our framework considers to securely expose services beyond the firewall.
Web services execution experiences in the deci- Moreover, multiple Web services can be integrated
sion making process and is highly adaptable to either to provide a new, value-added service to
the service requester constraints. The framework the end-user or to facilitate co-operation between

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

various business partners. This integration of Web services matchmaking. Next we discuss the design
services is called “Web services composition” and of our matchmaking algorithm, its implementa-
is feasible to achieve because of the Web services tion highlights, and analyze preliminary results.
advantages of being platform, language neutral Finally we investigate how case adaptation can
and loosely coupled. further extend our matchmaking algorithm to
Automatic Web service discovery and match- cater for service composition and review related
making is the principal aspect for dynamic services work.
composition. The accuracy of the matchmaking
(selection) process enhances the possibility of
successful composition, eventually satisfying the Motivation
user and application requirements. The current
standard for Web service discovery, the Universal The most practically deployed Web services
Description, Discovery and Integration (UDDI) composition techniques use the theory of business
registry is syntactical and has no scope for auto- workflow-management as composition process
matic discovery of Web services. Hence, current model to achieve formalization for control and
approaches attempting to automate the discovery data flow. Mainly based on the Business Process
and matchmaking process apply semantics to the Execution Language (BPEL) standard (Andrews
service descriptions. These semantics are interpre- et al., 2003), these techniques also have practical
table by the service (software) agents and should capabilities that fulfil the needs of the business
include WSDL-based functional parameters such environment, such as fault handling and state
as the Web services input-outputs (Martin et al., management. However, the main shortcoming
2004a)‎ (Akkiraju et al., 2005), and non-functional of these techniques is the static selection and
parameters such as domain-specific constraints composition approach, where the service selec-
and user preferences (Aggarwal, Verma, Miller, tion and flow management are done a priori and
& Milnor, 2004). manually.
The accuracy of automatic matchmaking of A popular research direction attempts to
web services can be further improved by taking improve BPEL composition by introducing se-
into account the adequacy of past matchmak- mantics to workflow-based composition (Osman,
ing experiences for the requested task, which Thakker, & Al-Dabass, 2005). However, these
gives us valuable information about the services approaches also match the static behaviour of
behaviour that is difficult to presume prior to Web services in terms of whether the service has
service execution. Hence, there is a need for a similar description for functional and non-func-
methodology that uses domain-specific knowledge tional parameters. While for the candidate Web
representation of the required task to capture the services it is highly likely that these parameters
Web services execution experiences and utilise are semantically similar, it is the execution values
them in the matchmaking process. Case Based for such functional and non-functional parameters
Reasoning (CBR) provides such methodology as that provide valuable guidance for the decision-
its fundamental premise is that experience formed making process regarding the service adequacy
in solving a problem situation can be applied for for the task. This is because service behaviour
other similar problem situation. is difficult to presume prior to service execution
The article begins with describing the moti- and can only be formed based on the experience
vation behind the work. In the following section with the service execution.
we review theory of Case Based Reasoning and Hence, the problem requires a methodology,
describe how it can be utilised for modelling Web which has the domain-specific knowledge repre-

605
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

sentation system for capturing the Web services that address the problem and inform about the
execution experiences and reason based on those relevance of the solution. When a reasoner is
experiences. We adopted CBR (Case Based created, the elements of the case are defined ac-
Reasoning) as the engine for our Web services cording to the context. For example, the city of
discovery mechanism because CBR’s fundamen- departure or the number of passengers could be
tal premise that situations recur with regularity some elements to represent a travel experience as
(Aamodt & Plaza, 1994), i.e. experience involved a case. Case vocabularies are thus developed for
in solving a problem situation can be applied or each reasoner, to define what knowledge needs
can be used as guide to solve other contextually to be captured.
similar problem situations. Reasoner based on
CBR hence matches the previous experiences to Case Storage and Indexing
inspire a solution for new problems.
Cases are then stored in a case library or case base.
It is an important aspect for the designing of CBR
Overview of case based systems because it reflects the conceptual view
reasoning of what is represented in the case. The structure
of the library should permit efficient search by
The Case-Based Reasoning technology was the reasoner. This search can be facilitated by the
developed in 1977 based on the research effort use of indexing. Indices are therefore assigned to
of Schank and Abelson. They proposed that our cases in order to express information about the
general knowledge about situations is recorded case content.
in the brain as scripts that allow us to set up
expectations and perform inferences (Watson, Case Retrieval
1997) The processes involved in CBR can be
represented by a schematic cycle comprising four Whenever a new problem needs to be solved, the
phases (Aamodt et al., 1994): case library is searched for the cases which can
be a potential solution. The first phase of this
• RETRIEVE the most similar case(s); search is case retrieval, which aims to find cases
• REUSE the case(s) to attempt to solve the that are contextually similar to the new problem.
problem; The retrieval is done according to the index of
• REVISE the proposed solution if necessary, the cases.
and
• RETAIN the new solution as a part of a new Matchmaking
case.
Matchmaking performs the comparison between
There are 4 main stages in CBR reasoning: retrieved cases and the request to verify if a past
solution can be reapplied. There are several avail-
Case Representation able methods for matchmaking in CBR literature.
The Nearest-Neighbour Matching and Ranking
A case is a contextualised piece of knowledge is an interesting one because it involves the as-
representing an experience (Aamodt et al., 1994). It sessment of similarity between stored cases and
contains the problem, a description of the state of the input (request) case. It assigns importance
the world when the case occurred, and the solution ranking to properties of cases and then computes
to this problem. The solution contains elements the degree of matching by comparing the cases

606
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

for these properties (Kolodner & Simpson, 1989). services during the case retrieval or Web service
The matchmaking process is thus performed on discovery phase respectively.
each retrieved case, and the most similar case to The apparent compatibility confirms our
the input case is assigned the highest ranking. If thesis that the CBR methodology is well suited
the system finds a matching case, it is possible to build automatic Web service composition
to reuse the solution suggested by the retrieved frameworks
case for the new problem.
In our CBR matchmaking approach, Web ser-
vices execution experiences are modelled as cases. Matchmaking Web Services
The cases are the functional and non-functional Using Case Based Reasoning
domain specific Web services properties described
using semantics. In this modelling, the case The Framework Architecture
library will be the storage place for such execu-
tion experiences and is identical to Web service In our Semantic CBR matchmaking, there are two
registry in that it stores Web services references, main roles: case administrator who is responsible
but unlike registries case libraries also describe for case library maintenance by entering or delet-
execution behaviour. ing cases from the library and case requestor who
Case retrieval is similar to Web services dis- searches the case library to find solution for the
covery problem in that both mechanisms seek problem. Figure 1 illustrates a schematic diagram
to find potential Web services for the current for our framework.
problem. Case matchmaking is similar to Web The dynamics of the framework operation is
services matchmaking as both attempts to select as follows:
acceptable Web services, from the retrieved Web

Figure 1. Architecture of the CBR matchmaking framework

Admin

candidate
domain web service
representation

problem new
service description problem Semantic
case Description
requester web service repository semantic case Genrator
reference represntation
annotated problem description
matched cases
Indexing
Case
Matchmaking new problem index

case with
Case Retrieval
similar index
case library

607
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

• Initially, the administrator populates the driven design of framework makes it application-
repository with semantic case representa- independent and allows for its seamless reuse
tion formats for specific application domain. for other applications domain. In order to enable
This representation is used to semantically matchmaking for the financial markets domain
annotate both the user requests for suitable for instance, it would suffice to enter a new case
services and the execution experiences of representation format into the repository, keeping
Web services for the specific domain. the rest of the reasoning logic intact.
• The user inputs the service requirements
and as a result receives Web service refer- Ontology Support for Case
ences via the framework interface. The same Representation and Storage
interface is used by service provides or the
system administrator to subscribe a Web The most common use of ontologies is the recon-
service as a candidate for available services ciliation between syntactically different terms that
for the specific domain. are semantically equivalent. Applied to CBR case
• The case representation repository retrieves descriptions for Web services, ontologies can be
the appropriate semantic case representation used to provide a generic, reasoner-independent
format for the requested service and forwards description of their functional and non-functional
it together with the problem description to parameters. Moreover, ontologies can also be
the Semantic Description generator mod- used to further index and structure cases with key
ule, which semantically annotates the new domain features that increase the efficiency of the
problem according to the representation matchmaking process. For instance, we can add a
format. feature to the travel domain ontology to indicate
• The annotated problem is then passed to whether a trip is domestic or international. Web
the indexing module, which computes a services QoS parameters are also indexed using
suitable index for the new problem based ontologies to further improve the accuracy of
on the domain feature and/or the functional case matchmaking.
parameters of the requested service. The In our framework, ontologies are also used to
index is passed for case retrieval. describe the rules of the CBR reasoning engine,
• The case retrieval module queries the case which not only streamlines the intercommunica-
library for cases with similar indexes. Out- tion between the Web service, user request, and
put at this stage will be the cases that have the case library, but promotes exploring the col-
similar index to the current problem, which laboration at the reasoning level between different
will be candidates for matchmaking. composition frameworks.
• The case matchmaking module takes the
retrieved cases and the annotation of prob- Case Vocabulary
lem description from the semantic descrip-
tion generator module, runs them through In CBR theory, the first step is to define all the
a matchmaking algorithm and forwards elements contained in a case and the associated
the closest match Web service to the re- vocabulary that represents the knowledge associ-
quester. ated with the context of a specific domain (our
case study is the travel domain).
Although the chosen case study for this work This vocabulary includes functional and non-
is from the travel domain, the modular, ontology- functional parameters:

608
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

• Functional parameters are the service input semantic net representations largely borrowed
(e.g. the travel details) and the service output from the frame structures(Elaine & Kevin, 1992),
or results (e.g. the travel itinerary). Input which makes natural transition to the Semantic
corresponds to the request of the user (e.g. Web descriptions possible. Table 1 shows such
date or city of departure) whereas output a frame structure for our travel domain case
corresponds to the response given to the vocabulary.
user (e.g. price, flight number). The slot Travel Request corresponds to the In-
• Non-functional parameters are constraints put, i.e. all the travel details for a travel agent. The
imposed by the user (e.g. exclusion of par- Travel Response slot corresponds to the Output,
ticular travel medium) or preferences over i.e. the answer given to the user at the end of the
certain specific parameters (e.g. Price range, process. The elements of the answer are the price
Quality of Service expected). In addition, and the corresponding currency, the access point
execution experiences stored in the case to the WSDL file of the corresponding Web Ser-
library should also include the solution (i.e. vices and the Services Used (companies involved
Web services effectively used) and a notion in the trip, e.g. an airline and a hotel).
to specify if the solution is acceptable for
the end-user. Features that characterise the Semantic Encoding of the Frame
domain are extremely useful for top-level Structure
indexing and can also be included as non-
functional parameters. In the developed framework, we map the frame
structures to ontologies. We derive rules for such
Case Representation Using Frame mapping as described in Figure 2.
Structures According to this mapping, frame and slot are
represented as classes. The relationship between
After deciding on the knowledge and corre- frame and slot is expressed in terms of properties
sponding vocabulary to be represented as a case, of a frame, i.e. the range for these properties are
we need to decide how this knowledge can be the slot classes. Dimensions are the properties of
represented. the slots. Possible range for these properties is the
In our approach, we adopt frame structures values the respective filler can derive from.
(Kolodner et al., 1989) the case representation. We use Web Ontology Language (OWL), a
In frame structures, frame is the highest repre- Semantic Web standard for constructing these
sentation element consisting of slots and fillers. ontologies. OWL is the most expressive Semantic
Slots have dimensions that represent lower level Web knowledge representation so far. The layered
elements of the frame, while fillers are the value approach adopted by semantic web, allows reason-
range the slot dimensions can draw from. In ing and inference based on ontologies, which is the
our implementation, slot dimensions represent most powerful and ubiquitous feature of Semantic
case vocabulary in modular fashion while fillers Web. After applying the mapping, the ontology for
describe the possible value ranges for the slot the travel domain case representation is created,
dimensions. where for instance the CaseRepresentation class
The frame representations are highly struc- has: hasTravelResponse, hasConstraintsOnGoal,
tured and modular which allows handling the and hasFeature object properties. Range for these
complexity involved in representation. More- properties are TravelResponse, Constraints, and
over, frame structure has a natural mapping to Feature classes respectively.
the semantic OWL description language as the

609
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

Table 1. Representation of a case


Slot Dimension Filler
Name of Traveler Any text
Travel Request Date of Arrival Any valid date
City of Departure Any valid city

Solution Service WSDL file


Travel Response Price Range Any positive Double
Currency Any valid currency
On Domain Any Valid Travel Domain
Constraints On Price range Any positive Double
On QoS parameter Any possible QoS parameter(s)
Features Travel Regions Domestic/International

Figure 2. Mapping between frame structure and semantic case representation

CaseRepresentation frame

hasFeatures
property
hasTravelResponse hasConstraintsnGoal

TravelResponse Features Constraints slot

price
city dimension

$520 London filler

globalization of semantic descriptions, we range in total of $220, He does not want to travel
used external ontologies where appropriate. For by road. The dates of Travel will be 27-02-2005
instance, the property cityOfArrival is an ob- for departure and 01-03-2005 for return. He pre-
ject property referring to a publically available fers to pay in USD and requires a fast response
ontology(PORTAL, 2003) , where other useful (approximately in 1.5 seconds)”.
information about the specific city can be found
such as country, the number of inhabitants, etc. Case Storage
An example of a semantically-encoded travel
request is illustrated in Table 2. “Find a Trip for All the Web service execution experiences, i.e.
a single person, Mr Lee; Mr Lee wants to travel solutions deemed valid for a particular request,
from Boston to New York, with a maximum price are stored in the Case Library to be reused by

610
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

In order to exercise the noble objective of Development of the CBR


globalization of semantic descriptions, we used matchmaking framework
external ontologies where appropriate. For in-
stance, the property cityOfArrival is an object Case Indexing and Retrieval
property referring to a publically available
ontology(PORTAL, 2003) , where other useful To facilitate the search procedure, cases are in-
information about the specific city can be found dexed based on vocabularies. In our framework,
such as country, the number of inhabitants, etc. we use “partitioning the case library” method,
An example of a semantically-encoded travel which is a variation of “flat memory indexing”
request is illustrated in Table 2. “Find a Trip for technique (Kolodner et al., 1989). In this indexing
a single person, Mr Lee; Mr Lee wants to travel method, case library is partitioned based on certain
from Boston to New York, with a maximum price vocabularies and the new problem is recognized
range in total of $220, He does not want to travel based on the identical vocabularies to decide
by road. The dates of Travel will be 27-02-2005 which partition the problem falls into.
for departure and 01-03-2005 for return. He pre- In our architecture, cases are stored based on
fers to pay in USD and requires a fast response vocabulary element Features as presented in Table
(approximately in 1.5 seconds)”. 1, which corresponds to hasFeatures property from
the CaseRepresentation ontology class. For our
Case Storage travel agent case study, the possible values for this
property are either Domestic or International (pre-
All the Web service execution experiences, i.e. defined instances from the TravelRegion class),
solutions deemed valid for a particular request, hence indexing will partition case library into
are stored in the Case Library to be reused by two parts. In more complex examples more than
the reasoner. The Case Library itself is also an one vocabulary term or a combination of terms
ontology. It contains some instances of the class can be used for more sophisticated indexing. As
CaseRepresentation (e.g. a travel experience or in relational databases selection, the efficiency
a travel case). of the retrieval process largely depends on the
precision of the indexing.

Table 2. Example of a case


Name of passengers Lee <TravelRequest:namePassengers>Lee
<TravelRequest:cityArrival rdf:resource= “http://localhost/
City of Arrival Boston
uk/2005/City.owl#Boston”/>
<TravelRequest:dateArrival>2005-03-01
Date of Arrival 1-3-05

Constraint on <Constraints:OnDomain rdf:resource=


Road
domain “http://localhost/ntu/TravelDomain.owl#Airline”/>
Constraint on price 220 <Constraints:OnPrice>220
Constraint on <Constraints:OnCurrency rdf:resource= “http://ecs.soton.ac.uk/
USD
currency currency.daml#USD”/>
Constraint on QoS 1.5 s <QoS:ExecutionDuration>1.5

611
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

Whenever a new Web service needs to be Data Property Comparison


fetched, the problem description involving the
functional parameters and non-functional param- To compare data type properties, like the price
eters are encoded using the case representation range or the value of QoS (e.g. execution time),
frame structure, i.e. as an instance of CaseRepre- we use the qualitative regions based measurement
sentation ontology as illustrated in Table 2. method (Kolodner et al., 1989). The closer the value
in a retrieved case is to the value in the request,
Matchmaking and Ranking the higher the similarity coefficient is.
For each data type property, this formula used
Case retrieval fetches Web services that are a is: |Vr − Vc| ≤ X.[Vr|, where V is the value of the
potential solution to the problem. The matchmak- property in the request r or in the retrieved case
ing process narrows down the retrieved cases to c and X the factor of tolerance. Thus, a factor of
present acceptable solution(s). From the available tolerance of 0.9 means the value of the retrieved
methods for matchmaking in CBR literature, we case should be in ±10% region in relation to the
choose Nearest-Neighbour Matching and Rank- value of the request. The optimum tolerance value
ing using numeric evaluation function (Remind, is determined by the administrator and can be
1992) method for our framework. The method calculated heuristically.
operates as follows:
Object Property Comparison
• Compare the similarity for each property,
between the new problem and the cases For the dimensions annotated as object proper-
retrieved. The method used for comparison ties, the possible filler values will be an instance
depends on the type of the property. of slot class. Hence, for semantically matching
• Quantify the weight of the similarity. A object property value of the new problem and
ranking is assigned to each property in ac- the retrieved cases, the algorithm compares the
cordance with its importance as exemplified instances. If the instances match, then the degree
in Table 3. of match is 1. Otherwise, the algorithm traverses
back to the super (upper) class that the instance
For each case retrieved, the similarity degree is derived from and the comparison is performed
is computed and the case with the highest score at that level.
corresponds to the best-match. Similarity takes The comparison is similar to traversing a tree
values between 0 and 1, which is attributed to each structure (Zhang, Arpinar, & Aleman-Meza,
property for each retrieved case. Our similarity 2003), where the tree represents the class hierar-
comparison method depends on the type of the chy for the ontology element. The procedure of
dimension: data or object. traversing back to the upper class and matching

Table 3. Quantifying the travel domain case dimensions


Slot Dimension Importance (0-1)
City Departure 1.0
Travel Request
City Arrival 1.0

On Instance 0.2
Constraints on Goal
On Domain 0.8

612
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

Figure 3. Semantically matching object properties (dimensions)

Continent America Europe America

Country USA UK USA

Dimension New York London New Jersey

Request Case#1 Case#2

instances is repeated until there are no super 1992) for each retrieved case according to the
classes in the class hierarchy, i.e. the top node for following equation:
the tree is reached, giving degree of match equal to
0. The degree of match (DoM) degree is calculated n
according to the following equation: ∑W × sim( f
i =1
i i
N
, fi R )
ADoM = n
(2)
∑W
MN
DoM = (1)
i
GN
i =1
Where the MN is Total number of matching
nodes in the selected traversal path, and GN Total Where, n is the number of ranked dimensions,
number of nodes in the selected traversal path Wi is the importance of dimension i, sim is the
For example, for the request in Figure 3, case#1 similarity function for primitives, and fiN and fiR
will return a degree of match of 0 because no are the values for feature fi in the new problem
matches are found while traversing the ontology and the retrieved case respectively.
tree until the leaf node is reached. However, for The evaluation function sums the degree of
case#2, the degree of match will be 2/3=0.67 as match for all the dimensions computed in the
the instances (New Jersey, New York) does not previous step, and takes aggregate of this sum by
match but the instances of the Country super considering the importance of dimensions.
class match.
It is worth to note that Constraints on object
properties are handled by omitting that path in Implementation
the case ontology tree that renders the constraint highlights
invalid. For example, if the passenger is reluctant
to travel by air, then the Brit Air, Flight path will The implementation of our framework uses se-
not be traversed. mantics extensively to implement both the util-
ity ontologies describing the components of the
Computing The Overall Similarity Value Case-Based Reasoner (Case representation), and
the domain ontologies that describe the profile
Overall similarity is evaluated by computing the of the Web services in the Case library with a
aggregate degree of match (ADoM) (Remind, semantic representation (Case Storage).

613
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

Figure 4. Admin and user interface

OWL was our ontology language of choice. Preliminary Results


We used Pellet (Parsia & Sirin, 2004) - a Java
based OWL reasoner, as our ontology engine At this initial stage of development, the focus of
in favour of the more popular Jena (HP-JENA, our experiments was to validate the logic of our
2003), because it supports user-defined simple matchmaking framework, rather than testing a
types. Pellet was used to load and verify (type fully working prototype. Hence, we tested our
and cardinality) ontology class instances of user framework with simple in-house developed Web
requests and candidate cases. services and compatible wrappers for external
Figure 4 illustrates a snapshot of the GUI publicly available services.
developed for the matchmaking framework. The In order to consolidate the test process, we
interface allows different options to two kinds of applied different rankings against each test case
users: The case administrator, who is responsible and associated them with a specific profile. The
for maintaining the case library, and a standard profile represents a group of users that have similar
client, who wants to retrieve Web services for a requirements for the travel request. For instance,
trip. The case administrator has admin privileges the Business profile stands for corporate users,
to perform case maintenance activities like case who have to travel frequently; therefore a high
seeding, modifying the ranking system or delet- standard of comfort is a significant element of
ing old cases. The client can also setup a ranking choice. These users also need reliability of ser-
system, which will be applicable for a particular vices. Price is not very important because firms
session. very often have contracts with travel companies.
While seeding the case library with a new On the other hand, for regular users, represented
case or making a new trip request, the interface by the Personal profile, cost is of paramount
assists the client in creating the required ontol- importance.
ogy instances. The value entered for a particular The three other types of users are mainly based
property is validated in relation to the range and on specific comparison properties: Economic
cardinality drawn from the ontologies. retrieves cases which price never exceeds a user-
The solutions (cases) resulting from the defined maximum amount; Travel Medium is
matchmaking process are presented to the client specific for constraints on travel domain as well as
are stored into the case library.

614
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

instances; and Enterprise is useful for companies Figure 5 shows the matchmaking degree for
which are interested in using reliable services. different cases using the criteria above. Some cases
The latter can be important if contracts between (Web service execution experiences) present satis-
the company and different Web services exist so factory results to all users (CaseInst10511611478).
that they can restrict other services. Another interesting highlight is that the chosen
The rankings are currently administered cen- ranking systems provide different results only if
trally, but in the future we would like to give the the coefficients are significantly different. This is
users the opportunity to tweak some of them using probably due to the fact that that our case library
a user-friendly interface. Table 4 shows the rank- is not richly populated at the moment.
ing of our profile system. Example of constraint The average execution time of our matchmak-
on Domain is reluctance to travel using a certain ing program at the time of the experiment was
transport and constraint on instance the exclusion approximately 40 seconds, relatively slow consid-
of certain airline from the search. Quality of ser- ering we only have 30 cases stored in the library.
vice is represented as a single parameter, but in Using semantics has the disadvantage of being
this experiment it is expressed as the availability more time-consuming than scanning databases.
and response time of the service. We identified the use of imported ontologies as the

Table 4. User profiles


Property
Constraint on Constraint. Quality of
Profile Category Price
Domain on Instance Service
Business 0.6 0.6 0.4 0.1 0.5
Personal 0.2 0.4 0.7 0.5 0.2
Economic 0.2 0.4 0.2 1 0.1
Travel Medium 0.2 1 0.8 0.3 0.2
Enterprise 0.5 0.3 0.1 0.2 1

Figure 5. User profiles


0,65
factor describing the equivalence

0,6 CaseInst10511611478
0,55 CaseInst105116115144
0,5 CaseInst105116114943
0,45 CaseInst10511611579
0,4 CaseInst105116115410
0,35
0,3
0,25
0,2
0,15
0,1
0,05
0
Business Personal Economie Medium Enterprise
different users

615
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

main performance leak for our program. We plan between the retrieved case and the current case
to develop an off-line caching system to enable and then applies formulae or rules that take those
us to access the public ontologies locally. differences into account when suggesting a solu-
tion (Watson & Marir, 1994).
Applied to the current framework, when the
Extending the matchmaking existing web services experiences in their original
framework to web services form are not sufficient to satisfy current request,
composition the framework should look for relaxing the case
restrictions under which a solution is acceptable.
The current framework addresses the problem of If the latter fails, the framework should attempt
automatic Web services discovery and matchmak- to merge potential cases to suggest a composite
ing by annotating Web services execution experi- solution.
ences and storing them into case base (Osman et
al., 2006). The search considers domain-specific Case adaptation can be defined by the follow-
criteria and user preferences to find Web services ing formula (Maher & Garza, 1997)
execution experience that solved a similar prob-
lem in the past. However the framework assumes C ' = (C ) (3)
that the case library contains suitable cases for
every possible problem. This assumption is not Where, C’ = new case, C = old case(s) and α
always satisfied considering the vast number of indicates adaptation operator.
problems and problem parameters. Moreover, The adaptation operator indicates the process
the framework also needs to deal with situations of identifying and substituting or transforming an
where the aggregate degree of match (ADoM) existing solution to fit new situations and is used
is below the domain-specific expected degree of in knowledge-based substitution adaptation.
match set by the domain administrator or to deal
with negative user feedback, where the matched Knowledge Based Substitutions
services are not acceptable to the user.
Work under progress involves exploring case In CBR matchmaking process previous cases
adaptation, which is termed as the REVISE phase cannot be always reused without making some
(Figure 6 - The REVISE phase in CBR) in CBR changes. Reasoning about these changes requires
theory. Adaptation is applicable when the available general and domain specific knowledge to mould
cases cannot fulfil the problem requirements, so case adaptation. For example, if an existing case-
matchmaking is attempted by adapting available solution is applicable for the current travelling
cases. Adaptation looks for prominent differences problem with the exception of the travel medium

Figure 6. The REVISE phase in CBR

REVISE
IF ADoM < {Cresults, ADoMexpected}
OR
IF{ Cresults} Uexpected

Uexpected User expected results


Cresults proposed case solutions

616
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

Figure 7. Local search • In a situation, where the reasoner cannot


find an exact or reasonable match or where
exact match is not possible or not desired
or not required.
• To make absolutely sure that only possible
solution is transformation, which is an ex-
pensive operation involving an AI planner,
which is a resource expensive exercise.

Hence, if for the current problem C, the avail-


able cases in the case library are:

Clibray = { C1 (S1+S2, F1), C2 (S2+S3, F2), C3 (S1,F3),


– Bus (a road based travel medium which is con- C4 (S2,F4) , C5 (S1+S2, F5) }
strained in the current problem), then the reasoner
should use local search on knowledge structure Where, C (S, F) indicates cases with Web services
(see the ontology in Figure 7) and conclude the as solution S applied under circumstances defined
Taxi domain as a possible substitution. Similarly, by F. The circumstances can be characterized by
interpolating parameter of old solution to adapt service description, problem description, con-
for the new problem can solve the problem of straints and preferences applied while solving
parameter mismatch. the problem.
Another substitution method can use the Then the solution for new problem C, C6 (S1 +
semantic “sibling” rule for equivalent classes to S2, F6) must be reached by exploring the matching
enables them to replace each other in order to cases C1 and C5 first, before transforming C3 and
present an appropriate solution. C4 to find a solution from a scratch.
Under this circumstance, the Equation 3 can
be reformulated as: Planning Based Transformations
C ' = (C , K ) (4) The planning based transformations can be ap-
plicable when the available solutions can not
Where K indicates the influence of general or fulfil the problem requirements with normal
domain specific knowledge. matchmaking and discovery mechanism or by
The role of knowledge in repairing the existing applying minor modifications using substitution
cases can be described as follows: based transformation. Under these circumstances,
the equation 3 can be reformulated as:
• Relaxing the service descriptions (functional
parameters) to find a sufficiently similar de- C ' = (C , ) (5)
scription (applied at the description level).
• Relaxing the execution values of candidate ρ indicates the application of planner for
cases (their non-functional parameters) in transformation, where classical planner handles
an attempt to adapt the solution. the task of coming up with a sequence of actions
that will achieve a goal (Russell & Norvig, 2003).
The criteria for applying knowledge based The planning phase is a resource-intensive and
substitution are: computational expensive (Selman, 2000; Long &

617
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

Figure 8. CBR methodology for Web services composition ( modified from [(Aamodt et al., 1994)])

RETRIEVE REUSE
6. Reuse and retain.
1. Search f or cases (composite services
/atomic services), already existing in the
library which will solve the current
problem.
If such case exists, reuse.

2. If similar cases exist and require minor


modification, retrieve cases and go to
revise. Previous
OR Cases

If similar cases do not exist, but partial


matches exist then go to revise. Knowledge
Base

REVISE

3. Use knowledge base to apply minor


changes to existing cases
(Knowledge based substitution) RETAIN
4. Use planning to adapt the solution
7. S tore in case lib
(Planning based transformation)

5. Verify and retain.

Fox, 2002) yet inevitable option, hence the two input parameters of other available Web services.
previous phases should narrow down the number A constraint driven composition framework in
of possible services planner can successfully use (Aggarwal et al., 2004) also uses functional and
to generate a composed service. data semantics with QoS specifications for select-
Figure 8 on the shows the holistic CBR meth- ing Web services. DARPA’s OWL-S (Ontology
odology to achieve Web services composition Web Language for Web services) is the leading
using the REVISE cycle. semantic composition research effort. OWL-S
(Martin et al., 2004b; Ankolekar et al., 2001) on-
tologies provide a mechanism to describe the Web
Related work services functionality in machine-understandable
form, making it possible to discover, and integrate
Semantic descriptions are increasingly being Web services automatically. An OWL-based
used for exploring the automation features re- dynamic composition approach is described in
lated to Web services discovery, matchmaking (Sirin, Hendler, & Parisa, 2003), where semantic
and composition. In (Zhang et al., 2003) such description of the services are used to find match-
semantic-based approach is described. They use ing services to the user requirements at each step
ontology to describe Web services templates and of composition, and the generated composition is
select Web services for composition by compar- then directly executable through the grounding
ing the Web service output parameters with the of the services. Other Approaches use Artificial

618
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

Intelligence planning techniques to build a task and targets highly adaptive systems that can react
list to achieve composition objectives: selection themselves to different business and organization
of services and flow management for performing settings. The adaptation is achieved through the
composition of services to match user preferences. CBR based exception handling, where the CBR
(Mcllraith & Son, 2002) uses Golog – AI planning system is used to derive an acceptable excep-
Reasoner for automatic composition, while in a tion handler. The system has the ability to adapt
similar spirit some other approaches (Wu, Parisa, itself over time, based on knowledge acquired
Hendler, Nau, & Sirin, 2006; Nau, Cao, & Lotem, about past execution experiences that will help
1999) have used the paradigm of Hierarchical Task solve new problems. Our approach concentrates
Network (HTN) planning to perform automated on Web services as a unit of computation to
Web service composition. These approaches use take advantage of highly accessible and loosely
semantics for automatic Web services discovery, coupled nature of Web services technologies. We
but they overlook the Web service execution be- focus on utilising service execution experiences
haviour in the decision-making process. to best serve user requirements and encode the
Use of CBR, Semantic Web and Web services framework with semantics.
are common technologies in our effort and the Experience based learning using CBR is a
efforts in (Nern et al., 2006) with different objec- relatively old branch of Artificial Intelligence and
tives, their’s being to consume these technologies Cognitive Science and is being used (Hammond,
to assist the procedure of Semantic Web services 1986; Ashley & Rissland, 1988) as an alternative
creation using Case-Based reasoning approach, to rule-based expert system for the problem do-
while our main concern is services composi- mains, which have knowledge captured in terms
tion.. Their INFRAWEBS project has Semantic of experiences rather than rules. However, Case
Web Unit (SWU) – a collaboration platform and based reasoning for Web services was initially
interoperable middleware for ontology-based han- documented in (Limthanmaphon & Zhang, 2003),
dling and maintaining of Semantic Web services. where the developed framework uses CBR for
The framework provides knowledge about a spe- Web services composition. In their approach,
cific domain and relies on ontologies to structure the algorithm for Web services discovery and
and exchange this knowledge to Semantic Web matchmaking is keyword based and has no notion
services development process. for semantics. This affects the automation aspects
There is also a number of existing approaches for Web services search and later for composition.
which applies CBR for workflow modelling. Similar approach described in (Diaz, Salgado,
(Madhusudan, Zhao, & Marshall, 2004) proposes Moreno, & Ortiz, 2006) proposes an extension
an approach to support workflow modelling and of UDDI model for web services discovery us-
design by adapting workflow cases from a reposi- ing category-exemplar type of CBR, where web
tory of process models where workflow schemas services are categorized in domains and stored as
are represented as cases and are stored in case exemplar (Porter & Bareiss, 1986) of particular
repositories. The cases are retrieved for a problem domain. Their implementation of CBR reasoner
which requires similar business process to solve facilitates UDDI registry by indexing the cases
the problem. The description and implementation based on the functional characteristics of Web
language of framework is based on XML and services. However, the approach does not take
main focus is on assisting workflow designer in into consideration the importance of non-func-
creating business process flows. In similar line, tional parameters in service selection and the
(Cardoso & Sheth, 2005) represents adaptive use of semantics at CBR level is peripheral as
workflow management system based on CBR they primarily use the UDDI based component

619
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

for service discovery. UDDI is text-based leav- achieves developer transparency and makes the
ing little scope for automation. Our framework framework extensible and reusable.
consumes semantics extensively and achieves the A problem that research in semantic-based
automation required for Web service discovery matchmaking and composition has not addressed
and matchmaking. Use of ontologies also makes sufficiently is the interoperation between inde-
our framework extensible and reusable. pendently developed reasoning engines. Without
this interoperation, the reasoning engines remain
imprisoned within their own framework, which
Conclusion is a drawback, especially that most engines usu-
ally specialise in servicing a particular domain,
Semantic description of Web service profile hence interoperation can facilitate inter-domain
paves the way for automating the discovery and orchestration. We believe that in this work we
matchmaking of services since it allows intelligent took a small step towards standardization at the
agents to reason about the service parameters reasoner level by describing the CBR reasoning
and capabilities. However, the accuracy of such model semantically
automatic search mechanism largely relies on In this article we also presented the preliminary
how soundly formal methods working on such experimental results of our framework, which
semantic descriptions consume them. informally proved the correctness of our approach
In this article, we argued for the importance of despite the relatively slow response time of the
considering the execution values for semantically matchmaking process. The latter is primarily at-
described functional and non-functional Web tributed to exporting external ontologies, which
services parameters in decision making regarding can be countered by utilising off-line caching of
Web service adequacy for the task. This is because public ontologies. The experimental results also
the service behaviour is impossible to presume demonstrated the advantages of classifying user
prior to execution and can only be generalized groups into profiles that have standard set of
if such execution values are stored and reasoned constraint rankings.
for deciding service capability. AI planning and The final contribution of the article was
Intelligent Agent based reasoning methods offer documenting our investigation into extending the
rule-based reasoning methodology rather than discovery and matchmaking algorithm to cater
experience-based. Hence, we used Case Based for web services composition. We discuss how
Reasoning method that allows capturing experi- we envisage exploiting the REVISE stage of the
ences and reasoning based on them. CBR cycle, i.e. case adaptation, to facilitate service
We implemented a Semantic Case based composition. The article advocates an exhaustive
Reasoner, which captures Web service execution knowledge-based substitution approach to adapt
experiences as cases and uses these cases for the functional and non-functional attributes of the
finding a solution for new problems. The imple- candidate case to the requested solution before
mented framework extensively uses ontologies, as suggesting more complex and computationally
semantics are used both for describing the problem taxing AI-based planning-based transformations
parameters and for implementing components that integrate the service profile of a number of
of the CBR system: representation, indexing, cases to deliver candidate solutions.
storage, matching and retrieval. Our approach The next stage of this research will involve
for modelling CBR as ontology-based reasoner the formal validation and implementation of our
adaptation-based composition model.

620
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

References Elaine, R. & Kevin, K. (1992). Artificial Intelli-


gence. McGraw-Hill.
Aamodt, A. & Plaza, E. (1994). Case-based
Hammond, K. J. (1986). Learning to anticipate and
reasoning: foundational issues, methodological
avoid planning problems through the explanation
variations, and system approaches. AI Commu-
of failures. In AAAI-86 conference (pp. 556-560).
nications, 7, 39-59.
Cambridge, MA: AAAI press/MIT press.
Aggarwal, R., Verma, K., Miller, J., & Milnor, W.
HP-JENA (2003). Jena- A semantic Web Frame-
(2004). Constraint Driven Web Service Composi-
work for Java. http://jena.sourceforge.net/ [On-
tion in METEOR-S. In 2004 IEEE International
line]. Available: http://jena.sourceforge.net/
Conference on Services Computing (SCC 2004)
(pp. 23-30). Kolodner, J. & Simpson, R. (1989). The MEDIA-
TOR: Analysis of an early case based problem-
Akkiraju, R., Farrell, J., Miller, J., Nagarajan,
solver. Cognitive Science, 13, 507-549.
M., Schmidt, M.-T., Sheth, A. et al. (2005). Web
Service Semantics - WSDL-S. In A joint UGA- Limthanmaphon, B. & Zhang, Y. (2003). Web
IBM Technical Note. service composition with case-based reasoning.
In Fourteenth Australasian database conference
Andrews, T., Curbera, F., Dholakia, H., Goland,
on Database technologies (pp. 201-208). Adelaide,
Y., Klein, J., Leymann, F. et al. (2003). Business
Australia.
Process Execution Language for Web Services
version 1.0. http://www-128.ibm.com/developer- Long, D. & Fox, M. (2002). Progress in AI
works/library/specification/ws-bpel/ [On-line]. Planning Research and Applications. Upgrade/
Novatica, 3, 10-25.
Ankolekar, A., Burstein, M., Lassila, O., Martin,
D., Mcllraith, S., Narayanan, S. et al. (2001). Madhusudan, T., Zhao, L. J., & Marshall, B.
DAML-S: Semantic Markup for Web Services. In (2004). A case-based reasoning framework for
International Semantic Web Working Symposium workflow model management. Data & Knowledge
(SWWS) (pp. 411-430). Engineering archive, 50, 87-115.

Ashley, K. D. & Rissland, E. L. (1988). A case- Maher, M. L. & Garza, A. G. d. S. (1997). Case-
based approach to modelling legal expertise. Based Reasoning in Design. IEEE Expert: Intelli-
IEEE Expert: Intelligent Systems and Their Ap- gent Systems and Their Applications, 12, 34-41.
plications, 3, 70-77.
Martin, D., PaolucciSycara, M., McIlraith, S.,
Cardoso, J. & Sheth, A. (2005). Adaptation and Burstein, M., McDermott, D., McGuinness, D. et
Workflow Management Systems. In International al. (2004a). Bringing Semantics to Web Services:
Conference WWW/Internet 2005 (pp. 356-364). The OWL-S Approach. In First International
Lisbon, Portugal. Workshop on Semantic Web Services and Web
Process Composition (SWSWPC 2004) San Diego,
Diaz, O. G. F., Salgado, R. S., Moreno, I. S., &
California, USA..
Ortiz, G. R. (2006). Searching and Selecting Web
Services Using Case Based Reasoning. In Com- Martin, D., Burstein, M., Hobbs, J., Lassila, O.,
putational Science and Its Applications (ICCSA McDermott, D., Mcllraith, S. et al. (2004b). Se-
2006) (pp. 50-57). mantic Markup for Web Services. http://www.
w3.org/Submission/OWL-S/ [On-line].

621
Utilisation of Case-Based Reasoning for Semantic Web Services Composition

Mcllraith, S. & Son, T. C. (2002). Adapting Golog Porter, B. W. & Bareiss, R. E. (1986). PROTOS:
for Composition of Semantic Web Services. In the An experiment in knowledge acquisition for heu-
Eighth International Conference on Knowledge ristic classification tasks. In First International
Representation and Reasoning (KR2002) (pp. Meeting on Advnces in Learning (IMAL) (pp.
482-493). 159-174). Les Arcs, France.
Nau, D., Cao, Y., & Lotem, A. (1999). SHOP: Remind (1992). ReMind Developer’s Reference
Simple Hierarchical Ordered Planner. In Interna- Manual. Cognitive systems.
tional Joint Conference on Artificial Intelligence
Russell, S. & Norvig, P. (2003). Artificial Intel-
(IJCAI-99) (pp. 968-973).
ligence: A Modern Approach. (2 ed.) Prentice
Nern, J., Agre, G., Atanasova, T., Marinova, Z., Hall, Pearson Education, Inc, Upper Saddle River,
Micsik, A., László, K. et al. (2006). INFRAWEBS New Jersey.
Semantic Web Service Development on the Base
Selman, B. (2000). Compute-Intensive Methods
of Knowledge Management Layer. International
for Artificial Intelligence. Annals of Mathematics
Journal “Information Theories and Applications,
and Artificial Intelligence, 28, 35-38.
13, 161-168.
Sirin, E., Hendler, J., & Parisa, B. (2003). Semi-
Osman, T., Thakker, D., & Al-Dabass, D. (2005).
automatic composition of web services using
Bridging the Gap between Workflow and Seman-
semantic descriptions. In Web Services: Model-
tic-based Web services Composition. In the 2005
ing, Architecture and Infrastructure workshop in
IEEE/WIC/ACM International Joint Conference
ICEIS 2003 Angers, France.
(pp. 13-23). Compiègne, France.
Watson, I. (1997). Applying Case-Based Reason-
Osman, T., Thakker, D., & Al-Dabass, D., Lazer,
ing: Techniques for Enterprise Systems. (1st ed.)
D., & Deleplanque, G. (2006). Semantic-Driven
Morgan Kaufmann Publishers.
Matchmaking of Web services using Case-Based
Reasoning. Accepted for the 2006 IEEE Interna- Watson, I. & Marir, F. (1994). Case-Based Rea-
tional Conference on Web Services (ICWS 2006), soning: A Review. The Knowledge Engineering
held on September 18-22, 2006, Chicago, USA. Review, 9(4), 1994, 9, 355-381.
Parsia, B. & Sirin, E. (2004). Pellet: An OWL DL Wu, D., Parisa, B., Hendler, J., Nau, D., & Si-
Reasoner. In the Third International Semantic Web rin, E. (2006). HTN Planning for Web Service
Conference (ISWC2004) Hiroshima, Japan. Composition Using SHOP2. Journal of Web
Semantics, 1, 1-30.
PORTAL (2003). Portal Ontology. http://www.
aktors.org/ontology/portal [On-line]. Zhang, R., Arpinar, B. I., & Aleman-Meza, B.
(2003). Automatic Composition of Semantic Web
Services. In International Conference on Web
Services (pp. 38-41). Las Vegas, Nevada, USA.

This work was previously published in International Journal of Intelligent Information Technologies, Vol. 5, Issue 1, edited by
J.N.D. Gupta, S.K. Sharma, and J. Hsu, pp. 24-42, copyright 2009 by IGI Publishing (an imprint of IGI Global).

622
623

Chapter 2.19
Rule Markup Languages and
Semantic Web Rule Languages
Adrian Paschke
Freie Universität Berlin, Germany

Harold Boley
National Research Council, Canada

Abstract for general Web rule languages to fulfill these


tasks. Characteristics of several important general
Rule markup languages will be the vehicle for using standardization or standards-proposing efforts for
rules on the Web and in other distributed systems. (XML-based) rule markup languages including
They allow publishing, deploying, executing and W3C RIF, RuleML, R2ML, SWRL as well as
communicating rules in a network. They may also (human-readable) Semantic Web rule languages
play the role of a lingua franca for exchanging rules such as TRIPLE, N3, Jena, and Prova are discussed
between different systems and tools. In a narrow with respect to these identified issues.
sense, a rule markup language is a concrete (XML-
based) rule syntax for the Web. In a broader sense,
it should have an abstract syntax as a common basis INTRODUCTION AND MOTIVATION
for defining various concrete languages addressing
different consumers. The main purposes of a rule Web rule languages provide the required expressive-
markup language are to permit the publication, ness enabling machine-interpretation, automated
interchange and reuse of rules. This chapter intro- processing and translation into other such Web
duces important requirements and design issues languages, some of which also being the execution
syntaxes of rule engines. One of these languages may
DOI: 10.4018/978-1-60566-402-6.ch001 act as a “lingua franca” to interchange rules and in-

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Rule Markup Languages and Semantic Web Rule Languages

tegrate with other markup languages, in particular design issues and characteristics of the introduced
with Web languages based on XML and with Se- rule languages. Section 4 presents future research
mantic Web languages (e.g. W3C’s RDF Schema, issues in Web rule language design. Section 5
OWL and its new OWL 2 version) for ontologies concludes this chapter with a summary.
serialized in RDF/XML or directly in XML. Web
rule languages may also be used for publication
purposes on the Web and for the serialization of WEB RULE LANGUAGES
external data sources, e.g. of native online XML
databases or RDF stores. Recently, there have Rule markup (serialization) languages have been
been several efforts aiming at rule interchange and developed for the Web-based interchange of, e.g.,
building a general, practical, and deployable rule privacy policies, business rules, and - as focused
markup standard for the (Semantic) Web. These here - Semantic Web rules. Rules are central to
encompass several important general standard- knowledge representation for the Semantic Web
ization or standards-proposing efforts including (Boley, 2007), hence are increasingly considered
RuleML (www.ruleml.org), SWRL (www.w3.org/ as being side by side with ontologies, e.g. in W3C’s
Submission/SWRL/), SWSL (http://www.w3.org/ layered Semantic Web architecture (2007 version
Submission/SWSF-SWSL/), R2ML (oxygen. shown in Figure 1).
informatik.tu-cottbus.de/rewerse-i1/?q=R2ML), Rule interchange in an open format is important
RIF (www.w3.org/2005/rules/), and others such for all higher Semantic Web layers, including a
as XCL (http://www.altheim.com/specs/xcl/1.0/), Web of Trust and, generally, a Pragmatic Web
designed as a concrete (serialization) syntax for (Paschke et al, 2007), and is crucial for applications
ISO’s Common Logic (CL) standard. in eBusiness, eGovernment, eHealth, etc. This
In this chapter, a system of general require- section introduces major rule markup languages
ments and design choices for Web rule languages including RuleML, R2ML, and RIF, as well as
will be introduced and instantiations discussed human-readable Semantic Web rule languages
in the context of the current prominent general such as TRIPLE and N3, and platform-specific
Rule Markup Languages and Semantic Web rule rule engine languages such as Jena and Prova.
languages. This chapter is intended to be of help
to a wide audience. In particular, it is targeted to Rule Markup Languages
rule practitioners who want to serialize the declara-
tive rules of their applications in a general rule We characterize rule languages as rule markup
markup language, and publish and interchange languages if they are serialized in XML, employ
them on the Web. Rule practitioners will find URIs/IRIs for constants etc., and can interface
here a discussion of general design criteria with with Web ontology languages.
examples from the current rule markup languages.
These examples, together with a discussion of RuleML
advantages and drawbacks, will offer guidance to
readers when declaratively representing their own The Rule Markup Language (RuleML, www.
rule-based applications in a Web rule language. ruleml.org) is a markup language developed to
The structure of the rest of this chapter is as fol- express a family of Web rules in XML for deduc-
lows: Section 2 introduces current rule markup tion, rewriting, and reaction, as well as further
languages and rule interchange formats as well as inferential, transformational, and behavioral tasks.
Semantic Web rule languages. Section 3 comprises It is defined by the Rule Markup Initiative (www.
the main part of this chapter, discussing important ruleml.org), an open network of individuals and

624
Rule Markup Languages and Semantic Web Rule Languages

Figure 1. Semantic Web Layer Cake [adapted


Logic, has subbranches with Negation-As-Failure,
from (W3C, 2007)]
strong-Negation, or combined languages, and is
parameterized by ‘pluggable’ built-ins.

SWRL

The Semantic Web Rule Language (SWRL,


www.w3.org/Submission/SWRL/) is defined as
a language combining sublanguages of the OWL
Web Ontology Language (OWL DL and Lite)
with those of the Rule Markup Language (Unary/
Binary Datalog).
The specification was submitted to W3C in
May 2004 by the National Research Council of
Canada, Network Inference (since acquired by
webMethods), and Stanford University in associa-
tion with the Joint US/EU ad hoc Agent Markup
Language Committee.
groups from both industry and academia that was Compared to Description Logic Programs
formed to develop a canonical Web language (DLP), a slightly earlier proposal for integrating
for rules using XML markup and transforma- description logic and Horn rule formalisms by
tions from and to other rule standards/systems. an overlapping authoring team, SWRL takes the
It develops a modular, hierarchical specification opposite integration approach: DLP can be seen
for different types of rules comprising facts, as the ‘intersection’ of description logic and Horn
queries, derivation rules, integrity constraints logic; SWRL, as roughly their ‘union’. For DLP,
(consistency-maintenance rules), production rules, the resulting rather inexpressive language cor-
and reaction rules (Reaction RuleML, http://ibis. responds to a peculiar looking description logic
in.tum.de/research/ReactionRuleML), as well as imitating special rules. It is hard to see the DLP
tools and transformations from and to other rule restrictions, which stem from Lloyd-Topor trans-
standards/systems. formations, being either natural or satisfying. On
Datalog RuleML is defined over both data the other hand, SWRL retains the full power of
constants and individual constants with an optional OWL DL, but adds rules at the price of undecid-
attribute for IRI (URI) webizing . Atomic formulas ability and a lack of complete implementations,
have n arguments, which can be positional terms although the SWRL Tab of Protégé has become
or, in Object-Oriented Datalog, slots (F-logic- quite popular (http://protege.cim3.net/cgi-bin/
like key->term pairs); OO Datalog also adds wiki.pl?SWRLTab).
optional types and RDF-like oids/anchors, via Rules in SWRL are of the form of an implication
IRIs (Boley, 2003). Inheriting all of these Datalog between an antecedent (body) conjunction and a
features, Hornlog RuleML adds positional or slot- consequent (head) conjunction, where description
ted functional expressions as terms. In Hornlog logic expressions can occur on both sides. The
with equality, such uninterpreted (constructor- intended interpretation is as in classical first-order
like) functions are complemented by interpreted logic: whenever the conditions specified in the
(equation-defined) functions. This derivation rule antecedent hold, then the conditions specified in
branch is extended upward towards First Order the consequent must also hold.

625
Rule Markup Languages and Semantic Web Rule Languages

R2ML system of built-ins drawn from Datatypes and


Built-Ins (RIF-DTB). The connection to other
R2ML (http://oxygen.informatik.tu-cottbus. W3C Semantic Web languages is established
de/rewerse-i1/?q=R2ML) was developed as a via RDF and OWL Compatibility (RIF-SWC).
subproject in the EU Network of Excellence Moreover, RIF-BLD is a general Web language
REWERSE (http://oxygen.informatik.tu-cottbus. in that it supports the use of IRIs (International-
de/rewerse-i1/). The R2ML project is about the ized Resource Identifiers) and XML Schema data
design of integrity and derivation rules on the types. The RIF Working Group has also defined
basis of the Rule Markup Language (RuleML) the Framework for Logic Dialects (RIF-FLD),
and the Semantic Web Rule Language (SWRL). of which RIF-BLD was shown to be the first
R2ML defines a general markup framework for instantiation. RIF-FLD uses a uniform notion of
integrity rules, derivation rules, production rules terms for both expressions and atoms in a Hilog-
and reaction rules. Rule concepts are defined like manner.
with the help of MOF/UML, a subset of the UML Current efforts of the RIF Working Group are
class modeling language proposed by the Object expected to introduce a Core (RIF-Core) in the
Management Group (OMG) for the purpose of intersection of RIF-BLD and a new Production
‘meta-modeling’, i.e. for defining languages con- Rule Dialect (RIF-PRD) influenced by OMG’s
ceptually on the level of an abstract (semi-visual) PRR, which can then be further extended or
syntax. From these MOF/UML language models supplemented by reaction rules.
concrete markup syntax is obtained by applying a
mapping procedure for generating corresponding Semantic Web Rule Languages
languages from parameterized schemas.
In contrast to the XML-based rule markup lan-
W3C RIF guages in the previous section, the Semantic
Web rule languages described in this section are
The W3C Rule Interchange Format (RIF) Work- human-readable rule languages, using an ASCII
ing Group (http://www.w3.org/2005/rules/wiki/ syntax based, e.g., on the ISO Prolog syntax
RIF_Working_Group) is an effort, influenced by standard. Typically, they are designed as compact
RuleML, to define a standard Rule Interchange presentation languages for human consumption.
Format for facilitating the exchange of rule sets While they may be serialized in an XML-based
among different systems and to facilitate the de- rule markup language such as RuleML or RIF,
velopment of intelligent rule-based application for e.g. for interchange purposes, they can also be
the Semantic Web. For these purposes, RIF Use employed directly: dynamically interpreted by
Cases and Requirements (RIF-UCR) have been platform-specific rule engines (at runtime) or
developed. The RIF architecture is conceived as a statically translated into executable code (at
family of languages, called dialects. A RIF dialect compile time).
is a rule-based language with an XML syntax and
a well-defined semantics. TRIPLE
So far, the RIF working group has defined the
Basic Logic Dialect (RIF-BLD), which semanti- TRIPLE (http://triple.semanticweb.org/) was de-
cally corresponds to a Horn rule language with signed as a practical rule language for linked-data
equality. RIF-BLD has a number of syntactic applications. It is an RDF query, inference, and
extensions with respect to ‘regular’ Horn rules, transformation language for the Semantic Web
including F-logic-like frames, and a standard extending F-logic with modules. TRIPLE rules

626
Rule Markup Languages and Semantic Web Rule Languages

have been used to implement RDFS and other advantages of Prova is its separation of logic,
schema languages. data access, and computation as well as its tight
integration of Java, Semantic Web technologies
N3 / Turtle and enterprise service-oriented computing and
complex event processing technologies.
Notation3 (w3.org/TeamSubmission/n3/), more
commonly known as N3, is a shorthand non-XML
serialization of Resource Description Framework DESIGN AND CHARACTERISTICS
(RDF) models, designed with human readability in OF WEB RULE LANGUAGES
mind: N3 is much more compact and readable than
RDF/XML serializations. N3 has several features General requirements that need to be addressed
that go beyond the serialization of RDF models, by a rule markup language include semantic
such as support for RDF-based rules. Support- expressiveness and clarity, computational effi-
ing the triple pattern syntax of SPARQL, Turtle ciency and Web scalability, machine-readable and
(w3.org/TeamSubmission/turtle/) is a simplified, machine-interpretable syntaxes, usability by both
RDF-only subset of N3. human users and automated agents, compact rep-
resentation, interchangeability with other formats,
Jena Rules means for serialization and persistence, as well
as tool support in authoring, parsing/generating,
The default representation format in Jena (jena. and verifying rules. An important property that
sourceforge.net/) for a rule in the rule-based rea- refers to development-time software engineering
soner is a Java Rule object with a list of body terms quality is the extensibility of the language and its
(premises), a list of head terms (conclusions) and interoperability with other representation formats.
an optional name and an optional direction. How- In this section, general language design principles,
ever, in Jena2 a rather simple parser is included together with a selection of four important issues
which allows rules to be specified in reasonably and criteria for rule markup language design, are
compact form in text source files. identified and characteristics of the current rule
markup languages RIF, RuleML, R2ML, SWRL
Prova (DAML Rules) as well as specific Semantic Web
rule languages are exemplified with respect to
Prova (http://www.prova.ws/) is both a (Seman- them. Further design issues and requirements for
tic) Web rule language and a highly expressive Web rule languages have been elaborated in, e.g.,
distributed (Semantic) Web rule engine which (Wagner et al, 2005), (Bry and Marchiori, 2005),
supports complex reaction rule-based workflows, (Boley, 2007), and (Paschke, 2007).
rule-based complex event processing, distributed
inference services, rule interchange, rule-based Language Design Principles
decision logic, dynamic access to external data
sources, Web Services, and Java APIs. Prova Given the large design space of rule languages
follows the spirit and design principles of the and rule concepts, the specification of a rule
W3C Semantic Web initiative and combines markup language is a difficult integration and
declarative rules, ontologies and inference with conceptualization challenge that calls for bal-
dynamic object-oriented programming and access ancing many (interrelated) design choices with
to external data sources via query languages such respect to semantics, syntax, and pragmatics.
as SQL, SPARQL, and XQuery. One of the key In this subsection, we will raise four (markup)

627
Rule Markup Languages and Semantic Web Rule Languages

language design principles and will illustrate the preserving syntax with many highly specialized
actual design choices of the current rule markup constructs. For instance, variables in R2ML are
languages with examples. provided in the form of ObjectVariable (i.e.
variables that stand for objects), DataVariable
1. Criteria of Good Language Design (i.e. variables that stand for data literals), and
GenericVariable (i.e. variables that do not have
Rule markup language should be clear, compact, a type), whereas RuleML (as well as SWRL and
precise and easily adaptable. They should strive RIF) only provide a generic Var construct. Like
to fulfill typical criteria for good language design RuleML, R2ML defines the notion of an individual
(Codd, 1971) - as known from logic, databases (constant) and distinguishes between objects and
and programming - such as minimality, referential data with the notions of an object name and data
transparency and orthogonality: value.
The main design goal of the specific Seman-
• Minimality requires that the language pro- tic Web rule languages such as TRIPLE, Jena,
vides only a small set of needed language and (following ISO Prolog syntax) Prova is to
constructs, i.e., the same meaning cannot be provide a terse scripting syntax with a minimal
expressed by different language constructs set of needed constructs. RuleML’s POSL syntax
• Referential transparency is fulfilled if the combines and extends the terse ISO Prolog and
same language construct always expresses F-logic syntaxes.
the same semantics regardless of the con-
text in which it is used 2. Different Syntactic and
• Orthogonality asks for pairwise indepen- Semantic Layers
dent language constructs, thus permitting
their meaningful systematic combination A complete specification of Web rule languages
consists of a formalization of their syntax, se-
The RuleML family follows these design mantics and, often left implicit, pragmatics. As
principles as far as possible and provides only a implied by their name, the syntax of markup
set of needed language constructs which can be languages always includes the concrete syntax of
applied in every meaningful combination. This (XML) markup, perhaps indirectly through other
leads to a compact homogeneous syntax which is languages such as via RDF/XML. Often, there
easier to maintain, learn, read and understand by is another more or less concrete syntax such as a
end users, as well as easy to process automatically compact shorthand or presentation syntax, which
by machines (e.g. translators). may be parsed into the XML markup. While a
SWRL and RIF, which build on RuleML, ba- presentation syntax can already disregard certain
sically follow this compact minimalistic design details, an abstract syntax systematically replaces
approach. However, SWRL introduces a more fine- character sequences with abstract constructors,
grained distinction of constructs than RuleML, often in a (UML) diagram form or as an abstract
e.g. of Atoms into various types of specialized syntax tree (AST). Together with different token
atoms such as classAtom, datarangeAtom, and dictionaries, it can be used to generate correspond-
invidiualPropertyAtom, which can be formed ing concrete syntaxes. The semantics is formalized
from unary predicates (classes), binary predicates in a model-theoretic, proof-theoretic, or procedural
(properties), and equalities or inequalities. manner, sometimes in more than one. When rules
R2ML introduces further differentiated types and speech-act-like performatives, such as queries
of terms and atoms. This leads to a rich structure- and answers, are transmitted between different

628
Rule Markup Languages and Semantic Web Rule Languages

systems, their pragmatic interpretation, including are directly written in the concrete SWRL XML
their pragmatic context, becomes relevant, e.g. syntax. RIF, being a rather new standard under
in order to explain the effects of performatives - development, currently does not provide any such
such as the assertion or retraction of facts - on the tool support.
internal knowledge base (Paschke et al, 2007). The PIM level should enable platform-
A general distinction of three modeling lay- independent machine interpretation, processing,
ers can be adopted from OMG’s model driven interchange and translation into multiple PSM
architecture (MDA) engineering approach (http:// execution syntaxes of concrete rule engines.
www.omg.org/mda/): Hence, the concrete XML (or RDF/XML-based)
syntax of a Web rule language such as RuleML,
• A platform specific model (PSM) which SWRL or R2ML resides on this level, whereas
encodes the rule statements in the language the abstract syntax is on the borderline between
of a specific execution environment the PIM and CIM levels. The abstract syntax
• A platform independent model (PIM) can be defined, e.g., with the help of a suitably
which represents the rules in a common general grammar definition language such as the
(standardized) interchange format, a rule EBNF formalism, used, e.g., in the definition of
markup language the abstract syntax of OWL, RuleML, RIF, and
• A computational independent model (CIM) SWRL, or with the help of a MOF/UML model,
with rules represented in a natural or visual as, e.g., in PRR, R2ML, and RuleML. (Wagner
language et al, 2004) (Giurca and Wagner, 2005).
The PSM level is the result of translating/
The CIM level comprises visual and verbal mapping PIM rule (interchange) languages into
rendering and rule modeling, e.g. via graphical execution syntaxes which can be directly used
representation or a controlled natural language in a specific execution environment such as a
syntax for rules, mainly intended for human con- rule engine. A general distinction can be made
sumption. Graphical representations such as UML between a compiled language approach, where
diagrams or template-driven/controlled languages the rules are statically translated into byte code
can also be used as presentation languages. (at compile time), as e.g. done in the rule engines
In order to facilitate rule modeling, R2ML Take (http://code.google.com/p/take/) and Drools
provides a UML-based Rule Modeling Language (www.jboss.org/drools/) versus interpreted script-
(URML) (Lukichev and Wagner, 2006) which ing languages, which are dynamically interpreted
allows visual rule modeling based on UML class (at run-time), as e.g. in the rule engines Prova
models and OCL constraints. RuleML on the CIM (Paschke, 2006b) and OO jDREW (Ball et al.,
level provides several tools that use a controlled 2005). While the compiled approach has obvi-
natural rule language approach. Among them are ous efficiency benefits, the interpreted approach
TRANSLATOR (Hirtle, 2006), which is based on is more dynamic and facilitates, e.g., updates at
Attempto Controlled English (ACE) (Fuchs et al, run-time. Often, Semantic Web Rule Languages
2006), the open source Reaction RuleML editor are directly executable by their respective rule
(http://ibis.in.tum.de/research/ReactionRuleML/ engines; hence reside on the PSM level. As an
index.htm#editor), which uses a template driven intermediate step between the concrete PSM level
approach, and the commercial RuleManager and the PIM level an abstract representation is
(Ensig, 2007). The Protégé tool (http://protege. often introduced, such as N3, which provides an
stanford.edu/) provides facilities for modeling abstract rule syntax based on the RDF syntax, or
SWRL rules, but only on the PIM level, i.e. rules POSL and Prova, which both provide ANTLR

629
Rule Markup Languages and Semantic Web Rule Languages

grammars (http://www.jdrew.org/oojdrew/demo/ which the interchange takes place is important


translator, http://www.prova.ws/gram.html) which for the target environment, in order to know how
are transformed into ASTs as the basis for further the received information should be used and
translation into interchange markup languages which actions should be taken with respect to
such as RuleML or other, specific execution the pragmatic aspects. A standard nomenclature
formats. of pragmatic performatives is defined by the
The correct execution of an interchanged Knowledge Query and Manipulation Language
PIM-level rule set serialized in a rule markup (KQML) (www.cs.umbc.edu/kqml/) and the FIPA
language depends on the semantics of both the rule Agent Communication Language (ACL) (FIPA,
program and the platform-specific rule inference 2000), which define several speech-act-theory-
engine (IE). To address this issue, the IE and the based communicative acts.
interchanged rule set must reveal their intended/
implemented semantics. This may be solved via 3. Modular Specialized Schema
explicit annotations based on a common vo- Layers vs. Flat General Schema
cabulary, e.g. an (Semantic Web) ontology which
classifies the semantics. Annotations describing There are two basic design principles for the
the semantics of an interchanged rule set could concrete rule markup syntax. The language (or
even be used to find appropriate IEs on the Web language family) may be implemented in one
to correctly and efficiently interpret and execute flat (monolithic) general XML schema or in a
the rule program; for example, (1) by configuring layered structure, where semantically related
the rule engine for a particular semantics in case constructs are defined within separate modules
it supports different ones, (2) by executing an ap- that are added to the different language layers of
plicable variant of several interchanged semantic the Web rule language (cf. http://www.ruleml.
alternatives of the rule program, or (3) by automatic org/modularization/) . This leads to a hierarchi-
transformation approaches which transform the cal structure where higher language layers build
interchanged rule program into a rule program on sublayers and add more expressiveness by
with an applicable semantics; cf. XTAN (http:// extending them. The layers are not necessarily
www.w3.org/2008/02/xtan/). Another approach organized around expressiveness/efficiency to
is to specify additional meta test cases for testing the language core.
typical properties of well-known semantics, where R2ML follows the first approach and pro-
by the combination of succeeded and failed meta vides one quite large, flat XML schema for all
tests the unknown semantics of an IE can often be different rule types and language constructs. In
uniquely determined (Paschke, 2006). contrast, RuleML (also SWRL and RIF) follow
We remark that, traditionally, rule-based the layered design principle and define new con-
systems have been supported by two types of structs within separate modules which are added
inferencing algorithms: forward-chaining and to the respective layers in the RuleML language
backward-chaining. A general rule markup lan- family. The layered and uniform design makes it
guage, as a lingua franca, should support transla- easier to learn the language and to understand the
tion and interpretation of both reasoning directions, relationship between the different features, and it
perhaps again using pragmatic annotations (where provides a certain guidance to users who might
by default chaining should be bidirectional, as be interested only in a particular subset of the
with the direction attribute in RuleML). features and who do not need support for the full
Independently from the semantics of an inter- expressiveness of the language. The modulariza-
changed rule program, the pragmatic context in tion allows for easy extension of the language’s

630
Rule Markup Languages and Semantic Web Rule Languages

representation capabilities, using the extensibil- • If the information is required to be in a


ity of XML Schema (e.g. a redefine of an XML standard DTD-like attribute type such as
Schema group definition), without breaking the ID, IDREF, or ENTITY, use an attribute.
core language standard. This development path • If the information should not be normal-
provides a stable, useful, and implementable lan- ized for white space, use elements. (XML
guage design for rule developers to manage the processors normalize attributes in ways
rapid pace of change on the Semantic Web and that can change the raw text of the attribute
modern rule systems. Apart from that, modules value.) (cf. http://www.ibm.com/develop-
facilitate the practical and extensible development erworks/xml/library/x-eleatt.html)
of a rule language family by bundling language
constructs into layers which can be developed, Accordingly, RuleML’s general markup
compiled, tested and managed separately. The conventions provide common principles for its
modularization also enforces the principle of language hierarchy. XML elements are used for
information hiding and can provide a basis for representing language constructs as trees while
data abstraction. However, a monolithic schema XML attributes are used for distinguishing
is easier to read by humans than an unevenly variations of a given element and, as in RDF,
modularized one and, by now, some of the extant for webizing. Variation can thus be achieved by
XML processing tools and editors do not fully different attribute values rather than requiring
support modular XML Schema definitions. This different elements. Since the same attribute can
calls for flattening a layered schema on demand occur in different elements, an orthogonal, two-
via automatic modular-to-monolithic translators, dimensional classification ensues, which has the
thus combining the advantages of modular de- potential of quadratic tag reduction.
velopment and maintenance with the advantages For example, recent work in RuleML led to
of monolithic delivery (for some validators and orthogonal dimensions extending the RuleML 0.9
(object model) transformers, e.g. based on JAXB role tags for argument, <arg . . .>, and slots, <slot>.
https://jaxb.dev.java.net/). So far, the unkeyed <arg index=”...”> was always
ordered, as indicated by the index attribute, and
4. XML Elements vs. Attributes the keyed <slot> was always unordered, as indi-
cated by the lack of an index attribute. This was
A general question regarding the implementation generalized by allowing an optional index attribute
of a concrete rule markup language is where to use for both role tags, as shown by the independent
XML elements and where attributes to define the distinctions in the following key-order matrix:
rule constructs and the rule information content. A
general discussion on this element-vs.-attribute is-
sue can be found in the OASIS Cover pages (http:// ordered unordered
xml.coverpages.org/elementsAndAttrs.html): keyed <slot index=”...”> <slot>
unkeyed <arg index=”...”> <arg>
• If the information in question could be it-
self marked up with elements, put it in an Two extra orthogonal combinations are ob-
element. tained from this system. First, keyed, ordered
• If the information is suitable for attribute children permit positionalized slots, as in this cost
form, but could end up as multiple attributes fact (see Figure 2).
of the same name on the same element, use Here, slot names item, price, and taxes are
child elements instead. provided, e.g. for readability, as well as index

631
Rule Markup Languages and Semantic Web Rule Languages

Figure 2.

positions 1-3, e.g. for efficiency. Expressive Layering


Second, unkeyed, unordered children permit
elements acting like those in a bag (finite multiset), From the perspective of knowledge representation,
as in this transport fact: the main adequacy criterion for a rule markup
language is its epistemological adequacy, which
addresses the ability of the language to represent
<Atom> all relevant knowledge under consideration.
<Rel>transport</Rel> Among other representation issues, a general rule
<arg><Ind>chair</Ind></arg> interchange format should allow to coherently
<arg><Ind>chair</Ind></arg> represent derivation rules, reaction rules, integrity
<arg><Ind>table</Ind></arg> rules, and deontic rules in a homogeneous syntax
</Atom> (Wagner et al, 2005). We use the following general
rule classification:
Here, the arguments are specified to be com-
mutative and ‘non-idempotent’ (duplicates are • Facts may comprise various kinds of infor-
kept). mation such as asserted atoms (formulas),
For a general discussion of positional vs. un- individual-class memberships (of ontology
ordered representations see (Boley, 2006). classes), (object-oriented) instances, stored
R2ML differs from RuleML, SWRL and RIF data (e.g., relational, XML), states and
as it implements an attribute solution and defines event occurrences which might be quali-
user information content in attributes. For instance fied, e.g., by priorities, temporally, etc.
a typed object variable “?driver” is represented • Derivation rules infer conclusions from
as follows: conditions (as in Datalog and Horn logic),
where facts (see above) are a special case
with constantly true conditions
<r2ml:ObjectVariable • Transformation rules specify term rewrit-
r2ml:name=”driver” ing, which can be considered as derivation
r2ml:classID=”userv:Driver” /> rules of logics with (oriented) equality
• Integrity rules (or integrity constraints)
A distinction of positional vs. slotted (named- are assertions which express conditions
argument) predicates and functions, as in RuleML (or queries) that must always be satisfied.
and RIF, does not exist in R2ML. Besides enforcing data integrity, they can
constrain, e.g., the rule system structure, its
information content, or its behavior:

632
Rule Markup Languages and Semantic Web Rule Languages

Figure 3. RuleML rule language family

◦◦ Structural constraints (deontic expressive languages; e.g., a sublanguage of con-


assignments) ditions is shared by derivation rules and reaction
◦◦ State constraints rules. A basic classification of rule languages is
◦◦ Process constraints introducted by the RuleML family with Deriva-
• Deontic rules describe rights and obliga- tion Rule Markup Languages (that is, Derivation
tions, e.g., of institutions and agents in the RuleML), Production Rule Markup Languages
context of evolving states (situations trig- (that is, Production RuleML which is a subfamily
gered by events/actions) and state transi- of Reaction RuleML), Reaction RuleML Markup
tions, where integrity rules (see above) are Languages (that is, Reaction RuleML, http://ibis.
a special case (‘introspectively’) affecting in.tum.de/research/ReactionRuleML/), and Trans-
the rule set itself formation Markup Languages (that is, Functional
• Reaction rules are (behavioral / action) RuleML, http://www.ruleml.org/fun/), as well as
rules that react on occurred events (exter- other specializations, e.g. dialects for deontic rules
nal events or changed conditions) by ex- (e.g. covered by the RBSLA language (Paschke,
ecuting actions, where production rules 2005)), defeasible rules (Defeasible RuleML,
are a special case with events restricted to http://defeasible.org/RuleML/) and uncertainty
changed conditions / fuzziness (Fuzzy RuleML, http://www.image.
ntua.gr/FuzzyRuleML/).
Because of this broad variety of rules relevant The coverage of rule markup languages can
to the Semantic Web, a general rule markup lan- be roughly divided into logic-based rule formal-
guage such as RuleML should have a hierarchical isms, usually variants of first-order predicate
structure that reflects the relevant rule dialects logic, and not logic-based rule formalisms such
or sublanguages, covering the knowledge repre- as (early) production rule systems. In this chapter,
sentation needs of various subcommunities (cf. we mainly focus on logic-based derivation rule
Figures 3 and 4). markup languages, but most of the general rule
The main branches contain subbranches, e.g. markup languages such as RuleML and R2ML
derivation rules monotonically disallowing or also support the serialization of production rules
non-monotonically allowing negation as failure and reaction rules (e.g., Reaction RuleML cov-
(NAF). Branches also exhibit a layering structure ers different types of reaction rule languages).
with sublanguages of different expressive power, In fact, RuleML and R2ML provide a roughly
e.g. for monotonic derivation rules proceeding similar coverage, whereas, e.g., SWRL acts as a
from function-free Horn logic (Datalog) through more specialized language for homogeneously
either full Horn logic or disjunctive Datalog up to combining Datalog rules with OWL, hence does
first-order logic (FOL). This hierarchy does not not cover, e.g., reaction rules.
form a strict tree (but a directed acyclic graph),
i.e. some sublanguages are shared by several more

633
Rule Markup Languages and Semantic Web Rule Languages

Figure 4. RuleML 0.91 derivation RuleML subfamily

Rules and Object Descriptions sociated URIs with constant symbols, predicate
names, and other language constructs for reference
With its URIs (http://www.w3.org/Addressing/), and disambiguation. For example, the constant
the Web provides a global addressing (URL) and symbol Georgia could be associated with the unique
naming (URN) mechanism for objects. A URI ‘homepage’ URI http://www.georgia.gov to refer
consists of a URI scheme name (http:, file:, etc.) to and disambiguate the state in the Southeastern
followed by other information that is interpreted U.S. in contrast to other entities having the same
relative to the URI scheme. The method for as- English name such as the country at the east coast
signing meanings to names varies from one URI of the Black Sea.
scheme to the next, and within each scheme for There have been attempts to differentiate
different sets of names. Each scheme’s specification the Web notion of URIs into two subnotions, as
describes how its URIs are intended to be used in discussed in (Halpin, 2006): URLs (Uniform Re-
certain contexts. As a result, any naming framework source Locators), for access, and URNs (Uniform
must provide mechanisms to enable the creation of Resource Names), for naming. This distinction is
new names while avoiding conflicts with existing independent from the recent IRI (Internationalized)
ones. URIs are also central to the Semantic Web, versions of URIs. In the context of Web knowledge
where RDF metadata are used to describe those representation, especially for Web rules as explored
objects or resources with classes and properties, in POSL, RuleML, and RIF, three central URI uses
which are themselves defined by ontologies (in are emerging (Boley, 2007), given here in the order
RDF Schema or OWL). Since SHOE (Heflin et of further needed research (orthogonal to research
al, 1999), Semantic Web rule languages have as- in URI normalization (Boley, 2003)).

634
Rule Markup Languages and Semantic Web Rule Languages

First, a URI can be used, URL/access-style, dereferencing (‘clicking’) will then retrieve Pluto’s
for module import (transitive import for nested Wikipedia entry.
modules), where it is an error if dereferencing the Third, for certain formal purposes a URI like
URI does not yield a knowledge base valid with http://termeg.org#MiniVan is needed just to pro-
respect to the expected representation language. vide a name; for other formal purposes, also to pro-
Second, a URI can be used, URN/naming-style, vide a total or partial definition found by using that
as the identifier of an individual constant in the same URI access-style (say, the partial definition of
representation language, where URI dereferencing being rdfs:subClassOf both http://termeg.org#Van
is not intended as part of the formal knowledge and http://termeg.org#PassengerVehicle).
representation. If dereferencing is attempted as In most rule markup languages as well as the
part of the metadata about the informal knowledge specific Semantic Web rule languages, the (user-
representation, it should retrieve a descriptive defined) vocabulary names are globally unique
‘homepage’ about the individual. standard identifiers in the form of URI refer-
Third, a URI can be used, naming-style, as the ences. Morover, they often define specific builtins
identifier of a class, property, relation, or function, for handling URIs such as the SWRL builtins
and at the same time, access-style, where derefer- swrlb:resolveURI (from XQuery op:resolve-uri),
encing yields (a “#”-anchor into) a knowledge base which is satisfied iff the URI reference in the first
formally defining that identifier (albeit perhaps argument is equal to the value of the URI reference
partially only, as for an RDF Schema knowledge in the second argument resolved relative to the
base just giving the superclasses of a class). base URI in the third argument, or swrlb:anyURI,
Here are examples for the three URI uses in which is satisfied iff the first argument is a URI
connection with rules. reference consisting of the scheme in the second
First, a module of U.S. states could be imported argument, host in the third argument, port in
into the current rulebase using the URL/access- the fourth argument, path in the fifth argument,
style URI http://modeg.org#us-state. query in the sixth argument, and fragment in the
Second, the URI http://en.wikipedia.org/wiki/ seventh argument.
Pluto can be used URN/naming-style to refer to All classes in R2ML are URI references. A
a celestial body originally considered a planet, class is a type entity for R2ML objects and ob-
as in this rule specifying its years of planethood ject variables. Similarily, a reference property as
(a URI is enclosed in a pair of angular brackets, well as a datatype predicate in R2ML is a URI
<... >): reference.
RIF uses internationalized resource identifiers
or IRIs (symbol space rif:iri) as constants similar
planet(<http://en.wikipedia.org/ to RDF resources.
wiki/Pluto>,AD[?year]):-
lessThanOrEqual(1930,?year), Rule-Ontology Combination
lessThanOrEqual(?year,2006).
A rule markup language should be reasonably
As part of the formal rule knowledge, the Pluto integrated with the Semantic Web and should be
URI is employed only for naming. The rule can able to refer to external Semantic Web vocabu-
also be employed as metadata about informal laries by means of URIs or IRIs, e.g. to use their
knowledge through (‘semantic search engine’) taxonomic vocabularies as type systems and their
queries like planet(?which,2005), because one of individuals as external constants/objects. Domain-
its solutions will bind ?which to the URI, whose independent rules can then be interpreted (relative

635
Rule Markup Languages and Semantic Web Rule Languages

to each vocabulary) in a domain-dependent man- A terminological predicate is not permitted in the


ner (with a precise semantics). Accordingly, the head of a rule. Intuitively, terminological classes
original rule set can be much easier interchanged cannot be (re)defined by assertional clauses,
and managed/maintained in a distributed envi- because a terminology establishes more stable
ronment. Also, the core Web rule language stays ‘background’ knowledge extended by asser-
compact and can be easily extended for different tions that constitute more volatile ‘foreground’
vocabulary languages (RDFS, OWL, OWL 2, knowledge.
etc.) on a “per-need-basis”. Such a hybrid lower layer can use sort restric-
In recent years, quite an effort has been made tions as simple terminological queries in Datalog
to develop a dual expressiveness layering of as- rule bodies, which in higher layers are extended
sertional and terminological knowledge as well to terminological queries involving properties,
as their blends (Antoniou et al., 2005, Kifer et al, ALC expressions, etc. In the spirit of (Kifer et al,
2005). To retain decidability of querying, the as- 2005), this should lead to a more realistic Seman-
sertional bottom layer usually consists of Datalog tic Web architecture with simplified foundations
(function-free) assertions, perhaps restricted to and better computational properties. Our fine-
unary/binary predicates. For the terminologi- grained bottom-up approach also complements
cal bottom layer, an irreflexive version of RDF the recent differentiation of OWL-Lite into OWL
Schema’s subClassOf can be employed, which 1.1 (later: OWL 2) Tractable Fragments (Grau et
could later be extended towards the rhoDF (Munoz al., 2006).
et al., 2007) fragment of RDF. The two layers can The following example uses classes of a sub-
be blended through a hybrid combination (rhoDF ClassOf terminology as variable sorts (types) of
classes used as types for Datalog constants and slightly extended Datalog rules, namely of Horn
variables, and subClassOf defined with order- logic rules employing (unary) functions only for
sorted semantics) or a homogeneous integration measurement units.
(rhoDF classes used as unary predicates in the The terminology forms a DAG (cf. Figure 5)
body of Datalog rules, and subClassOf defined as that introduces Vehicle-rooted classes and ex-
special rules with Herbrand-model semantics). emplifies multiple inheritance of MiniVan from
The higher layers can develop Datalog into Van and PassengerVehicle (the “>” infix is used
Horn (as in Prova’s or OO jDREW’s hybrid between a superclass and a subclass):
implementation) and FOL (First-Order Logic)
assertions, rhoDF into ALC and SHIQ terminolo-
gies with classes and properties, and appropriate Vehicle > Van
blends (Rosati, 2006) (Rosati, 2006a), e.g. as Vehicle > PassengerVehicle
advancements of our hybrid DatalogDL (Mei et Van > MiniVan
al, 2007b) or homogeneous ALCuP (Mei et al, PassengerVehicle > MiniVan
2007). For certain purposes, especially in the early PassengerVehicle > Car
modeling phases, the assertional layers can move
even beyond FOL, including towards higher-order The following RuleML/POSL rules specify
and modal logics, as started as part of the RuleML registration fees for vehicles. The first rule speci-
family (Boley, 2006). fies a vehicle variable typed by the Van class, while
To permit the specification of terminologies the second refers to the Car class (the “:” infix is
independent of assertions, a hybrid approach is used between a variable and its type):
proposed here adopting the CARIN (Levy and
Rousset, 1998) principle as a working hypothesis:

636
Rule Markup Languages and Semantic Web Rule Languages

Figure 5. A vehicle terminology (cf. (W3C, 2000))

set. R2ML distinguishes between plain and


registration(?V:Van,CAD[?R:Deci typed literals and typed and untyped variables.
mal]):- A DataTerm is a DataVariable, a DataLiteral,
emission(?V,CO2[?E]), or a DataFunctionTerm. An ObjectVariable is a
weight(?V,kg[?W]), variable that stand for objects of a particular class
emiweight(CAD[?R],CO2[?E],kg[?W type, a DataVariable is a variable that stand for
]). data literals, and a GenericVariable is a variable
registration(?V:Car,CAD[?R:Deci that does not have a type. For instance, an Ob-
mal]):- jectVariable contains an optional reference to a
emission(?V,CO2[?E]), class which is used as its type:
speed(?V,kmh[?S]),
emispeed(CAD[?R],CO2[?E],kmh[?S
]). <r2ml:ObjectVariable
r2ml:name=”driver”
A registration query for a given vehicle class r2ml:classID=”userv:Driver” />
will thus unify only with correspondingly sorted
rule conclusions, hence directly branch into the R2ML allows both typed and untyped individu-
appropriate rule premises (the emiweight and als. For instance, a TypedLiteral consist in a lexical
emispeed premises compute the fees from the value and a type that is an RDF datatype or a user
emissions as well as the weights and speeds for defined datatype (subclass of rdfs:Literal):
Vans and Cars, respectively). The previous sec-
tion has shown URI-’webized’ versions of these
terminological classes. <r2ml:TypedLiteral r2ml:type
R2ML rules may refer to a vocabulary which Literal=”xs:positiveInteger”
can be R2ML’s own vocabulary or an imported r2ml:lexicalValue=”90”/>
one (such as RDF(S) and OWL). R2ML’s in-
ternal default vocabulary is a serialization of a RuleML supports order-sorted terms permit-
UML fragment of class diagrams. R2ML uses ting typed individuals, variables (exemplified
XML Schema datatypes as its standard datatype above) and data literals (Boley, 2003). Therefore,

637
Rule Markup Languages and Semantic Web Rule Languages

RuleML/XML defines an optional type attribute As in RuleML, RIF provides an optional type
for specifying a term’s (user-defined) type. Besides attribute for typed constants / individuals and a set
referring to the default XML Schema datatypes, of default XML Schema primitive data types such
typed terms may also link into external object as xsd:long, xsd:integer, xsd:decimal, xsd:string,
class hierarchies via their fully qualified class rdf:XMLLiteral and rif:text. For example:
names (e.g. Java classes) or taxonomies such as
RDF Schema class hierarchies, thus reusing the
OO class models and Semantic Web’s light-weight <Const
ontologies as pluggable external order-sorted type type=”rif:iri”>dc:creator</
systems. For example: Const>
<Const type=”xsd:string”>abc</
Const>
<Var type=”rdf://
owl:Vehicle”>V</Var> However, variables are not typed directly in
<Ind type=”xml:// a prescriptive form using, e.g., the type attribute
xsd:string”>abc</Ind> in the variable construct to denote that a vari-
<Var type=”java://java.lang. able X is of type T, i.e. X:T. Instead, RIF defines
Number”>X</Var> classification terms for class memberships, and
also for subclass relationships (cf. F-logic’s “:”
A Data term in RuleML contains a fixed value, and “::”):
like an RDF literal. It may be optionally associated
with an XML Schema built-in datatypeusing the • t#s is a membership term if t and s are
xsi:type attribute. For example: terms.
• t##s is a subclass term if t and s are terms.

<Data These classification terms are used to describe


xsi:type=”xs:dateTime”>2002-10- subclass hierarchies and membership constraints,
10T17:00:00Z</Data> e.g. expressing that a variable is of a certain class
(type).
This open order-sorted typing approach of In SWRL, a homogeneous combination of
RuleML provides higher levels of abstractions OWL (OWL DL and OWL Lite) and RuleML
and allows ad-hoc polymorphism with respect to (Unary/Binary Datalog), atoms can be of the form
coercion, i.e., automatic type conversion between C(x), P(x,y), sameAs(x,y) differentFrom(x,y), or
subtypes, and overloading, i.e., defining multiple builtIn(r,x,...) where C is an OWL description or
cases (rules with the same head except for types) data range, P is an OWL property, r is a built-in
taking different argument types. The ability to in- relation, x and y are either variables, OWL indi-
tegrate external Semantic Web vocabularies, data viduals or OWL data values, as appropriate. In
types and object oriented class hierarchies as type the context of OWL Lite, descriptions in atoms
systems provides syntactic expressiveness for easy of the form C(x) may be restricted to class names.
extension of the language with domain-specific That is, SWRL defines a rule language on top of
terminologies, and it facilities rule interchange OWL ontologies and hence directly supports the
across domain boundaries due to the explicit definition of class ontologies and their proper-
semantic definition of the used vocabulary, e.g., ties which can be used to type variables. For an
a Semantic Web ontology. example, see Figure 6.

638
Rule Markup Languages and Semantic Web Rule Languages

Figure 6.

This homogeneous integration approach is these secondary data storages as facts or object
adopted by R2ML from SWRL, while in RuleML values into the rules in order to reduce redundancy
without OWL one would refer to the external and high memory consumption. It should also
ontology which defines the vocabulary classes support outsourcing of expensive (pre-)processing
and their properties. of data to external systems, e.g., of mathematical
functions to procedural implementations such as
Java, or of SQL/SPARQL aggregation queries
EXTERNAL DATA INTEGRATION (constructive views) to database management sys-
AND DATA PROCESSING tems or RDF triple stores. A tight combination of
declarative and object-oriented programming with
Often Web rules refer to or describe functions and rich procedural attachments and language built-
queries over data stored in an external database ins, e.g. for querying, will facilitate the integration
which can be anything from log files to Web of existing functionalities, tools, and external data
sources or relational databases, data warehouses, sources into rule executions at run time.
or XML or RDF databases such as native XML Procedural attachments are procedure calls to
databases or RDF triple stores. The rule language external user-defined computational models of a
must allow for the direct dynamic integration of standard programming language, e.g., directly to

639
Rule Markup Languages and Semantic Web Rule Languages

Figure 7.

Java or C# methods. Therefore, procedural attach- date, time, intervals, lists, etc.
ments are a crucial extension of a modern Web All Web rule languages discussed in this chapter
rule language. They permit the combination of provide support for built-ins and some of them also
the benefits of declarative (rule-based) as well as for general procedural attachments.
procedural and object-oriented languages, e.g., to SWRL provides an extensible library of built-
delegate computation-intensive tasks to optimized in functions (http://www.daml.org/2004/04/swrl/
object code or to invoke procedure calls on object builtins.html) co-developed with RuleML. SWRL’s
methods which cannot be easily expressed in a built-ins approach is based on the reuse of existing
declarative rule-based way. Procedural attach- built-ins in XQuery and XPath, which are them-
ments should be supplemented with a typed logic selves based on XML Schema Part 2: Datatypes.
approach with external type systems such as Java SWRL built-ins are called via a built-in atom,
or Semantic Web ontologies, e.g. to assign external swrlx:builtinAtom, which identifies a built-in using
objects to typed variables, and with mode declara- the swrlx:builtin attribute and lists its arguments as
tions in order to safeguard the usage of built-ins subelements. SWRL built-ins are identified using
and calls to external functionalities. the http://www.w3.org/2003/11/swrlb namespace,
(Procedural Attachments). A procedural currently also used by RuleML. This is an example
attachment is a function or predicate whose of calling the multiply built-in (see Figure 7).
implementation is given by an external proce- SWRL does not provide direct support for
dure. Two types of procedural attachments are procedural attachments, but it could easily adopt
distinguished: this feature from Reaction RuleML.
R2ML by default supports SWRL and
• Boolean-valued attachments (or predicate XPath2 built-ins as predicate names of atoms
attachments) which call methods that return r2ml:DatatypePredicateAtom and symbols of
a Boolean value, i.e., that are of Boolean functions r2ml:DatatypeFunctionTerm. Func-
sort (type). tions and operators like addition, subtraction, etc.
• Object-valued attachments (or functional are translated into corresponding R2ML function
attachments) which are treated as functions terms. The operands of the functions implied by
that take arguments and return one or more the built-ins are enclosed by r2ml:dataArguments
objects, i.e., that are of a function sort. This and might be class attributes, class operations, data
also includes access to public object fields. variables, typed variables, or further nested built-in
functions. However, mode declarations are missing
(Built-Ins). Built-in predicates or functions are (see Figure 8).
special restricted predicate or function symbols R2ML supports procedural attachments in
in the rule language for concrete domains, e.g., order to access public data fields of objects
built-ins for strings, numerics, Boolean values, which might be bound to object variables (see

640
Rule Markup Languages and Semantic Web Rule Languages

Figure 8.

Figure 9.

Figure 9).
For reactive rules such as production rules, ‘External’ ‘(‘ Expr ‘)’
R2ML supports assignments of action expressions
in order to call object methods as actions in the where Expr is a UNITERM, i.e. either a
action part (see Figure 10). Boolean-valued function expression / predicate or
The RIF built-ins (http://www.w3.org/2005/ an object-valued functional expression. Since RIF
rules/wiki/DTB) overlap with the functions and does not support a general typed rule language,
predicates defined in XQuery 1.0 and XPath 2.0 it requires special guard predicates for all of its
Functions and Operators. supported datatypes to ensure the correct usage
Syntactically, built-in predicates and func- of the arguments of built-ins:
tions in RIF are enclosed by external terms of
the form:

Figure 10.

641
Rule Markup Languages and Semantic Web Rule Languages

Figure 11.

“effect” is interpreted impurely both for its value


External(“op:numeric-greater- and its (side-)effect action, e.g. by a procedural
than”^^rif:iri( attachment, using “modal” is interpreted as pure
?diffdays modality, and using “builtin” as a built-in (Figure
“10”^^xsd:integer)) 11).
External( The mode (“+”: input; “-”: output; “?”: input
“www.w3.org/2007/rif-builtin- or output) and type declarations ensure the correct
predicates#isInteger”^^rif:iri( usage of arguments in built-ins, i.e. that built-ins
?diffdays)) are called with ground values (not free variables)
of the expected types.
Note that the above example shows the RIF RuleML provides a concise integration of pro-
presentation syntax, not the concrete RIF XML cedural attachments. Methods of external object
syntax. classes can be called, including calls to object
RIF-FLD foresees procedurally attached constructors and calls to object instance and static
user-defined function terms or predicates to be object methods as well as access to public object
wrapped as external terms but does not define a fields. Constructed objects and returned result ob-
concrete approach for calling procedural actions jects can be assigned to variables. Nested selection
yet. However, it supports frame terms t[p1->v1 patterns can be defined over the result object col-
... pn->vn] which can be used to describe proper- lections such as “forall ?X where ?X=Person(age
ties of objects. > 30 and age < 40)” (see Figure 12).
The RuleML family (through its Reaction Most of the specific Semantic Web rule
RuleML branch) provides an open flexible ap- languages such as Jena and Prova support all
proach for pluggable external built-in libraries “standard” built-ins of Web Rule languages as
safeguarded by type and mode declarations. It well as many additional built-ins e.g. for meta
explicitly denotes the usage by the attribute per= interpretations of literals, exception handling,
”plain|value|effect|modal|builtin” on functional console printouts, collections, iterations/enu-
expressions and atomic relations. A <Rel> or merations, object property constraints, or access
<Fun> using “plain” is left uninterpreted, using to system environment properties. For instance,
“value” is interpreted purely for its value, using Prova supports several query built-ins to access

642
Rule Markup Languages and Semantic Web Rule Languages

Figure 12.

files, XML data sources via DOM, XPath, and families. Some of the language families such as
XQuery, RDF data sources via RDF triples and classical production rules historically only define
SPARQL and RDFS/OWL ontologies, as well as an operational semantics, while other rule families
various homogeneous or heterogeneous inference such as logical rules (see RuleML family in the
queries using external DL reasoners such as, e.g., section about expressive layering) are based on a
Pellet. Prova also provides a tight and natural Java model-theoretic and/or proof-theoretic semantics.
integration. Methods of classes in arbitrary Java A general research question is whether there exists
packages can be dynamically invoked from Prova a unifying semantic framework for all different
rules. The method invocations include calls to Java rule types. Work in this direction is pursued, e.g.
constructors creating Java variables and calls to in RIF (http://www.w3.org/2005/rules/wiki/FLD)
instance and static methods for Java classes as and Reaction RuleML (transactional transition
well as public object data fields. semantics for reaction rules subsuming all other
RuleML rules). However, since there is no general
consensus on one particular semantics for all ex-
FUTURE TRENDS pressive rule languages, an exclusive commitment
to one particular semantics for a Web rule language
A general rule markup language such as RuleML should be avoided (even in well-researched fields
or RIF covers many different rule types and rule such as logic programming several semantics

643
Rule Markup Languages and Semantic Web Rule Languages

such as well-founded semantics and answer set Another domain of research is the engineering
semantics are competing). Nevertheless, for cer- and maintenance of large rule-based applications,
tain subfamilies a preferred semantics can still where the rules are serialized and managed in a
be given and semantic mappings between rule distributed manner, and are interchanged across
families be defined. domain boundaries. This calls for support of
General rule markup languages need to in- verification, validation and integrity testing
clude practical language constructs which might (V&V&I), e.g. by test cases that are written in
not (yet) have a standard formal semantics based the same rule markup language and are stored
on classical model-theoretic logic. For instance, and interchanged together with the rule program.
procedural calls to external (object) functions, A proposal for self-validating rule bases adapting
operational systems, data sources and termino- a test-driven development approach from extreme
logical descriptions, are often vital to deal with programming in Software Engineering has been
practical real-world settings of distributed Web made for RuleML (Paschke et al, 2006) and for
applications. Recent research, e.g. in RuleML RIF (Paschke et al, 2005).
and RIF-PRD, is done on adopting such practical
language constructs without a standard formal se-
mantics but with a non-standard one. While there CONCLUSION
is a risk that these concessions to non-standard
semantics might endanger the benefits of formal In this chapter several important requirements and
semantics for the overall rule language, they turn design choices for a rule markup language have
out to be a crucial means to avoid limitations of been described. It was shown how the current Web
standard rule representations in the exploration of rule language proposals address these issues and
rule markup languages. The rule component will what characteristics derive from those solutions.
rarely run in isolation, but interact with various Commonalities as well as differences between
external components, hence call for functionalities the languages were presented and illustrated with
such as efficient object-oriented or relational/SQL- concrete examples. Discussions of the advantages
style retrieval and aggregation methods that are and disadvantages of the language design ap-
common in modern information systems. proaches reveal that all approaches legitimately
Further examples of useful practical constructs coexist at this stage, as all have their strengths
are the annotation of rules and rule sets with and weaknesses.
additional metadata such as rule qualifications,
rule names, module names, Dublin Core annota-
tions, etc., which eases, e.g., the modularization REFERENCES
of rules into rule sets (bundling of rules), the
creation of constructive views over internal and W3C (2000). W3C Resource Description Frame-
external knowledge (scoped reasoning), as well work. Candidate Recommendation. Retrieved
as the publication and interchange of rules / rule from, http://www.w3.org/TR/2000/CR-rdf-
sets on the Web. Advanced rule qualifications schema-20000327/
such as validity periods or rule priorities might W3C (2007). Semantic Web Layer Cake. Retrieved
for example safeguard dynamic updates (e.g. from, http://www.w3.org/2007/03/layerCake.
the incorporation of interchanged rules into the png
existing rule base), where conflicts are resolved
by rule prioritizations.

644
Rule Markup Languages and Semantic Web Rule Languages

Antoniou, G., Damasio, C. V., Grosof, B., Hor- Fuchs, N.E., Kaljurand, K., & Schneider, G.
rocks, I., Kifer, M., Maluszynski, J., et al. (2005). (2006). Attempto controlled English meets the
Combining rules and ontologies — A survey. challenges of knowledge representation, reason-
Deliverables I3-D3, REWERSE. Retrieved from, ing, interoperability and user interfaces. FLAIRS
http://rewerse.net/deliverables/m12/i3-d3.pdf 2006.
Ball, M. Boley, Hirtle, H.D., Mei, J., & Spencer, Giurca, A., & Wagner, G. (2005). Towards an ab-
B. (2005). The OO jDREW reference implemen- stract syntax and direct-model theoretic semantics
tation of RuleML. In Proceedings of Rules and for RuleML. RuleML, 2005, 45–55.
Rule Markup Languages for the Semantic Web
Grau, B. C., Calvanese, D., De Giacomo, G., Hor-
(RuleML-2005) (pp. 218-223). (LNCS 3791).
rocks, I., Lutz, C., Motik, B., et al. (2006). OWL
Boley, H. (2003). Object-oriented RuleML: User- 1.1 Web ontology language tractable fragments.
level roles, URI-grounded clauses and order- W3C Member Submission. Retrieved from, http://
sorted terms ( . LNCS, 2876, 1–16. www.w3.org/Submission/owl11-tractable/
Boley, H. (2006). The RuleML family of Web Halpin, H. (2006, May). Identity, reference, and
rule languages. In Proceedings of the Fourth meaning on the Web. WWW2006 Workshop on
Workshop on Principles and Practice of Seman- Identity, Reference, and the Web. Retrieved from,
tic Web Reasoning, Budva, Montenegro (LNCS http://www.ibiblio.org/hhalpin/irw2006
4187, pp. 1-15).
Heflin, J., Hendler, J., & Luke, S. (1999). SHOE:
Boley, H. (2007). Are your rules online? Four A knowledge representation language for Internet
Web rule essentials. RuleML (pp. 7-24). applications (Technical Report CS-TR-4078)
(UMIACS TR-99-71).
Bry, F., & Marchiori, M. (2005). Ten theses on
logic languages for the Semantic Web. PPSWR Hirtle, D. (2006). TRANSLATOR: A TRANSlator
(pp. 42-49). from LAnguage TO rules. Canadian Symposium
on Text Analysis (CaSTA), Fredericton, October
Codd, E. (1971). Alpha: A data base sublanguage
2006.
founded on the relational calculus of the database
relational model. In Proceedings of ACM SIG- Kifer, M., de Bruijn, J., Boley, H., & Fensel, D.
FIDET Workshop on Data Description, Access (2005). A realistic architecture for the Semantic
and Control, San Diego, CA. Web. In A. Adi, S. Stoutenburg, & S. Tabet (Eds.),
RuleML (LNCS 3791, pp. 17-29).
Ensing, M. (2007). The rule manager: A
graphical business rules environment. Demo Levy, A. A., & Rousset, M.-C. (1998). CARIN: A
at RuleML-2007. Retrieved from, http://2007. representation language combining horn rules and
ruleml.org/docs/Acumen%20Business%20-%20 description logics. Artificial Intelligence, 104(1-2),
RuleML.pdf 165–209. doi:10.1016/S0004-3702(98)00048-4
FIPA. (2000). FIPA agent communication lan- Lukichev, S., & Wagner, G. (2006). Visual rules
guage. Retrieved December 2001, from http:// modeling. In I. Virbitskaite & A. Voronkov (Eds.),
www.fipa.org/ Proceedings of the 6th International Conference
Perspectives of Systems Informatics (LNCS 4378,
pp. 467-673).

645
Rule Markup Languages and Semantic Web Rule Languages

Mei, J., Lin, Z., & Boley, H. (2007). ALCuP: Paschke, A., Boley, H., Kozlenkov, A., &
An integration of description logic and general Craig, B. L. (2007). Rule responder: RuleML-
rules, Web reasoning and rule systems. In First based agents for distributed collaboration
International Conference, RR 2007 (LNCS 4524, on the pragmatic Web. ICPW, 2007, 17–28.
pp. 163-177). doi:10.1145/1324237.1324240
Mei, J., Lin, Z., Boley, H., Li, J., & Bhavsar, V. C. Paschke, A., Dietrich, J., Giurca, A., Wagner, G., &
(2007b). The DatalogDL combination of deduc- Lukichev, S. (2006). On self-validating rule bases.
tion rules and description logics. Computational Int. Semantic Web Enabled Software Engineering
Intelligence, 23(3), 356–372. doi:10.1111/j.1467- Workshop (SWESE’06), Athens, Georgia.
8640.2007.00311.x
Rosati, R. (2006). The limits and possibilities of
Munoz, S., Perez, J., & Gutierrez, C. (2007). combining description logics and datalog. In T.
Minimal deductive systems for RDF. In E. Fran- Eiter, E. Franconi, R. Hodgson, and S. Stephens
coni, M. Kifer, & W. May (Eds.), ESWC (LNCS (Eds.), RuleML (pp. 3-4). IEEE Computer So-
4519, pp. 53-67). ciety.
Paschke, A. (2005). RBSLA: A declarative rule- Rosati, R. (2006a). Integration ontologies and
based service level agreement language based on rules: Semantic and computational issues. Rea-
RuleML. CIMCA/IAWTIC 2005 (pp. 308-314). soning Web.
Paschke, A. (2006). Verification, validation and Wagner, G., Antoniou, G., Tabet, S., & Boley, H.
integrity of distributed and interchanged rule (2004). The abstract syntax of RuleML - towards
based policies and contracts in the Semantic a general Web rule language framework. Web
Web. Int. Semantic Web and Policy Workshop Intelligence, 2004, 628–631.
(SWPW’06), Athens, Georgia.
Wagner, G., Damásio, C. V., & Antoniou, G.
Paschke, A. (2006b). A typed hybrid description (2005). Towards a general Web rule language.
logic programming language with polymorphic Int. J. Web Eng. Technol, 2(2/3), 181–206.
order-sorted DL-typed unification for Semantic doi:10.1504/IJWET.2005.008483
Web type systems. OWL-2006 (OWLED’06),
Athens, Georgia.
Paschke, A. (2007). Rule-based service level KEY TERMS and definitions
agreements - Knowledge representation for au-
tomated e-contract, SLA and policy management. Built-in Predicates or Functions: Special
IDEA Verlag GmbH, München. restricted predicate or function symbols in the
rule language for concrete domains, e.g., built-ins
Paschke, A., Boley, H., & Dietrich, J. (2005). for strings, numerics, Boolean values, date, time,
RIF use case. Rule interchange through test- intervals, lists, etc.
driven verification and validation. Retrieved Deontic Rules: Describe rights and obliga-
from, http://www.w3.org/2005/rules/wg/wiki/ tions, e.g., of institutions and agents in the context
Rule_Interchange_Through_Test-Driven_Veri- of evolving states (situations triggered by events/
fication_and_Validation, 2005. actions) and state transitions, where integrity rules
(see above) are a special case (‘introspectively’)
affecting the rule set itself.

646
Rule Markup Languages and Semantic Web Rule Languages

Derivation Rules: Infer conclusions from production rules are a special case with events
conditions (as in Datalog and Horn logic), where restricted to changed conditions.
facts are a special case with constantly true con- Rule Interchange Format: A common inter-
ditions. change format, such as e.g. W3C RIF or RuleML,
Facts: Various kinds of information such as for different rule types and rule families.
asserted atoms (formulas), individual-class mem- Rule Markup Language: A concrete markup-
berships (of ontology classes), (object-oriented) based rule syntax using e.g. XML for the Web.
instances, stored data (e.g., relational, XML), Semantic Web Rule Language: A rule lan-
states and event occurrences which might be guage specifically tailored for the semantic web
qualified, e.g., by priorities, temporally, etc. with a human-friendly syntax, e.g., a scripting
Integrity Rules (or integrity constraints): syntax (as opposed to a Rule Markup Language
Assertions which express conditions (or queries) for the Semantic Web).
that must always be satisfied. Transformation Rules: Specify term rewrit-
Procedural Attachment: A function or ing, which can be considered as derivation rules
predicate whose implementation is given by an of logics with (oriented) equality.
external procedure.
Reaction Rules are (Behavioral / Action)
Rules: (Re)act on occurred events (external events
or changed conditions) by executing actions, where

This work was previously published in Handbook of Research on Research on Emerging Rule-Based Languages and Technolo-
gies: Open Solutions and Approaches, edited by A. Giurca, D. Gasevic, and K. Taveter, pp. 1-24, copyright 2009 by Information
Science Reference (an imprint of IGI Global).

647
648

Chapter 2.20
Semantic Web Rule Languages
for Geospatial Ontologies
Philip D. Smart
Cardiff University, UK & University of Glamorgan, UK

Alia I. Abdelmoty
Cardiff University, UK & University of Glamorgan, UK

Baher A. El-Geresy
Cardiff University, UK & University of Glamorgan, UK

Christopher B. Jones
Cardiff University, UK & University of Glamorgan, UK

Abstract extends Description Logic Programs. A hybrid rep-


resentational approach is adopted where the logical
Geospatial ontologies have a key role to play in the component of the framework is used to represent
development of the geospatial-Semantic Web, with geographical concepts and spatial rules and an ex-
regard to facilitating the search for geographical ternal computational geometry processor is used for
information and resources. They normally hold storing and manipulating the associated geometric
large volumes of geographic information and un- data. A sample application is used to demonstrate
dergo a continuous process of revision and update. the proposed language and engine and how they
Limitations of the OWL ontology representation address the identified challenges.
language for supporting geospatial domains are
discussed and an integrated rule and ontology
language is recognized as needed to support the introduction And Background
representation and reasoning requirements in this
domain. A survey of the current approaches to The Internet is the single largest information re-
integrating ontologies and rules is presented and source in the world that is however still not being
a new framework is proposed that is based on and used to its full potential. To fully unlock the po-
tential of such a large knowledge resource and to
DOI: 10.4018/978-1-60566-402-6.ch007 enable its effective utilisation by both human and

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Semantic Web Rule Languages for Geospatial Ontologies

machine agents, information on the Web needs presented. Two approaches are identified, namely
to be machine-understandable using semantic as a hybrid approach where both systems of ontolo-
opposed to syntactic (e.g. HTML) markup lan- gies and rules are kept distinct and communicate
guages and tools. At the heart of this vision are only through an interface, and a homogeneous
ontologies which, in the context of the web, are approach where one system is mapped to and
logical theories that act to constrain and derive becomes accessible from the other.
information (Guarino,1995). They provide the In the second section, we first discuss the
necessary semantics and machine understanding representational and manipulation challenges
to the sheer volumes of information contained facing ontology management systems that aim
on the Web. to support geospatial domains. OWL as an ontol-
A significant proportion of information re- ogy representation language is evaluated against
sources on the web are geographically referenced. those challenges and the need for a integrated rule
Nearly 17% of all web queries contain place layer is highlighted. In the third section, current
names (Sanderson & Kohler, 2004) and the web, approaches to integrating rules and ontologies
powered by the simplicity of recent applications (logic programming and Description Logic) are
such as Google Maps, is increasingly being seen identified and classified. Based on a comparative
as a medium for the storage and exchange of geo- evaluation of both approaches, a homogenous ap-
graphic data in the form of maps. A geographic proach to integration, namely, Description Logic
or geospatial ontology is a model of terminol- Programs is chosen as a suitable platform for the
ogy and structure of geographic space as well development of geospatial ontology management
as records of entities in this space (Egenhofer, systems. In the fourth section, the potential and
2002). This chapter considers the development further extensions of this new approach are de-
and management of geospatial ontologies on scribed. In the fifth section, the implementation of
the Semantic web. By analyzing the nature and the approach is briefly sketched and demonstrated
complexity of the geographical concepts and data using a sample geospatial ontology described in
to be handled by these ontologies, we evaluate the chapter, followed by conclusions and future
the suitability of the current semantic web tools outlook in the final sections.
and suggest an appropriate platform to represent
and develop these ontologies.
In particular, geographical concepts are Managing Geospatial
complex, normally associated with geometric Ontologies
representations of their boundaries and location
and exhibit implicit spatial relationships that need In this section we consider a typical geospatial
to be computed and derived. Qualitative spatial ontology model, as shown in Figure 1. The model
reasoning as well as computational geometry is based on OGC guidelines for simple geographic
procedures are both established complementary features, see (OGC Technical Committee, 1999;
techniques for the representation and manipula- Vretanos, 2005), and other models commonly used
tion in this domain. In addition, maintaining the in existing geospatial ontology development (Fu et
spatial integrity of large geospatial ontology al., 2005, Smith & Frew, 1995). The terminology
bases is crucial for their realization. Ontology of the geo-ontology is relatively plain with regards
representation languages such as OWL are lim- to the number and type of constructs used. This
ited in their ability to handle the challenges in reflects typical geographic ontology developments
this domain. In this chapter, a survey of current which, beyond the complex representation of ge-
approaches to integrating rules and ontologies is ometry, are relatively sparse (parsimonious (Jones

649
Semantic Web Rule Languages for Geospatial Ontologies

Figure 1. An example geospatial ontology

et al., 2001)) and fit to purpose. In this section, is- can be represented using OWL as shown below.
sues related to the representation and management Examples from the axioms in the TBox (terminol-
of such geospatial ontologies are discussed and ogy or model) and the ABox (asserted knowledge)
OWL’s ability to handle them is evaluated. are shown. The logic-based syntax used here is
In Figure 1, a geofeature is a representation of that of OWL-DL.
any geographic phenomena that exist in space, e.g.
a forest, a building or a road. As such, its location TBox
and boundary can be specified using a geometric Geofeature ⊆ =1.GeofeatureID ∩
entity of point, line or polygon. Also, as it is located ≥1.RelatedTerm ∩ ∀RelatedTerm.Geofeature
in space, the relationships it exhibits with other …
geographical features are of interest, e.g. it may Region ⊆ Geofeature
be inside (topological), north of (directional) or Country ⊆ Region
near to (proximity) another feature. Some of these Ward ⊆ Region
relationships may be stored explicitly or need to be UnitaryAuthority ⊆ Region
computed from the features geometric entities. ABox
Cardiff: UnitaryAuthority
Representational Requirements Newport: UnitaryAuthority
Roath: Ward
Consider the map scene in Figure 2. The map shows Wales: Country
sample administration regions in Wales. Two Uni- Cathays: Ward
tary Authorities are shown; Cardiff and Newport Cardiff Meets Newport
along with some of their contained Wards. Roath Inside Cardiff
The ontology in Figure 1 can be used to Cardiff Inside Wales
represent this map of Places. In particular, it Roath Meets Cathays

650
Semantic Web Rule Languages for Geospatial Ontologies

Figure 2. A sample geographic scene showing administrative regions in Wales

The geometry of geographic regions is normally To support the geospatial domain, an ontology
complex and can consist of a large number of points. management system needs to consider the follow-
For example, the Unitary Authority of Cardiff in ing requirements.
Figure 2 is represented by over 1,000 points, pre-
senting a large storage and management overhead • Support the representation of basic spa-
on the ontology language. An experiment with a tial data models, spatial data types and
small geospatial ontology of geographic regions relationships.
was constructed in OWL, using the Jena2 library. • Provide a scalable capacity for handling and
The base OWL ontology, without the associated searching over large geometric data stores.
geometry, contained 1,000 individual places with
two data attributes (ID and name). The ontology oc- Manipulation Requirements
cupied approximately 2.2mb of persistent storage
space and 16mb of system memory to reason with. Manipulating geospatial ontologies involves
Adding the associated geometry for each place, the search, computation and retrieval of spatial
represented as general classes and data properties, properties and relationships. Two paradigms
the ontology grew to around 100mb on persistent are possible: quantitative, using computational
storage and the memory overhead increased to geometric procedures for structuring and search
over 1gb. Even then, spatial queries e.g. all Wards along with qualitative, using qualitative spatial
in Wales, are not supported within OWL as cur- reasoning techniques. Indeed, both paradigms are
rently its associated reasoning engines e.g. Racer complementary and can be used together.
(Haarslev et al., 2001) or Fact (Horrocks, 1998), In this section, both paradigms are explored
do not contain the necessary algorithms to perform to understand the issues they pose to geospatial
polygon in polygon geometry operations etc. ontology management systems.

651
Semantic Web Rule Languages for Geospatial Ontologies

Table 1. Part of a spatial composition table for topological relations between simple regions. 1 denotes
the universal relation.

DISJOINT MEET OVERLAP INSIDE CONTAINS


DISJOINT 1 DISJOINT, MEET, DISJOINT, MEET, DISJOINT, MEET, DISJOINT
OVERLAP, OVERLAP, OVERLAP
COVEREDBY, IN- I N S I D E , C O V- COVEREDBY, IN-
SIDE EREDBY SIDE
MEET D I S J O I N T, M E E T, DISJOINT, MEET, DISJOINT, MEET, OVERLAP, DISJOINT
OVERLAP OVERLAP OVERLAP, COVEREDBY,
COVERS, CONTAINS COVEREDBY, COV- I N S I D E , C O V- INSIDE
ERS, EQUAL EREDBY
OVERLAP D I S J O I N T, M E E T, DISJOINT, MEET, 1 OVERLAP, DISJOINT,
OVERLAP OVERLAP, COVEREDBY, MEET,
CONTAINS COVERS COVERS, INSIDE OVERLAP
CONTAINS COVERS, CON-
TAINS
INSIDE DISJOINT DISJOINT DISJOINT, MEET, INSIDE 1
OVERLAP
I N S I D E , C O V-
EREDBY
CONTAINS DISJOINT,MEET, OVERLAP, COVERS OVERLAP, OVERLAP, COVERS CONTAINS
OVERLAP CONTAINS COVERS COVERED-
CONTAINS COVERS CONTAINS B Y, INSIDE,
CONTAINS EQUAL

Geometric Manipulation • Support basic geometric computational


and spatial search functions to manipulate
As noted in the previous section, geospatial on- the geometric data stores.
tologies will normally be associated with large
geometric ontology bases representing the ground Qualitative Spatial Representation
location associated with geographic phenomena. and Reasoning
Simple manipulation of these phenomena will
involve the computation of their spatial prop- Over the past two decades much work has been
erties, such as length or area and relationships conducted on the development of qualitative
such as near or inside. Traditional computational spatial approaches to represent and reason over
geometry algorithms need to be implemented to space and spatial relations (Frank, 1992; Freska,
compute these properties. Also, spatial databases 1992; Gahegan, 1995; Egenhofer et al., 1999; El-
and information systems normally employ dif- Geresy, 2004). Qualitative spatial manipulation is
ferent forms of spatial data structures and index- import and complements quantitative geometric
ing techniques to facilitate searching over large processing in space, especially so when precise
geometric stores. geometric information is missing or simply not
Hence, a geospatial ontology management needed for the context of operation. Qualitative
system needs to consider the following require- approaches are based on the exploitation of the
ment. nature of the structure of space and the qualities
of the spatial relationships themselves for deriving
implicit information. Results of these approaches
are documented in what is known as composition

652
Semantic Web Rule Languages for Geospatial Ontologies

table of spatial relations such as the one shown complete or accurate. Such errors can propagate
in Table 1. to inconsistencies in the spatial relationships and
Entries in the table are possible relationships consequently to wrong information being stored
between two regions (A and C) resulting from the in the ontology bases. Erroneous updates to the
composition of relationships with another region data may go undetected unless appropriate spatial
B, i.e. R1(A,B) ⊗ R2(B,C) → R3(A,C). integrity rules are declared and applied.
As an example, consider the spatial relations Cockcroft (1997) catagorised spatial integrity
defined in the ontology in the previous section. A rules as 1) topological, maintaining the accuracy
new relationship can be derived as follows, where of topological information (which applies to all
⊗ denotes the composition of two relations: spatial relations) 2) Semantic, concerning the
Inside ⊗ Inside → Inside meaning of geographical features and how they
(RoathInsideCardiff) ⊗ (CardiffInsideWales) should legally be allowed to interact 3) user de-
→ (RoathInsideWales) fined, analogues to user defined business rules.
Entries in spatial composition tables can be seen Topological constraints can be further subdivided
as a set of first order compositional inferences, to structural errors, geometric errors, and topo-
and can be represented as a set of deduction or semantic constraints (Servigne et al., 2000).
inference rules of the form: Consider again our example ontology and a
∀ x,y,z: R1(x,y) ∧ R2(y,z) → R3(x,z) new fact to be inserted as follows.
where R1, R2 and R3 are spatial relations, for Roath Disjoint Wales
example, Although this fact is valid from the point of
∀ x,y,z Inside(x,y) ∧ Meet(y,z) → view of the ontological model, i.e. it asserts the
Disjoint(x,z) existence of a spatial relationship between two
where x,y and z are region variables, substituted regions, its spatial consequence implies an in-
for geofeature instances in the geospatial ontol- consistency as the explicit relationships already
ogy. The rule entails that the region bound to the stored implies that Roath must in fact be inside
variable x is disjoint from the region bound to the Wales.
variable z, if x is inside another region y, and y (RoathInsideCardiff) ⊗ (CardiffInsideWales)
meets (touches) z. → (RoathInsideWales)
Hence, another requirement for the geospatial To detect this inconsistency, we need not only
ontology management system is as follows. to encode the appropriate qualitative spatial rule
but also the following integrity constraint
• Support the representation of spatial ∀x,y,z Inside(x,y) ∧ Inside(y,z) ∧ Disjoint(x,z)
composition rules for qualitative spatial → error
reasoning. Different types of spatial integrity constraints
can be encoded to maintain the consistent spatial
Maintaining the Consistency structure and properties of the geospatial ontol-
of Geospatial Ontologies ogy bases. For example the following constraint
indicates that if an object is inside another it must
‘Consistency describes the absence of any logical also be smaller in size.
contradictions within a model of reality’ (Nectaria ∀x,y Inside(x,y) ∧ SmallerThan(y,x) → error
& Egenhofer, 1996). Errors in the description of A final consideration for the geospatial on-
the location and shape of geographical entities tology management system can therefore be as
are common, especially on the web, when data follows.
provided and manipulated by users may not be

653
Semantic Web Rule Languages for Geospatial Ontologies

• Support the expression and implementa- typical form of a spatial compositional in-
tion of spatial integrity constraints over ference rule.
geospatial ontology bases. • The description logic underpinning OWL-
DL does not support spatial data types.
Summary Representation of geometric objects using
generic class and property constructs is not
Description logics (DL) are a powerful represen- ideal and will have potentially high impli-
tational tool for describing real world concepts, cations on storage overheads (Haarslev et
their attributes and relationships. Key reasoning al. 1998).
mechanisms of any DL are checking concept • OWL does not support geometric compu-
satisfiability and inferring subsumption hierar- tation, analysis or spatial indexing. Simple
chies. On the terminological level a DL reasoner computational geometry calculations such
will infer concept hierarchies based on concept as area and distance calculations are not
subsumption. On the level of asserted knowledge possible to represent and consequently
(instance level) each individual’s type is inferred more complex spatial search queries are
if not already explicit. also not possible.
The description logic underpinning OWL-DL • Tableaux based reasoners (as used in most
(SHOIN(D)) has purposefully been restricted to DL reasoners) are poor for query answer-
preserve decidability and in large tractability of ing over individuals (Bruijn et al., 2005).
the language. Such restrictions lead to represen- Instance bases of geospatial ontologies are
tation and reasoning limitations. In addition to likely to be very large. Logic programming
these, several limitations can be recognized with reasoning engines are more suitable plat-
OWL as a platform to support the management of forms for reasoning in this case.
geospatial ontologies (Abdelmoty et al., 2005).
These limitations, some as highlighted in previous Realising the limitation of OWL, a rule layer
sections, are now summarized below. has recently been proposed in the semantic web
stack. Adding a rule layer to the existing ontology
• OWL’s first order, open world semantics layer (typically a DL variant) will help to over-
in combination with the non unique name come some of the representational and reasoning
assumption is not suitable for constraint limitations of OWL. In the next section a survey
checking (Bruijn et al., 2005). Extensions is given of existing approaches to the integration
to OWL have been proposed to overcome of logic programs and description logics, or less
this limitation, for example by translating formally rules and ontologies.
subsets of OWL to a logic program that as-
sumes both unique name and closed world
assumptions (Bruijn et al., 2005; Grosof et Integration of Rules
al., 2003), as well as by locally closing cer- and Ontologies
tain domain concepts using autoepistemic
constructs. Work on adding a rule layer to the semantic web
• “Triangular knowledge” can not be rep- technology stack was initiated by the W3C at the
resented directly in OWL (Horrocks, turn of the century. A fundamental challenge in
2005). In particular, inference patterns of facilitating the addition of this rule layer is find-
the form, ∀ x,y,c: rel1 (x,y) ∧ rel2(y,c) → ing means of integrating rules (Logic Programs
rel3(x,c) can not be represented. This is the - LPs) and ontologies (a descriptions logic- DL),

654
Semantic Web Rule Languages for Geospatial Ontologies

Figure 3. Different paradigms; a) ontologies in DLs and b) Logic programming systems

and in doing so handling the following semantic component is the subject of much work and
differences between logic programs and the DL debate. Current approaches can broadly be
subset of classical first order logic. classified into two categories based on the de-
gree of overlap between the two models in the
• First order languages have open world resulting system; the hybrid approach and the
semantics, while LPs adhere to closed homogeneous approach as described below.
world semantics.
• DLs do not assume the unique name as- Rule + Ontology: A Hybrid Approach
sumption, whereas all individuals in a
logic program are assumed unique. A hybrid approach is essentially a modular ap-
In addition, an integrated framework for proach to the integration of rules and ontologies.
rules and ontologies will face the following It is sometimes referred to as loose integration or
issues. the integration of rules and ontologies through
• Computational complexity of the result- strict semantic separation (Eiter et al., 2006).
ing system. The decidability and tracta- Reasoning distinction between the ontology
bility of the resulting reasoning system. (DL) component and the relational (rule) com-
The realisation and practical utilisa- ponent is maintained (Rosati, 2005). The ontol-
tion of the combined frameworks need ogy component is a description logic variant
to provide pragmatic reasoning proce- i.e. ALC and upwards, and the rule component
dures (with at most a polynomial time is typically some identified flavor of Datalog
complexity). (for example Datalog or Datalog∨).
• The modularity of the reasoning process. A complete hybrid knowledge base Κ is
The choice is between combining both represented by a pair Κ = <Σ,Π>, where Σ
systems into a single logical language represents the ontological (DL) component
with a uniform reasoner or retaining dif- and Π represents the relational (rule) compo-
ferent reasoners. nent. Π contains rule and ontology predicates
maintaining a strict separation between both.
Figure 3 illustrates the differences between Typically a rule r in the relational component
ontologies, as DL fragments of first order logic Π has the form:
and rule systems as logic programming frag- H ← B1 ∧... ∧ Bn: O1 ∧ Om m ≥1, n ≥1
ments of the same logic. The integration of the where H and B are both rule predicates
DL structural component with the LP relational (head and body predicates respectively) and O

655
Semantic Web Rule Languages for Geospatial Ontologies

Figure 4. Hyrbid Rule + Ontology Integration

represents an ontology predicate. The ontology information. Iterative reasoning is then performed
or structural predicates act as constraints on on both components until no more inferences can
the interpretation of the relational component. be drawn. This approach was adopted by Eiter at
Interaction between the rule and ontology al. (2004), namely Description Logic Programs,
reasoners take place through a safe interface and DL+log (Rosati, 2006).
(Eiter et al., 2006). The flow of information
through the interface is either unidirectional Homogenous Rule +
or bidirectional. Ontology Integration
In the unidirectional approach, reasoning is
performed over the ontology using an ontological Homogenous approaches present a complete or
reasoning engine (DL reasoner). Entailments from tight integration of both the ontology Σ and rule Π
the DL reasoner (Σ |= ω) are fed, as a starting point, components resulting in a single logical language
into the rule reasoner. Rules are interpreted such L. No syntactic or semantic distinctions are made
that they must satisfy the ontology predicates p in L between the ontology and rule predicates and
in ω. Early unidirectional approaches combined both can be interpreted under the same reasoning
unexpressive structural variants with unexpressive umbrella.
relational variants. For example AL-Log (Donini All works based in this approach employ a
et al., 1998) uses the foundational DL namely mapping (typically a recursive mapping) from
ALC with Horn Datalog while introducing the, one language to the other. Approaches that in-
now common, rule safety condition to maintain volve mapping the rule language to the ontology
decidability. That is, rule safety constraints the language exist. However, more common are ap-
use of each variable that appears in the head of proaches mapping from the ontology language
the rule to also appear in the body of the rule. to the rule language, thus opening the possibility
Moreover, only concepts are allowed as constraints of using existing rule engines for reasoning tasks
in the relational component. CARIN (Levy & (query answering etc). The mapping process
Rousset, 1995) overcame this limitation with a involves either completely combining both lan-
more expressive DL (ALCNR), and allowed both guages (expressive union), or embedding one
concept and role constraints whilst maintaining language into the other (intersection).
decidability using a form of role safety. Expressive union of the structural and relational
To further enhance reasoning, a complete syn- component is depicted in Figure 5 as the union
ergy of structural and relational reasoning can be of the entire LP and DL fragments within FOL.
catered for by introducing a bidirectional flow of The union brings about substantial computational

656
Semantic Web Rule Languages for Geospatial Ontologies

Figure 5. Homogenous Rule + Ontology Integration

complexities that often leads to undecidability, representation paradigm of WRL and represents
even if the languages of each are simplistic, see an ontology language in its own right.
(Schmidt-Schauß, 1989) for the proof. W3C’s
candidate rule language SWRL (Patel-Schneider Comparison of the Two Approaches
et al., 2004) is an example of unifying OWL-DL
with a Horn based rule language which does lead Augmenting ontologies with rules is of benefit in
to undecidability. general and offer potential solutions to the repre-
Intersecting the structural and relational com- sentational limitations of ontology languages such
ponents into their common fragment can help to as OWL, or indeed to most description logics.
retain decidability and tractability if for example, Of the hybrid approaches, early techniques
their common fragment corresponds to the horn either used unexpressive DL variants (e.g. AL-Log
subset of FOL. Description Logic Programs ALC), or only permitted class constraints in the
(Grosof et al., 2003) represent a complete ontol- relational component. Unidirectional approaches
ogy paradigm that is formed by the intersection produce only a subset of all inferences possible
of the description logic underpinning OWL-DL from the combination of both components. On
with Horn Datalog. The result is a sound, highly the other hand, bidirectional approaches suffer
tractable and practical paradigm from which fur- from higher, often intractable, and therefore un-
ther extensions can be be layered (Hitzler, 2005). acceptable worst case complexities (EXPTIME
Over the past few years a number of extensions or NEXPTIME see (Eiter et al., 2004)). The use
have been considered to DLP. For example, by of two separate reasoning engines to obtain a
considering disjunctive logic programs a larger unified output could be an inconvenient obstacle
fragment of DL can be mapped into the combined to reasoning and may add an additional cost to
language L (Motik & Volz, 2003), and in (Krötzsch run-time performance.
et al., 2008) which extended the reasoning potential Of the homogenous approaches, SWRL is
of DLPs, while not adversely effecting the decid- undecidable and as such is not mature enough for
ability and tractability of the language. full scale implementation. WRL-Flight is promis-
DLPs are also at the heart of the W3C’s ing and supports integrity constraints thanks to
alternative (to SWRL) candidate rule language both a closed word and unique name assumption.
WRL (Angele et al., 2005). WRL-core is the base However, its perfect model semantics is not as

657
Semantic Web Rule Languages for Geospatial Ontologies

compatible with most existing ordinary logic position tables, based on the Region Connection
programming engines. Disjunctive logic programs Calculus (RCC) (Cohn et al., 1997), a theoretical
exhibit high computational complexities (typically variant of the 9-intersection model shown in the
higher than polynomial time i.e. NEXPTIME or previous section, again for topological reasoning.
NEXPTIMENP (Eiter et al., 1997)), due to the The RCC spatial role axioms can be used to check
possibility of multiple minimal models (each the topological satisfiability of spatial individuals.
minimal model increases the size of the search However, as noted in their work, the proposal is
space exponentially). only decidable if RCC-5 or a more general family
The most tractable of all approaches is, un- of the spatial logic is used – for example by replac-
surprisingly, the least expressive logic, namely, ing the distinction between within and coveredby
Description Logic Programs (DLPs). A DLP can with the more general part-of relation.
be trivially mapped to existing logic programming To be of practical use, Wessels (2003) also
or production system engines e.g. XSB (Sagona argues for layerd hybrid deductive geographic
et al., 1993) or Rete (Forgy, 1982) respectively. information system that employs the DL ALCIrcc,
Logic programs or productions systems do not and recognizes the limitations of adding geometry
assume classical first order semantics. Making to a description logic, previously mentioned.
intuitive closed world and unique name assump- The full potential of using an integrated logical
tions make them more suitable models for integrity and geometric framework to support a general
checking tasks. Finally, logic programs will scale spatial ontology is yet to be realized. The potential
well to reasoning over individuals (Krotzsch et of such a framework to serve geospatial domains
al., 2006); a very desirable property, in particular has been recognized in some recent proposals
for the geospatial domains. (Chen et al., 2005; O’Dea et al., 2005; Smart et
al.). Exiting approaches to combine spatial logics
Approaches to Integrating with description logics to form new geospatial
Spatial Logics and Ontologies ontology language, highlight some important
issues. Both proposals above are limited in their
Current approaches to utilizing spatial logic and application both from a representational and rea-
reasoning in ontologies are based on the homo- soning perspective. In particular, the following
geneous approach to integration and propose issues still need to be addressed.
extending existing description logics with spatial
concrete domains and qualitative spatial reason- • Existing DLs need to include complex role
ing algorithms. The method provides a means composition constructs in order to effec-
to employ spatial reasoning for deduction and tively store spatial compositional inference
satisfiability checking of spatial information as- rules such as those described in table 1, as
serted into the description logic. well as role compositions that allow head
Haarslev et al. proposed an extension to disjunctions. A separate rule representation
ALCRRP(D) DLs to include a concrete spatial layer is needed.
domain and hence support spatio-terminological • A spatial ontology will need to support
reasoning (at the concept level) (Haarslev et a general set of spatial data types, such
al., 1998). However, the proposal was limited as regions, points, and lines. Restricting
to support only one specific spatial data type; a geospatial ontologies to spatial logics that
polygon. work over simple polygonal shapes limits
Wessels (2001) proposed the ALCIrcc DL that their general applicability.
includes role axioms derived from the spatial com-

658
Semantic Web Rule Languages for Geospatial Ontologies

Figure 6. Proposed DLP approach to representing geospatial ontologies and rules

• Geometric processing of spatial data is bet- we examine how this framework can support the
ter suited to dedicated external geometric requirements identified in the second section and
reasoning engines. Hence, ‘outsourcing’ also outline any necessary further extensions.
the geometric component of the ontologies
to external components outside the DLs DLP for Representing
and logic programming seems a reason- Geospatial Ontologies
able choice.
The geospatial ontology in section 2 can be cap-
A Semantic Web Rule and tured, without loss in a DLP. It is of interest to
Ontology Language for note that a large percentage of currently developed
Geospatial Domains ontologies are also within the representational
abilities of the DLP fragment (Volz, 2004). A
From the above survey of approaches to inte- transformation function, named DLP-Fusion
grating rules and ontologies and the analysis of (Grosof et al., 2003), is needed to translate OWL-
the requirements of the geospatial domains, we DL to DLP. The representation of the geometric
propose the use of Description Logic Programs component of geo-features is decoupled from the
DLP (a homogeneous approach to integration of ontology component and delegated to an external
rules and ontologies) as a basis for representing specialized system as described below.
and reasoning in geospatial domains. To further
support the particular requirements of these do- DLP-Fusion, OWL-DL to DLP
mains, an integrated framework combining both
the logical and the computational approaches to DLP-Fusion is a syntactical, semantic preserving,
geospatial data processing is also proposed. Figure bidirectional mapping between OWL-DL and the
6 depicts the new proposed language framework for Horn fragment of FOL, resulting in a new ontol-
representing geospatial onotlogies. In this section, ogy paradigm namely a Description Horn Logic

659
Semantic Web Rule Languages for Geospatial Ontologies

Table 2. Example DLP-Fusion mapping

DL Axiom First Order Rule Syntax


Region ⊆ Geofeature Region(x) → Geofeature(x)
Ward ⊆ Region Ward(x) → Region(x)
Topological ⊆ Spatial_Relationship Topological(x,y) → Spatial_Relationship(x,y)
Meet ⊆ Topological Meet(x,y) → Topological(x,y)
Roath: Ward Ward(Roath)
Wales: Country Country(Wales)
Roath Inside Cardiff Inside(Roath,Cardiff)

(DHL) ontology. are limited to compositional inferences with a


The mapping function, denoted T, takes DL definite head (one head predicate). To employ
axioms of the form (C ⊆ D), (S ≡ B), (T ⊆ ∀ P.D), more general qualitative spatial reasoning calculi
(T ⊆ ∀ P-.D), (a: D), (<a,b>: P), (P ⊆ Q), (P ≡ Q), disjunctive rules are required; rules that allow
(P ≡ Q-)), (P+ ⊆ P) and converts them into a rule head disjunctive of the form:
of the form A → B. The geospatial ontology in R1(x,y) ∧ R2(y,z) → R3(x,z) ∨ … ∨ Rn(x,z)
figure 3 can be directly mapped using this trans- for example:
formation. Examples are shown in Table 2. Contains(x,y) ∧ Meet(y,z) → Overlap(x,z) ∨
Where x and y are spatial variables from the Contains(x,z)
domain of individuals in the geospatial ontol- To work around this problem, spatial calculi
ogy. have been proposed that can be directly mapped
to Horn rules (Schockaert& Cock, 2007). Such
DLP for Qualitative spatial calculi are currently being tested with the
Spatial Reasoning proposed DLP approach, but the results are not
included in this chapter.
Unlike OWL, the Horn fragment of first order In addition to representing qualitative rules, a
logic is sufficient to represent triangular knowl- DLP can be extended to include procedural attach-
edge or role composition of the form (where R1, ments that implements, through calls to external
R2 and R3 are spatial relations, and x,y and z are geometric processing systems, spatial operators
spatial variables): for the computation of spatial properties and rela-
∀x,y,z R1(x,y) ∧ R2(y,z) → R3(x,z) tionships. A reasoning synergy is then possible to
For example from the composition Table 1, we compute required properties which could not be
can now represent the compositional inference: derived automatically using the spatial logic.
∀x,y,z Disjoint(x,y) ∧ Contains(y,z) →
Disjoint(x,y) DLP for Integrity Maintenance
Of note, OWL-DL can represent transitive
roles, for example: R1(x,y) ∧ R1(y,z) → R1(x,z), Classical Horn Logic assumes a first order
but not those involving three different relations semantics, including the open world and non
as shown in the example above. unique name assumptions. However, the logic
As DLPs are the logic programming equivalent programming equivalent of Horn logic, that used
of a Horn rule, they too can capture such spatial by DLPs, assumes a more intuitive closed world
compositional inferences. However, Horn rules and unique name assumption and is consequently

660
Semantic Web Rule Languages for Geospatial Ontologies

Figure 7. DLP based geo-ontology software framework

a suitable language for expressing and implement- inconsistencies as and when they arise and as
ing integrity constraints. such need to work in a forward chaining mode.
Integrity rules are headless rules, where satis- However, forward and backward reasoning modes
faction of all body predicates leads to a conflicting can be made to work together, an interleaved mode
knowledge base. In addition to this general form (Eisenstadt & Brayshaw, 1990). Then, predicates
of integrity rules, representation of ‘default’ rules in the forward system can be determined on the
and rule exceptions is also desirable. Consider for fly from the backward system, and no longer have
example, a rule that states that rivers and roads to be contained explicitly in the geo-ontology as
do not intersect. It may be desirable to express facts. This is advantageous in reducing storage
exceptions to this rule in the case where the enti- and memory overheads, but has an adverse effect
ties in fact do intersect in a “ford”. Enumerating on reasoning speed. That is, not all facts need to
rule exceptions is therefore a desirable quality. be stored in the geo-ontology, but querying the
Extension to DLP to support a form of default backward system requires evaluation of, in the
logic, e.g. Courteous Logic (Grosof et al., 1997) worst case, the entire rule set for each query.
is needed.
A backward chaining mode of inference is
used in the implementation of logic programs, Prototype Application
where querying the system results in exhaustive
rule evaluation. On the other hand, integrity rules A system has been developed that implements the
must continually monitor the ontology to find DLP proposed framework above within the Jena2

661
Semantic Web Rule Languages for Geospatial Ontologies

Figure 8. Geo-ontology visualisation tool

Semantic Web toolkit (Carroll et al., 2003). The [[<label>InsideDisjointIntegrity</


rule engine is based on the Rete pattern matching label><ruleGroup>Topological</rule-
production system (Forgy, 1982). A complete spa- Group>: Region(?x) AND Region(?y)
tial rule base of topological spatial compositions AND Region(?z) AND inside(?x ?y) AND
has been developed. The geometric processing Disjoint(?y ?z) AND Inside(?x ?z) -->
component is developed using the Oracle 10g error(?x ?z)
Spatial database management system. Figure 7
shows a concrete instantiation of the framework Figure 8 shows a screenshot of the ontology
in Figure 6. depicted as a graph of regions (as red circles) and
Evaluation of the system is currently ongo- relations (different colored edges).
ing on synthetic as well as real data sets. A small A new region (Penylan) as well as two new re-
example based on the ontology in section 2 is lationships are introduced to the ontology as found
demonstrated below. The following are examples using the web mining technique in (Schockaert et
of both an integrity and deduction rule in the al., 2008), these are:
system.
Penylan: Ward
Topological deduction rule:

Penylan inside Roath


[<label>ContainsEqualContains</
label><ruleGroup>Topological</rule-
Group>: Equal(?x ?c) AND Contains(?c Penylan inside Cathays
?y) --> Contains(?x ?y)]
The reasoning engine detects the inconsisten-
cies as a result of this update as shown in Figure 9,
Topological integrity rule: where edges resulting in the error are highlighted
(in red).

662
Semantic Web Rule Languages for Geospatial Ontologies

Figure 9. Topological inconsistencies

The three relationships between Cathays, Where, Penylan is inside Cathays and Roath
Roath and Penylan are all highlighted indicating meets Cathays means that Penylan can not be inside
an inconsistent spatial scene. In reality, Penylan Cathays, but it is hence the rule implies an error.
and Roath are neighbours, as shown in the Google [[<label>Contains_Inside</
maps view in Figure 10. To find this inconsistency, label><ruleGroup>Topological</ruleGroup>:
the following integrity rules were triggered. Region(?x) AND Region(?y) AND Region(?z)
[[<label>Inside_Meet</ AND Contains(?x ?y) AND Inside(?y ?z) AND
label><ruleGroup>Topological</ruleGroup>: Meet(?x ?z) --> error(?x ?z)]
Region(?x) AND Region(?y) AND Region(?z) Where, Roath contains Penylan and Roath
AND Inside(?x ?y) AND Meet(?y ?z) AND meets Cathays means that Cathays can not con-
inside(?x ?z) --> error(?x ?z)] tain Cathays, but it is hence the rule implies an
error.

Figure 10. Google Maps View ©2008 Google Map Data ©2008 Tele Atlas

663
Semantic Web Rule Languages for Geospatial Ontologies

Future Trends Description Logic Programs is a homoge-


neous approach to integrating both OWL-DL
Further to the current development of the proposed ontologies with classical Logic Programs. DLPs
framework, the following are ongoing research was proposed as the most suitable approach as
tasks: it provides a tractable, and hence scalable, base
for the expression of spatial rules and integrity
• Evaluating the scalability using large on- constraints.
tology bases. In particular, large place on- The DLP approach was then examined to
tologies are developed using some real data analyse its benefits and a new framework was
sets obtained from national mapping agen- proposed in which it can be implemented to
cies and multiple resources on the web. serve this domain. The implementation of the
• Evaluating the value of encoding hybrid framework is briefly sketched and demonstrated
spatial reasoning rules using multiple types using a sample ontology used for demonstration
of spatial relations. purposes throughout the chapter. Initial testing of
• Explore more expressive DLP fragments the approach demonstrates its efficiency and ef-
for an enriched geo-ontology representa- fectiveness. Evaluation experiments are currently
tion paradigm with increased reasoning being done to test the scalability of the approach
and integrity potential. to realistic geospatial ontology development on
• Deployment or testing of the proposed the Web.
framework for realistic application scenar-
ios, for example, representing and main-
taining Wikipedia articles. References

Abdelmoty, A. I., Smart, P. D., Jones, C. B., Fu,


Conclusion G., & Finch, D. A. (2005). Critical evaluation of
ontology languages for geographic information
This chapter proposes a new framework for the retrieval on the Internet. In Journal of Visual
development and management of geospatial on- Languages & Computing (pp., 331–-358).
tologies on the Semantic web. Challenges in rep- Angele, J., Boley, H., Bruijn, J., & Fensel, D.
resenting and manipulating geospatial knowledge (2005). Web Rule rule Language language (WRL).
in large onotology bases are identified and used Retrieved 2008, from http://www.w3.org/Submis-
to specify requirements for the new framework. sion/WRL
OWL as the Semantic Web ontology language
was evaluated and its limitations identified to Bruijn, J., Lara, R., Polleres, A., & Fensel, D.
handle those requirements. (2005). OWL DL vs. OWL flight: Conceptual
To overcome the limitations of OWL, the cur- modeling and reasoning for the semantic Semantic
rent approaches to the integration of ontologies Web. In Proceedings of the 14th international
and rules are reviewed and classified. A rule layer conference on World Wide Web (pp. 623-632).
will allow for the complex semantics of geospa- New York: ACM Press. New York, NY, USA.
tial ontologies to be represented, including the
expression of qualitative spatial logic rules for the
specification and derivation of spatial properties
and relationships.

664
Semantic Web Rule Languages for Geospatial Ontologies

Carroll, J. J., Dickinson, I., Dollins, C., & Reyn- Eiter, T., Gottlob, G., & Mannila, H. (1997). Dis-
olds, D. (2003). Jena: Implementing the Seman- junctive Datalog. ACM Transactions on Database
tic Web recommendations. (Internet Report). Systems, 22 (pp., 364-418) Vol 22.
Retrieved 2007 from http://dsonline.computer.
Eiter, T., Ianni, G., Polleres, A., Schindlauer, R.,
org/0211/f/wp6jena.htm.
& Tompits, H. (2006). Reasoning with rules and
Chen, H., & Fellah, S. & Bishr., Y. A. (2005). ontologies. In Reasoning Web, Second Interna-
Rules for geospatial Semantic Web applications. tional Summer School 2006, Lisbon, Portugal
In W3C Workshop on Rule Languages for In- (LNCS 4126, pp. 93-127).
teroperability.
Eiter, T., Lukasiewicz, T., Schindlauer, R., &
Cockcroft, S. A. (1997). Taxonomy of spatial Tompits, H. (2004). Combining answer set
data integrity constraints. GeoInformatica, 4(1), programming with description logics for the se-
327–343. doi:10.1023/A:1009754327059 mantic Semantic Web. In KR2004: Principles of
Knowledge Representation and Reasoning (pp.
Cohn, A. G., Bennett, B., Gooday, J., & Gotts,
141–151). Menlo Park, CA: AAAI Press, Menlo
N. M. (1997). Qualitative spatial representa-
Park, California.
tion and reasoning with the region connec-
tion calculus. GeoInformatica, 1, 275–316. El-Geresy, B. A., & Abdelmoty, A. I. (2004).
doi:10.1023/A:1009712514511 SPARQS: A qualitative spatial reasoning engine.
In Journal of Knowledge-Based Systems, 17, (pp.
Donini, F. M., Lenzerini, M., Nardi, D., &
89-102): Vol 17.
Schaerf, A. (1998). AL-log: Integrating data-
log and description logics. Journal of Intel- Forgy, C. (1982). Rete: A fast algorithm for the
ligent Information Systems, 10(3), 227–252. many patterns/many objects match problem.
doi:10.1023/A:1008687430626 Journal of Artificial Intelligence, 19, 17037.
doi:10.1016/0004-3702(82)90020-0
Egenhofer, M. J. (2002). Toward the semantic
geospatial web. In Proceedings of the tenth Frank, A. (1992). Qualitative spatial reasoning
ACM international symposium on Advances in about distances and directions in geographic space.
geographic information systems (pp. 1-4). ACM Journal of Visual Languages and Computing
Press. (LNCS 639, pp. 343–371).
Egenhofer, M. J., Mark, D. M., & Herring, J. Freksa, C. (1992). Using orientation information
(1999). The 9-Intersection: Formalism and its use for qualitative spatial reasoning. In Theories and
for natural-language spatial predicates. Report, methods of spatio-temporal reasoning in geo-
National Center for Geographic Information and graphic space (pp. 162–178).
Analysis.
Fu, F., Jones, C. B., & Abdelmoty, A. (2005).
Eisenstadt, M., & Brayshaw, M. A. (1990). Knowl- Building a geographical ontology for intelligent
edge engineering toolkit: The first of a two-part spatial search on the Web. In IASTED - Interna-
series presenting a knowledge-engineering tool- tional Conference on Databases and Applications
kit for building expert systems. Byte Magazine, part of the 23rd Multi-Conference on Applied
15(10), 268–282. Informatics, Innsbruck, Austria (pp. 167-172).

665
Semantic Web Rule Languages for Geospatial Ontologies

Gahegan, M. (1995). Proximity operators for Horrocks, I. R. (1998). Using an expressive


qualitative spatial reasoning. In Spatial informa- description logic: FaCT or fiction? In KR’98:
tion theory: A theoretical basis for GIS (LNCS Principles of Knowledge Representation and
988, pp. 31-44). Reasoning (pp. 636-645).
Gerevini, A., & Renz, J. (2002). Combining topo- Jones, C. B., Alani, H., & Tudhope, D. (2001).
logical and size information for spatial reasoning. Geographical information retrieval with on-
Artificial Intelligence, 137, 1–42. doi:10.1016/ tologies of place. In Spatial Information Theory:
S0004-3702(02)00193-5 Foundations of Geographic Information Science,
International Conference: COSIT (LNCS 2205,
Grosof, B. N. (1997). Prioritized conflict handling
pp. 322-335).
for logic programs, In Proceedings of the Inter-
national Symposium on Logic Programming (pp. Krotzsch, M., Hitzler, P., Vrandecic, D., & Sintek,
197-211). Cambridge: MIT Press. M. (2006). How to reason with OWL in a logic
programming system. In Rules and Rule Markup
Grosof, B. N., Horrocks, I., Volz, R., & Decker,
Languages for the Semantic Web, Second Interna-
S. (2003). Description logic programs: combining
tional Conference (pp. 17–28). IEEE Computer
logic programs with description logic. In Proceed-
Society.
ings of the 14th international conference on World
Wide Web 03 (pp. 48-57). Krötzsch, M., Rudolph, S., & Hitzler, P. (2008).
Description logic rules. In Proceedings of the 18th
Guarino, N. (1995). Formal ontology, conceptual
European Conference on Artificial Intelligence
analysis and knowledge representation. Interna-
(ECAI-08) (pp. 80–84). IOS Press.
tional Journal of Human-Computer Studies, 43,
625–640. doi:10.1006/ijhc.1995.1066 Levy, A. Y., & Rousset, M. C. (1995). Combin-
ing rules and description logics: An overview
Haarslev, V., Lutz, C., & Moller, R. (1998). Foun-
of CARIN. In Proceedings of the International
dations of spatioterminological reasoning with
Symposium on Logic Programming (pp. 635).
description logics. In Principles of knowledge
Cambridge: MIT Press.
representation and reasoning (pp. 112–123).
Motik, B., & Volz, R. (2003). Optimizing query
Haarslev, V., & Muller, R. (2001). RACER system
answering in description logics using disjunctive
description. In Proceedings of the First Interna-
deductive databases. In Proceedings of the 10th
tional Joint Conference on Automated Reasoning
International Workshop on Knowledge Repre-
(pp. 701-706).
sentation meets Databases, Hamburg, Germany
Hitzler, P., Haase, P., Krtzsch, M., Sure, Y., & (Vol. 79).
Studer, R. (2005). DLP isn’t so bad after all. In
Nectaria, T., & Egenhofer, M. J. (1996). Multi-
Proceedings of the 1st Workshop on OWL: Ex-
resolution spatial databases: Consistency among
periences and Directions (OWLED-05). CEUR
networks. In Integrity in Databases-Sixth Inter-
Workshop Proceedings (Vol. 188).
national Workshop on Foundations of Models and
Horrocks, I. (2005). OWL rules, ok? In W3C Work- Langauges for Data and Objects (pp. 119–132).
shop on Rule Languages for Interoperability.

666
Semantic Web Rule Languages for Geospatial Ontologies

O’Dea, D., Geoghegan, S., & Ekins, C. (2005). Schockaert, S., & Cock, M. (2007). Reasoning
Dealing with geospatial information in the Seman- about vague topological information. In Pro-
tic Web. In Proceedings of the 2005 Australasian ceedings of the Sixteenth ACM Conference on
Ontology Workshop: AOW ’05 (pp. 69-73). Aus- Information and Knowledge Management (pp.
tralian Computer Society, Inc. 593-602).
OGC Technical Committee. (1999). OGC, ab- Schockaert, S., Smart, P. D., Abdelmoty, A. I., &
stract feature specification (Tech. Rep.) Retrieved Jones, C. B. (2008). Mining topological relations
2008, from: http://www.opengis.org/techno/ from the Web. In DEXA Workshops (pp. 652-656).
specs.htm. IEEE Computer Society.
Patel-Schneider, P. F., Horrocks, I., Tabet, H. B. Servigne, S., Ubeda, T., Puricelli, A., & Laurini,
S., Grosof, B., & Dean, M. (2004). SWRL: A Se- R. (2000). Methodology for spatial consistency
mantic Web rule language combining OWL and improvement of geographic databases. GeoInfor-
RuleML (Internet Report). Retrieved 2008, from matica, 7–34. doi:10.1023/A:1009824308542
http://www.w3.org/Submission/2004/SUBM-
Smart, P. D., Abdelmoty, A. I., Jones, C. B., & El-
SWRL-20040521/.
Geresy, B. A. (2007). A framework for combining
Rosati, R. (2005). Semantic and computational rules and geo-ontologies. In Web Reasoning and
advantages of the safe integration of ontologies Rule Systems: First International Conference,
and rules. In Principles and Practice of Semantic RR 2007, Innsbruck, Austria (LNCS 4524, pp.
Web Reasoning, Third International Workshop, 133-147).
PPSWR 2005, Dagstuhl Castle, Germany (LNCS
Smith, T. R., & Frew, J. (1995). Alexandria digi-
3703, pp. 50-64).
tal library. Communications of the ACM, 38(4),
Rosati, R. (2006). DL+log: Tight integration of 61–62. doi:10.1145/205323.205340
description logics and disjunctive datalog. In
Volz, R. (2004). Web Ontology reasoning with
Proceedings of the Tenth International Conference
logic databases. PhD thesis, AIFB, University
on Principles of Knowledge Representation and
of Karlsruhe.
Reasoning (pp. 68-78). AAAI Press.
Vretanos, P. (2005). Open geospatial consortium
Sagonas, K., Swift, T., & Warren, D. (1993). XSB:
filter encoding implementation specification
An overview of its use and implementation (Tech.
(Tech. Rep.). Retrieved 2008, from http://www.
Rep). Retrieved 2006, from http://citeseer.ist.psu.
opengis.org/techno/specs.htm.
edu/195039.html
Wessel, M. (2001). Obstacles on the way to
Sanderson, M., & Kohler, J. (2004). Analyzing
spatial reasoning with description logics – Some
geographic queries. In Workshop in Geographic
undecidability results. Working Notes of the 2001
Information Retrieval.
International Description Logics Workshop (DL-
Schmidt-Schauß, M. (1989). Subsumption in 2001) Vol. 49, Stanford, CA, USA.
KL-ONE is undecidable. In Proceedings of the
1st International Conference on Principles of
Knowledge Representation and Reasoning (pp.
421-431). Toronto, Canada.

667
Semantic Web Rule Languages for Geospatial Ontologies

Wessel, M. (2003). Some practical issues in build- Integration of Rules and Ontologies: How
ing a hybrid deductive geographic information to integrate description logic ontologies with clas-
system with a DL component. In Proceedings of sical rule based logic programs, while preserving
the 10th International Workshop on Knowledge semantics and maintaining decidability and trac-
Representation meets Databases (KRDB 2003) tability. Approaches are either based on the union
(Vol. 79). Hamburg, Germany. or the intersection of the two languages.
Interleaved Mixed Mode Reasoning: Allow-
ing knowledge querying using backward rule sets
during the course of forward inferencing. Facts
Key Terms and definitions
in the premise of forward rules can be found
Description Logics: Successors to Semantic directly from explicit facts or implicitly through
Networks and Frame Based Languages which can additional inferences.
represent both asserted and structural knowledge. Ontology: Those things that exist are those
Modern description logics stem from KL-ONE things that have a formal representation within the
which formalized the ideas of Semantic Networks context of a machine. Knowledge commits to an
and Frames. A description logic can describe the ontology if it adheres to the structure, vocabulary
world in terms of properties or constraints that and semantics intrinsic to a particular ontology i.e.
specific individuals have to satisfy. it conforms to the ontology definition. A formal
Geospatial Ontologies: A specilisation of ontology in computer science is a logical theory
ontology that represent only knowledge from the that represents a conceptualization of real world
geographic and spatial domain. concepts.
Geospatial Rules: Rules or logical rules of Qualitative Spatial Reasoning: Represen-
inference represent dynamic, relational knowl- tation of continuous properties of the world by
edge as opposed to static structural knowledge. discrete symbols, and then reasoning over such
A rule is an inference of the form PREMISE symbols without recourse to more expensive (com-
implies CONCLUSION, mimicking human putationall) quantitative knowledge. Qualitative
cognitive reasoning processes. If the PREMISE knowledge and reasoning better mimics human
condition holds, then the CONCLUSION condi- spatial reasoning processes.
tion is deducible. Geospatial rules are those that
represent and reason with geographical and or
spatial knowledge.

This work was previously published in Handbook of Research on Emerging Rule-Based Languages and Technologies: Open
Solutions and Approaches, edited by A. Giurca; D. Gasevic & K. Taveter, pp. 149-169, copyright 2009 by Information Science
Reference (an imprint of IGI Global).

668
Section III
Tools and Technologies

This section presents extensive coverage of the technology that informs and impacts Web technologies.
These chapters provide an in-depth analysis of the use and development of innumerable devices and
tools, while also providing insight into new and upcoming technologies, theories, and instruments that
will soon be commonplace. Within these rigorously researched chapters, readers are presented with
examples of the tools that facilitate and support the emergence and advancement of Web technologies.
In addition, the successful implementation and resulting impact of these various tools and technologies
are discussed within this collection of chapters.
670

Chapter 3.1
New Paradigms:
A Collaborative Web-Based Research Tool
Hamish Holewa
International Program of Psycho-Social Health Research, Central Queensland University, Australia

abstract Introduction

The chapter aims to document the challenges as- This chapter presents, through the medium of a case
sociated with the management of an international study, insights gained from practical experience de-
research program and to look at innovative, infor- veloping and implementing an original web based
mation technology (IT) based ways of tackling collaborative research tool to assist and enhance
these. Through the medium of a case study, insights the management of an existing, qualitative research
gained from practical experience developing and program. The case example used is that of the Inter-
implementing an original Web based collaborative national Program of Psycho- Social Health Research
research management tool are discussed. This tool (IPP-SHR). This case study provides the reader with
is based on a centralised model of information dis- insights into the ways in which information technology
tribution and access. It was designed following a (IT) processes can be used to overcome problems as-
reductionist analysis of existing research processes sociated with the post-modern research environment.
and procedures. The ways in which the integration Within this context the major challenges are to address
of responsive IT processes into the management of a the fragmented nature of research locations, staff and
large international research program have removed project administration within a global setting.
redundancies and increased automation and research Technological advances have paved the way for
efficiency are also discussed. global research, enabling it to transcend physical,
geographical and cultural boundaries. However, there
are still great challenges to be overcome in conducting
a truly international research program. The chapter
aims to document the challenges associated with
the process and management of a large international
research program and to look at innovative, IT based
DOI: 10.4018/978-1-59904-970-0.ch005 ways of tackling these.

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
New Paradigms

The International Program of Psycho-Social contributors to the success of such a system. The
Health Research (IPP-SHR) provides international system uses a central website, where users with
leadership through research, publication, educa- appropriate security credentials like correct user
tion, media, newsletters and podcasting activities name, password and encryption key can deposit
in the area of psycho-social health research. This files related to the research processes, implement
program, explores a broad range of psycho-social automatic work flow processes for dictation, tran-
health issues including: the lived experience of scription and coding processes, view work and
serious and terminal illness; haematology and project flows and progress, schedule appointments
oncology; palliative care; indigenous health; rural and stipulate task for other users or team members.
and remote health; mental health; obstetrics; bio- The system improves research efficiency and
ethics; and the interface between patients and the lowers research costs. This is achieved through a
health care system. The core aim of IPP-SHR is streamlined website portal offering best practice
to ‘make a difference’ by informing policy and security, enhanced ethics compliance, limiting or
service delivery in the real world of health care. reducing redundancy between processes and team
This program utilises qualitative, or natu- members, and providing accurate information on
ralistic, research methodology, which seeks to the process and state of each particular research
document the voice of the research participant project. The software also provides team building
from their own world view (Streubert & Carpenter and mentoring activities through the use of project
1995). Such methodologies are underpinned by reporting, a bulletin board, discussion forums and
a philosophical perspective that listens to, rather team feedback.
than imposes on the experience of others and has
a sensitivity to the disempowered and marginal-
ized (Latimer et al. 2003) The large and diverse Background and challenges
amount of the data gained from using such meth-
odologies, coupled with diverse and geographical The research paradigm and context within which
isolated data collection sites of an International IPP-SHR operates presents unique challenges.
program, necessitated the design and construction Although the program has developed gradually
of a central based management system. over the last decade, it has only recently evolved
After extensive literature searches in ma- to the level of national and international research
jor databases, consultation with software and data collection and collaboration. As a qualitative
project management vendors, collaboration and research program with a focus on the human in-
discussion with international leaders in qualita- terface of health care, the challenge is collecting
tive methodologies, it was evident that no such and managing the magnitude and complexity
program existed to meet the specific requirements of data gained from naturalistic methodologies
of IPP-SHR or collaborative multi-site qualitative over extensive geographical areas. This section
research projects. As such, to meet the challenges details the challenges and problems associated
and technical difficulties associated with IPP- with running a decentralized, location unspecific
SHR’s methodology and operation, an internet international research program. It also introduces
based research tool was designed. Server side the equity and ethical considerations associated
technologies were utilized to achieve a central with research.
research portal for IPP-SHR practitioners to use IPP-SHR operates in an environment, charac-
and collaborate through, independent of their terised by the fragmentation of location, staff skills
physical location. The maturing of server side and expertise, participant groups, disciplinary
and Internet connectivity and speed are major focus and broad topic interests. Translated to the

671
New Paradigms

practicalities of research activities, this means, produce over two hundred pages of language texts,
data collection and analysis occurs in many geo- excluding supplementary data such as descriptive
graphical locations and time zones and focuses statistics. The sheer quantity of data produced
on a multiplicity of research topics. Additionally by IPP-SHR’s qualitative methodologies neces-
staff management needs to address a multiplic- sitates that the discrete processes of qualitative
ity of duties and responsibilities, and access and research (i.e. verbatim recording and transcrip-
control of project information, some of which is tion of interviews, managing coding processes)
confidential. Also posing challenges are the prac- be streamlined.
tical necessities of enabling simultaneous access Although, the data gained in IPP-SHR projects
by multiple team members to a broad range of is usually qualitative, occasionally descriptive
specific documents, and creating processes for data is also included. Thus, any system that is
multi-site data entry and analysis. to facilitate and streamline IPP-SHR’s research
As a core component of IPP-SHR’s philosophy processes also needs to be scalable and flex-
is to ‘make a difference,’ and to, ‘document the ible so as not to disadvantage collaboration and
human experience of human illness,’ IPP-SHR’s research efficiency if projects required support
qualitative methodology focuses on a phenom- for different methodologies. Additionally, each
enological perspective using exploratory, iterative project has differing qualitative methodological
and open–ended interviews. A phenomenological requirements, which vary according to project
perspective is used with data analysis and process size, participant numbers, interviews per par-
as its inherent aim is to document and record the ticipant, and timeframes. As such, any software
particular phenomena or appearance of things as implementation needs to support IPP-SHR’s
a lived experience (Streubert & Carpenter 1995). diverse projects which requires flexibility, scal-
Data gathered during the interviewing process ability and durability, plus continuity of access
is then transcribed verbatim and analysed from over extended periods of time.
the view point of the participiant (McGrath & Additional requirements and challenges posed
Holewa 2006). Such analyses and exploration is to the development of a software system are
undertaken without imposing specific theoreti- stipulated by the regulatory and policy frame-
cal or conceptual frameworks on the interview works within which IPP-SHR operates. Human
or data analyses to ensure that the individual Research Ethic Councils (HREC) stipulate pri-
experience is recorded (Polit & Hungler, 1995). vacy, informed consent, confidentiality, and audit
As such, it is methodologically important for the requirements for research approval involving
data management system to not only store data humans (Australian Federal Government 1988;
correctly and without corruption but to ensure NHMRC 2003; AIATS 2004). Additionally, audit
that research staff can engage in a rigorous data requirements necessitate data be stored in a con-
analysis process. IPP-SHR has a particular pride fidential and secure location for a period of time
and a documented history of ensuring a rigorous from five to seven years. The challenges imposed
data analysis process by which the findings are by such policy and regulatory frameworks require
driven by a meticulous coding of all statement IPP-SHR to store and be legally liable for any
by participants. information which is gathered throughout the
Due to the rigour of IPP-SHR’s application research process. This is particularly important
of qualitative methodologies, even relatively within IPP-SHR’s operating paradigm due to the
small research projects produce large amounts decentralized composition of IPP-SHR research
of data. For example, a small project in which projects and staff. Document control, ethical re-
ten participants are interviewed will on average quirements and privacy issues represent a major

672
New Paradigms

concern and challenge for the research process see advances that technology can make and can
and for any software system designed to support suggest, modify and drive new innovations and
such research. uses. Understanding this user determined inno-
IPP-SHR research practitioners have not vation variable allows users and IT designers to
previously been exposed to a high level of IT implement user friendly software with practical
involvement and had a reticence based on lack outcomes which correctly and efficiently opera-
of familiarity. Consequently, an additional chal- tionalises research processes. That is, by a process
lenge for incorporation and implementation of the of continual feedback and discussion between
system was the development of comprehensive research practitioners and IT consultants, both
and supported training packages. The case study parties were able to understand and learn the
highlights the need for mutual understanding from constraints, practicalities and possibilities of au-
both academic and IT disciplines in developing tomating the research process. Development and
the system and the positive outcomes that can be specification of the software requirements was
gained from incorporating such diverse profes- partly informed by the Institute of Electrical and
sional viewpoints. This case study profiles the Electronics Engineers (IEEE) document (1998)
importance of IT leadership and innovation in referring to recommended practice for software
meeting the outlined challenges. requirement specifications.
Such in-depth discourse between the two
professional groups and the development of a
Design and implementation thorough understanding of the research processes
allowed for extra innovative features to be created
Design and implementation of the technological and other unnecessary ones to be omitted. This
interface created for the research program required assisted in producing an effective program with
a detailed understanding and high level analysis an efficient design. For example, features such
of the challenges, backgrounds, procedures, needs as user control and public and private files where
and desires of proposed users of the system. This not proposed in the original design, however after
required both research practitioners and IT con- extensive consultation this feature was warranted
sultants to have a high level of cooperation and as crucial. This was in particular reference to
effective discourse focussed on the needs and identifiable participant data which, in line with
wishes of the research practitioners. Although HREC agreements, only authorised persons should
research practitioners bring to their work an be able to view. Without consultation and ongoing
understanding of the role and function of IT, this discussion between both professional groups, this
was insufficient for translating their work into feature would have been omitted from the final
an innovative and original incorporated system. product. Additionally, features proposed by the
What became evident in this experience was IT professionals were omitted as they did not add
the need to provide IT leadership, collaboration value and could potentially complicate use of the
and experience in translating the research work final software interface.
components and wishes into a sustainable and All parties agreed on final software specifica-
useable system. tions before commencement of programming and
Design and implementation of the system software construction. The use of specification
needed to be in a bottom up fashion. Effective allowed for accurate budgeting and timeframe
translation, education and implementation as- projections. It also allowed for a medium in which
sisted in establishing a self perpetuating and users from both professional groups could suggest
learning experience for users. Users are able to and implement changes before programming.

673
New Paradigms

This is particularly important in relation to the produced by the research process were discretely
ethical imperative of efficient and effective use assigned a unique identifier. Components were
of resources, as changes made to the system once then tracked in order to ascertain a step by step
programmed are costly and expensive compared understanding of the research process. For ex-
to alterations made in the pre–programming stage. ample, the interviewing of a participant produced
(Diaz & King, 2002) a component, “audio voice file,” this file was then
This considered, laborious and collaborative forwarded to the transcriber, who by transcribing
consultative period before manufacture of the the audio voice file produced another component,
software began is an essential step for any project “text document of interview verbatim”. The flow
which bridges two disparate disciplines. Continual and process of information creation (arrange-
communication between research practitioners ment of participant interview) to research output
and IT consultants, coupled with a referable (publishing and dissemination of information)
specification documents, allowed for greater flex- was tracked and compiled into a flow diagram.
ibility, increased innovation and design features, (See Figure 1)
producing a software package that is of direct use After construction of the flow diagram each
and benefit to its intended users. process was analysed to identify redundant pro-
cesses and avenues through which automation
and efficiency could be increased. For example
Workflow analysis, a redundant process found was the duplication of
program specifications communication occurring between team mem-
and implementation bers in the activity of scheduling an interview.
Although, in its simplest form, scheduling can
Extensive research and market searches indicated be achieved through one person (enrols the par-
that no suitable, scalable and location-independent ticipant, schedules and conducts the interview,
software was available to serve the unique de- compiles and stores documents relating to the
mands and constraints faced by the research group. interview and transcribes the interview) this is a
Although there are numerous research software rarity due to the aforementioned operating chal-
packages specifically designed for qualitative lenges of the research program. It is IPP-SHR’s
methodologies (e.g. NUD*IST or Atlas/ti) (Barry experience that there can be up to four people
1998) there was an absence of project manage- involved in the process of arranging, conducting
ment software which incorporated qualitative and transcribing an interview. Similar issues of
methodologies and associated workloads. In redundancy and repetition of tasks were found
order to fill this gap, a software program was within the interface between transcription, coding
designed to handle such programmatic demands and recording keeping.
and incorporate data analysed within the major Analysis of data flow and information creation
qualitative data analyses programs. A full working in Figure 1 suggested that many tasks undertaken
commercial version of this program can be found within a project only need be completed once if
at www.quadrant-pm.com every member involved had access to such infor-
A reductionist analysis of the workflow and mation. As such, a centralised distribution model
procedures that IPP-SHR uses to operationalise of information storage, provision and access was
its research was critical for successful design and developed to enable this streamlining to occur.
implementation of the software program. This Using the centralised model, users of the system,
process involved detailing and referencing every with appropriate rights and security credentials,
activity conducted by researchers. All components have access to each component of information that

674
New Paradigms

is relevant to the specific needs of their task. This each particular research instance and co-ordinating
enables users to maintain up-to-date information changes between each other, only one record is
and accurate data repositories, which assists in kept and only one change is made. Using the model
smooth and efficient running of research processes. in relation to interviewing, scheduling would be
Furthermore, duplication, redundancy and errors provided to the centralised data source (such as
are reduced as only one record exists per each re- a web based calendar) and information given to
search instance per project. Instead, for example, of other members associated with such interview
four team members keeping individual records of will be provided from such source.

Figure 1. Procedural, sequential information creation and flow (©2007, Hamish Holewa. Used with
permission)

675
New Paradigms

Figure 2. Centralised and distributed information flow (©2007, Leena Hiltunen. Used with permis-
sion)

As shown in Figure 2, research staff interacts interview time; (3) organises HREC compliance
individually with the central dynamic website such as informed consent and project description
and information flow occurs via a centralised, procedures; (4) the interviewer (INT) completes
distributed model. This distinctly contrasts with the interview and produces a digital recording; (5)
the processes formerly used, whereby information the recording is sent to the transcriber (TRS) who
flow was procedural, sequential and task oriented. completes transcription processes; and, (6) send
Once information has been created, research team completed transcription to RA or project officer
members with appropriate security credentials can (PO), who initiates coding and qualitative analysis.
access the most up-to-date source of information At any point in the process, each member is reliant
through a website portal without needing to ask on another member to complete their scheduled
other team members for the status and latest ver- task, record appropriate details and ensure HREC
sion of such information. and other policy guidelines are complied with.
The depicted system in Figure 2 also removes Furthermore, the PO or investigator may have to
unnecessary and redundant tasks with the motiva- contact up to four people to receive an accurate
tion of increasing automation and realising effi- report on the progress of the research project.
ciencies through the process. Once, implemented A major advance has been achieved on the au-
and configured, simple procedural and reporting tomation and reduction of the described processes.
tasks are completed automatically. In previous This is done through the automatic storage and
work flow the researchers were involved in a six referencing of documents and information (par-
step processes which included: (1) research as- ticipant details, consent forms, digital interview
sistant (RA) enrols participant; (2) schedules an files, transcriptions) which is completed at each

676
New Paradigms

stage and by each individual person responsible based initiatives has been successful within this
for such tasks. Once each task is completed by the case study. The provision of a message board,
assigned research team member, the team member collaborative document sharing applications, ver-
interfaces with the website portal and flags its sion control and work processes has encouraged
completion and uploads the accompanied file or community building and increased collaboration
information. The website will then contact the and use of the service. It has also facilitated self
next research team member within the process. directed learning and enquiry and has supported
For example, once an interviewer has completed a forum for public discussion.
the interview, this is flagged within the system Evidence associated with the development of
and the corresponding sound file is uploaded to an online community and self directed learning
the website portal. Once uploaded, the website stemmed from online dialogue evident within the
sends an alert to the transcriber of the need for an bulletin board feature and the minimum formal
interview to be transcribed. This process reduces training needed for new users. Through anecdotal
record keeping times and ensures an up-to-date records and conversions with users of the system, it
report on progress. was noted that software training and learning was
The distributed, centralised model of project largely conducted within the system. Although,
work flow implementation has been realised software training was required for initial users of
through the use of four discrete online modules the system, the historical dialogue between such
operating within the project management portal. users provided a self directed learning environ-
These four modules are as follows: calendar ment for new users, and assisted in development
and scheduling; project workflow and progress; of an environment conducive to posting public
document and version control; and administra- comments and asking questions to the group of
tion. Included are different levels of user access users. However, it should be noted, that the de-
and control and differing levels of information velopment of such a community of self- directed
categories, ranging from public to private. Access learning may be further successful due to the
to the website portal occurs through a computer separate and disparate physical location of the
with internet connectivity and browser support. users. It is unclear whether such development of
The website server runs on a combination of cli- a bulletin board and self- directed learning would
ent and server side code attached to a centralised have developed if user where located in the same
database. Most computation and processing is physical space, e.g. in the same office.
completed on the remote server and, apart from The central place assumed by the software
internet connectivity, there is little computational within the research program creates a responsive
demand on client computers. medium through which staff concerns, ideas and
issues can be raised. This enables research staff
to pose research-related questions and comments,
Self directed learning and provide direct feedback on the program’s us-
communities ability. The facilitation of such open communica-
tion allowed for quick development of ideas and
The development of a community, centred upon creative problem solving. It also helped to foster
online collaborative applications, is a well regard- team culture and shared identity, whilst acting as
ed method for increasing the use and accessibility an effective training tool.
of a system and for encouraging users to engage The software program’s ability to dynamically
in self directed interaction and learning (Neus, provide users with up-to-date project information
2007). The implementation and use of community has encouraged individual and team ownership of

677
New Paradigms

the research projects. As the program allows for or disadvantaged groups. The adoption and inclu-
instant viewing of research project status and prog- sion of such technologies, has the opportunity to
ress, research members are also able to identify any lower research costs, providing greater potential
aspects of the research process hindering overall outputs from fewer resources.
progress. Additionally, staff training and supervi- Increased security represents another domain
sion, early identification of potential problems, in which future advancements may occur. This
and feedback were facilitated by the availability aligns with the ethical imperative of maintaining
and ease of gauging project progress. rigorous privacy and confidentiality protocols
within the research process, particularly when
using centralised models of information storage
Future advances and collaboration and project management work. The
ethical implications transfer, storage and processing of information
within the aforementioned system is governed by
It is anticipated the further development of the strict HREC and other policy requirements. Ef-
system will occur through a user driven, evolution- fective system security, regulating access and use,
ary response. Increased and synchronised commu- is paramount to fulfilling the ethical and policy
nication avenues are predicted with a decreasing requirements of privacy, confidentiality and cost
need of physical presence. This is particularly effective use of research resources. Although
evident with the advent of increased access and the system currently uses best practice security
data throughput, as a result of increasing band- practices, such as encryption, user access and
widths and the emergence of new and maturing password, security certificates and user training,
technologies (Choudrie & Dwivedi, 2007). It is the advent of increased processor and computa-
envisaged that access to the research collabora- tional capabilities and increased internet users
tion tool will occur through more diverse means, and connectivity means that security issues are a
and not be limited to a computer with Internet priority area for ongoing future advancement.
connectivity. This has strong positive implica- The software program has the potential to
tions for the processes of qualitative research, increase user engagement between researchers,
as many research activities are conducted in the health professionals and stakeholders in qualita-
setting of the participant and away from tradi- tive research processes, especially for groups
tional computing equipment. There is potential to whose lack of access to traditional research
limit the redundant activities presently associated institutions may have formerly acted as a bar-
with scheduling, interviewing and recording by rier to participation. This use of IT solutions
allowing handheld or portable devices access to within qualitative research has the potential to
the software. bridge geographical and social communication
Additionally, the immediacy of information gaps, opening up the potential for increased
access between research staff and the potential for collaboration and information sharing between
real-time communication is an important avenue researchers worldwide. From IPP-SHR’s experi-
for future advances. Instant Messaging and Voice- ence implementing and using such a program, it
Over-Internet-Protocol advances have the poten- is evident that the integration of IT management
tial to contribute to stronger group collaboration, structures into a qualitative research program
communication and a sense of community. The can be instrumental in bridging geographical
use of such technologies also has the potential to constraints and fostering global research col-
lower communication costs and providing greater laboration. The potential of this new technology
access to research opportunities for marginalised is clearly evidenced by the way in which it has

678
New Paradigms

enabled IPP-SHR to effectively engage in research Barry, C. A. (1998). Choosing Qualitative Data
collaboration that spans national boundaries and Analysis Software: Atlas/ti and Nudist Compared
multiple time zones. Sociological Research Online, 3(3). Retrieved
March 17, 2007, from http://www.socresonline.
org.uk/socresonline/3/3/4.html
Conclusion
Choudrie, J., & Dwivedi, Y.,K. (2007). Broadband
impact on households consumers: online habits
This chapter has presented, through the medium
and time allocation patterns on daily life activities.
of a case study, insights gained from the practical
International Journal of Mobile Communications,
experience of developing and implementing an
5(2), 225–241. doi:10.1504/IJMC.2007.011817
original web based collaborative research tool to
assist and enhance the management of an existing Diaz, M., & King, J. (2002). How CMM impacts
qualitative research program. This tool is based quality, productivity, rework and the bottom line.
on a centralised model of information distribution CrossTalk, 15(3), 9–16.
and access and was designed following a reduc-
IEEE – Software Engineering Standards Commit-
tionist analysis of existing research processes and
tee of the IEEE Computer Society. (1998). IEEE
procedures. The ways in which the integration
recommended practice for software requirements
of responsive IT processes into the management
specifications. USA
of a large international research program have
removed redundancies and increased automation Latimer, J., et al. (Eds.). (2003). Advanced Qualita-
and research efficiency has also been discussed. tive Research for Nursing. Oxford, Malden, MA:
The program opens up avenues for increased Blackwell Publishing.
research participation and collaboration and
McGrath, P., & Holewa, H. (2006). Missed Op-
makes an important contribution to overcom-
portunities: Nursing insights on end-of-life care
ing the challenges that a fragmented, globalised
for haematology patients. International Journal
environment poses.
of Nursing Practice, 12(5), 295–301. doi:10.1111/
j.1440-172X.2006.00585.x
References Neus, A. (2007). Managing Information Quality in
Virtual Communities of Practice. Lessons learned
AIATS. (2000). Ethical guidelines for research. for a decade’s experience with exploding Internet
Australian Institute of Aboriginal and Torres Strait communication. The 6th International Conference
Islander Research. Retrieved March 18,2007, from on Information Quality at MIT. Retrieved March
http://aiatsis.gov.au/research_program/gratns/ 18, 2007, at http://opensource.mit.edu/papers/
gratns_assests/ ethics_guidelines.pdf neus.pdf
Australian Federal Government. (1988). Pri- NHMRC. (2003). Values & ethics: guidelines for
vacy Act 1988 Australia Privacy Policy. Aus- ethical conduct in Aboriginal and Torres Strait
tralian Federal Government. Retrieved March Islander health research. National Health and
18,2007, from http://www.comlaw.gov.au/com- Medical Research Council. Retrieved March 18,
law/management.nsf/lookupindexpagesbyid/I 2007, http://www.nhmrc.gov.au/publications/
P200401860?OpenDocument synopses/e52syn.htm

679
New Paradigms

Polit, D., & Hungler, B. (1995). Nursing research: Psycho-Social Research: Research processes
principles and methods. 5th Ed. Philadelphia: which aim to explore and document social and
Lippincott. psychological aspects of the human experience.
Qualitative Research: A non-numerical re-
Streubert, K., & Carpenter, D. (1995). Qualitative
search methodology which aims to describe and
research in nursing: Advancing the humanistic
understand, rather than explain, human behaviours
imperative. Philadelphia: Lippincott.
and experiences.
Redundant Processes: Unnecessary process
or tasks that through analyses of such process
KEY TERMS can be removed without affect the output of such
tasks.
Information Technology (IT): As defined Server Side Technology: A form of web server
by the Information Technology Association of technology in which users’ requests are fulfilled
America, IT is “the study, design, development, by running a script directly on the web server to
implementation, support or management of generate dynamic HTML pages. It is used to pro-
computer-based information systems, particularly vide interactive web sites capable of interfacing
software applications and computer hardware.” with databases and other data stores.
Phenomenology: A method of inquiry based
around the exploration, description and analysis
of a particular phenomenon, untainted by presup-
posed theories, beliefs and assumptions.

This work was previously published in Handbook of Research on Digital Information Technologies: Innovations, Methods,
and Ethical Issues, edited by T. Hansson, pp. 57-67, copyright 2008 by Information Science Reference (an imprint of IGI
Global).

680
681

Chapter 3.2
Adaptability and Adaptivity in
The Generation of
Web Applications
Raoudha Ben Djemaa
MIRACL, Tunisie

Ikram Amous
MIRACL, Tunisie

Abdelmajid Ben Hamadou


MIRACL, Tunisie

Abstract evaluation and a validation of GIWA are given in this


article to prove our adaptation.
This article proposes a generator for adaptive Web
applications called GIWA. GIWA‘s objective is to
facilitate the automatic execution of the design and INTRODUCTION
the generation ofAdaptable WebApplications (AWA).
Characteristically, the effort in this work has to be The growing demand for data-driven Web ap-
pursued with special attention to both issues applied to plications has led to the need for a structured and
AWA: adaptability and adaptivity. The architecture of controlled approach to the engineering of such
GIWA is based on three levels: the semantic level, the applications. Both designers and developers need
conceptual level and the generation one. Using GIWA, a framework that in all stages of the engineering
designers specifies, at the semantic level the features process allows them to specify the relevant aspects
of Web application. The conceptual level focuses on of the application. The engineering becomes even
the creation of diagrams in WA-UML language; the more complicated when we include notions of
extended UML by our new concepts and new design adaptation. Here, we address both adaptations
elements for adaptation.At the generation level, GIWA during the presentation generation, for example to
acquires all information about users’ preferences and reflect user preferences or platform used, as well
their access condition. Consequently, the generated as adaptation inside the generated presentation.
pages are adaptable to all these information. An The need for adaptation arises from different
aspects of the interaction between users and Web

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Adaptability and Adaptivity in The Generation of Web Applications

applications. Users’ categories which deal with the functional space is determined by a semi
these systems are increasingly heterogeneous due automatic algorithm called AGCA.
to their different interests, preferences, and the use In the Conceptual Design step (Ben Djemaa,
of number of devices (PC, WebTV, PDA, WAP 2008), the functional space for each audience
phone, etc...). User’s preferences and interests can class is represented using traditional conceptual
be deduced from his and browsing history. modeling: use case diagram, sequence diagram,
Adaptive Web engineering is meant to pro- class diagram, etc. In GIWA, conceptual model
vide a systematic and disciplined approach for is represented in a specific notation called Web
designing, generating and maintaining adaptive Adaptive Unified Modelling Language (WA-
Web applications (Cingil, 2000). For this reason, UML) (Ben Djemaa, 2008). This new notation
recently several models and methodologies have increases the expressivity of UML while adding
been proposed for supporting the development of labels and graphic annotations to UML diagrams.
adaptive Web applications. The main goal of such This extension of UML defines a set of stereotypes
models is to help designers to reason in a struc- and constraints, which make possible the design
tured way about aspects that are specific to hy- of conceptual model. These models are translated
permedia, such as links, structure and navigation, and exported in XML files in a data repository.
and to express adaptation in the design process. The adaptation design level (Ben Djemaa,
Moreover, such models and methodologies should 2007) is based on the profile model, which
help engineers to manage the overall complexity takes into account the user’s devices capabili-
of Web development which requires a variety of ties (hardware and software), Users’ preferences
activities, such as organizing the structure, choos- presentation (desired layout, navigation patterns,
ing the contents and the presentation modality, etc.) and personal information (eg. Age, sex,
some of them involving automated generation of language, etc...).
Web page (Brusilovsky, 1998). So, methodologies In this article we concentrate on the genera-
usually provide guidelines for performing such tion level. At this level the designer is invited to
activities and suitable models for expressing the instantiate previous models using the specific
results of such operations. interfaces offered by GIWA. Only the aspects
In our previous works (Ben Djemaa, 2006a; related to the two first levels (requirement analysis
2006b, 2006c; Ben Djemaa, 2007; Ben Djemaa, and conceptual design) are instantiated by the
2008) we have presented a methodology for AWA designer. Information related to the devices’ ca-
which guides the designer through different steps pabilities are dynamically captured by the system
of the design process, each of them yielding a spe- (using Logs files) and then stored in the profile
cific model that is being interpreted by the GIWA model. At the end of the step of instantiation, the
tools. The GIWA methodology is based on several GIWA deployment can be launched.
following steps: requirement analysis, conceptual Characteristically, the effort in this work has
design, adaptation design and generation. to be pursued with special attention to both is-
The requirement analysis step (Ben Djemaa, sues applied to AWA: adaptability and adaptivity.
2005) represents the application domain. This Adaptability can be defined as the facility of an
step expresses the purpose and the subjects of the application to be configurable according to a set
Web application through the functionality model of decisions taken by the user, which usually de-
and defines the target audience through the audi- fine his preferences and/or background. Whereas
ence model. The result of these two models is a adaptivity denotes the capacity of the application
set of audience classes together with an informal to alter the profile model according to the user’s
description of their functional space. In GIWA, behaviour during the application run and adapt

682
Adaptability and Adaptivity in The Generation of Web Applications

dynamically to the current state of the user model All these methodologies were originally
to any user. designed for Adaptive Hypermedia Application
The article is structured as follows. In Section (AHA) and do not deal comfortably with Adaptive
2 we provide an overview of related works. In Web Application (AWA). These methodologies are
section 3 we present an overview of the different very much data-driven or implementation oriented
models of GIWA. Section 4 presents the architec- and do not covers the lifecycle of adaptive Web
ture of GIWA and some examples of interfaces applications. Still, most solutions have been origi-
which illustrated the prototype. In section 5 we nally developed for a manual hypermedia design
present our experimental design and the carrying process and are not particularly well-suited in the
out of the experiment using the design. Finally, context of automated hypermedia design.
section 6 concludes the article and suggests future Methodologies like RMM, UWE or AMA-
research directions. CONT are not specifically targeted to support
dynamic adaptation. For these methodologies,
personalization means that the application ac-
Related WorkS knowledges the user’s situation and its informa-
tion delivery are adapted. They may be able to
For a long time, Web application engineering has solve adaptability problems to some extent but
been synonymous with ad hoc development and they do not address the real problem of adaptivity
not supports a systematic process. Aspects like relating to the devices’ capability (hardware and/
adaptation and generation process complicate the or software). On the other hand, most of the cur-
design process of Web application engineering and rently existing methodologies lack a profile model
bring its complexity beyond the level that is easily that would allow for the design of truly adaptive
handled by a single human developer. Therefore, Web applications. In fact, this model can play a
to support a systematic development process, a significant role in such applications: our aim is
strong methodology (supported by a suite of tools) to include this aspect in the personalization of
can help to keep the design process at a practical the hypermedia presentations that get generated.
level. Recently, different approaches for model- Generating adaptive presentations requires a
ing and engineering adaptive hypermedia system clean separation of concerns, as is advocated in
have emerged. Approved hypermedia design (Frasincar, 2002).
principles, such as those defined in OOHDM
(Schwabe, 1998) or in RMM (Isakowitz, 1995;
Isakowitz, 1998) have been enhanced with the Overview of GIWA models
notions of adaptation and personalization in a
further extension of OOHDM (Rossi, 2001) or the In GIWA, Web applications data are defined by
RMM-based Hera methodology (Frasincar, 2001; different models such functionalities model, audi-
Frasincar, Houben, 2002). UWE (Koch, 2000, ence model and profile model.
2001) included a design methodology for adaptive
hypermedia applications (AHDM) and a develop- The Functionalities Model
ment process for such applications (AHDP). In
the AMACONT project authors have introduced Met with the increasing needs of Web applications
a component-based XML document format (Fiala, users, we propose a functionalities model, gath-
2003). This project enables to compose adaptive ering users’ informational and functional needs.
Web applications by the aggregation of reusable In this model, functionalities are classified into
document components. three functional classes: Static Informational (SI),

683
Adaptability and Adaptivity in The Generation of Web Applications

Dynamic Informational (DI) and Professional (P) At the lowest level of the Functionalities Model,
ones. The Static Informational functional class the concept “Functional Space” provides each
gathers all functionalities enabling the users to actor with a list of authorized functionalities.
have access to the system to acquire static infor- In Figure 1 we present the Functionalities’
mation being in a specific URL Web page. This meta-model (using the UML notation). This figure
class translates the set of the static informational illustrates the inheritance and composition links
users’ needs. Using hypertext and hypermedia existing between the concepts constituting the
links, the user is capable to exploit a functionality Functionalities Model (Functional Class, Func-
of information’s consultation. tionality and Functional Space). We notice that
The Dynamic Informational functional class the Informational Functionality class executes
regroups all functionalities enabling the users link (hypertext and hypermedia link classes)
to have access to the system to acquire dynamic and the Informational Functionality class uses
information. This class translates the set of the a query of consultation (Query research class)
dynamic informational users’ needs. Using tech- which makes it possible to select and to organize
niques of research by a search engine, the user the presented information. In this case, the Query
is capable to exploit a research functionality of class is composed of several classes representing
information. various clauses of a simple query like in SQL. In
The Professional functional class is devoted to this meta-model we present:
the representation of the users’ functional needs.
Indeed, Web applications are based on technolo- • The SQL Select clause by a Projection class
gies which make their contents dynamic, enabling, which defines the expected structure of the
thus, the user to modify the applicative state of the query result;
server by carrying out a set of functionalities. • The SQL From clause by the Source class
Each of these functional classes, represented which introduces the collections from which
above, will be decomposed by the designer into the result is built;
a set of elementary functionalities representing • The Where clause by the Selection class
both informational and functional needs. These which specifies a predicate allowing to filter
functionalities, describing every functional class, the collections;
are three:
A query produces a result modelled by a Result
• Static Informational Functionality (SIF) class. A relation of dependence exists between the
displaying a static Web page being in a classes Query and Result: any modification made
specific URL. to a query has a direct influence on the result.
• Dynamic Informational Functionality (DIF) In addition, the Profession functional class uses
displaying a dynamic Web page constructed another type of query which updates the database
from one (or several) “SELECT” server system (Query Updating class). It also implements
query (s). It is about a selection from the the composition of the classes representing the
database without affecting the applicative following clauses: The Insert clause represented
server’s state. by the class Insertion; the Delete clause repre-
• Profession Functionality (PF) displaying a sented by the class Delete and the Update clause
dynamic Web page constructed from one (or represented by the class Modification.
several) server query: UPDATE, ADD or In our approach, the Functionalities Model
DELETE. The execution of this functionality implementation is carried out in two stages. Dur-
affects the applicative server’s state. ing the first stage, the designer is invited to define

684
Adaptability and Adaptivity in The Generation of Web Applications

Figure 1. Representation of the meta-model of the model of the functionalities


functional class
code_CF
desc_CF

professional functional class Dynamic Inform ational functional class Static Inform ational functional clas s

functionnality
code_F functional space
desc_F code_FS
1. .* desc_FS
add_F()
modify_F() add_FS()
delete_F() modify_FS()
consult_F()
informational functionality

professional functionality
link
url
exécute
1
static informational functionality dynam ic inform ational functionality

utilis e
1..*
query
hypertxt link hypermedia link utilise
name modification(Update)
name 1
1 name_tab
query (updating)
query (research)
source(from)
name_tab 1
1

produit insertion produit suppression(delete)


(insert) name_tab
projection
(select) selection nom_tab 1
attributes (where)
result(updating)
predicate 1

result(research)

result

dis play()

the first three levels of the model. The fourth level, consult information (to execute an informational
relating to the functional space of the actors, is functionality) and possibly to modify the state of
approached only after having defined the list of the system (to execute a professional functional-
the actors involved in the application. However, ity). For example, in the case of an application of
this list is defined in the following section. library in line; Subscribers and Visitors are some
physical actors.
The Audience Model Logical actor represents a role played by a
human user (or a human user group) to assure
The goal of this model is to define the actors’ list the maintenance of the Web application. This
of a Web application. Indeed, for this type of ap- actor assures all actions and functionalities that
plication, besides the human users (defines as of participate in the configuration and the admin-
the physical actor) that exploits the Web system, istration of the system (to execute a professional
we can distinguish services (representing roles functionality). For example, in the case of a Web
played by the human users) or systems (devices, application of a library the Webmaster and the
data processing system, Web service,…). In this Bookseller are some logical actors.
context, to take account these distinctions, we System Actor represents a computer system,
propose three categories of actors classified as an access device or a Web service, etc. These
follows: physical actor, logical actor and system systems are connected generally to the applica-
actor. tion to provide news to the system or to update
Physical actor represents a human user (or hu- data. It is the external sources that will be charged
man user group) who visits the application Web. automatically in the system.
This actor interacts with the system to search or

685
Adaptability and Adaptivity in The Generation of Web Applications

In our approach, we have proposed an algo- tance and composition links existing between the
rithm which generates the list of these different concepts constituting the Functionalities Model
actors of the Web application starting from the (Functional Class, Functionality and Functional
concept of Functionality. We have presented in Space).
(Ben Djemaa, 2007), the process of operations To evaluate the audience model, we have
of this algorithm. proposed in (Ben Djemaa, 2005) an EPMA
The hierarchy of actors generated by this algo- (Evaluation Process of the Audience Model) which
rithm cannot describe a model for users because evaluates the result generated by the algorithm
the definite actors are not all human actors. There- of actor generation. This process is based on
fore, in the goal to define an audience model (that several mathematical symbols and formulas fol-
leans logically on the human users), we propose low to check the distribution of the informational
to differentiate the human users by the concept and functional needs between the actors for the
“audience class” to inhuman one (systems) and, application.
therefore to distinguish between “logical audience
class” and “physical audience class”. Profile Model
By definition, an audience class is a potential
user group which belongs to the application target In our approach we proposed a new model spe-
audience and which has the same informational cific for adaptive Web application called Profile
and functional needs. These classes are not nec- Model (Ben Abdallah, 2008). In Figure 3 we show
essarily disjoined (a user can belong to several the different dimensions treated in this model to
classes of audience). generate adaptive Web applications.
In Figure 2 we present the meta-model of our This model represents an abstract specification
audience model. This figure illustrates the inheri- of the presentation in terms of users’ profiles. In

Figure 2. Meta-model of the audience model


logical audience class 1..n 0..n physical audience class
inherit of
+is held by +play

audience class +is owner of


1
1..n
+execut 1 +corresponds
0..n
Functional class +belongs to Fonctionnality +belongs to +is constituted Functional space
1..n 1 1..n 1..n
correspond

SIF DIF PF execut external resource


0..n 1..n

686
Adaptability and Adaptivity in The Generation of Web Applications

Figure 3. Profile model


User Profile
Code_profil
blind : boolean
Session partially-sighted : boolean
deaf : boolean
id_session : int
id_user : int = adresse ip n 1
create()
state : string
Delete()
Update()

UP_Preferences
code_pref
1
1..n 1
UP_Context
UP_Presentation 1..n code_context
code_up_presentation UP_navigation 1
Code_navig 1 1
Update()
Typ_navig
1..n 1..n
Up_Contexte_Temporel 1..n Up_Localisation
UP_Context_Environmental code_context_temporel GPS
code_context_env : string time_FIS longitude
Protocol : string = HTTP | WAP time_FID latitude
time_FM

fact, this model needs to take into account the us- of attributes. In Figure 4 we defined three profiles
ers’ preferences (Up_preferences) and the user‘s components (UP_Network, UP_Hardware and
context (UP_context). The first aspect is composed UP_software). UP_Hardware component has a
of preferences of presentation and preferences of number of attributes (eg. Support_image speci-
navigation. The second aspect dealt with the en- fies if the device is able to display images and
vironmental context (hardware and software), the ScreenSize defines the dimensions of the device
temporal context (the preferred time to execute a display).
FIS, FID or FM functionality) and the localisation Both adaptability and adaptivity are considered
(GPS information’s) context. in our approach. These concepts are treated dif-
Users’ preferences presentation will be defined ferently in the profile model. Users’ preferences
through specific techniques of data presentation presentation will be defined through specific
and different media in the Web page. In fact, to charters (composition and graphical charters).
take into account the adaptation of the different Each of these charters can be choice by the user
media in a Web application we have defined a after the generation of his Web application (adapt-
model of media adaptation which presents pre- ability) and he can also modify some information
ferred choices for users for each type of media: at his system use (adaptivity). This dimension
visual, video and audio one. (cf. Figure 5). of the profile model will be detailed in the fol-
UP_context_Environmental has a number of lowing section. Users’ preferences navigation
components, each component grouping a number and information about the context of user is a

687
Adaptability and Adaptivity in The Generation of Web Applications

Figure 4. Environnemental context / user profile

Figure 5. Model of media adaptation in GIWA

UP_Presentation
code_up_presentation

Mettre à jour()

Mode
format

Visual Auditive
colors = all | mono Volume : float
Speciality = Mono | Stéréo | 3D
sound_duration
Video
Duration : int
Number_Slides : int
Streaming : boolean
Video_conf : boolean
Rafraichissement : int = nombre d'image par seconde
Differed : boolean voice synthesis
sub_title : String Suite of notes
Language = FR | EN | DE | ...
Type = man | women | child tempo : float
speed : float
stamps : float

Musique Jingle

Texte
Language = FR | EN | AR | DE | ...
caracter_seize : float
Length : int Graphisme
Police height : float
Style = normal | gras | italique width : float
code_caracters = ASCII | Unicode | IJS ...

Diagram
icon
Type = Histogram | Curve | Camembert cheese Image
type = 2D | 3D
animate : boolean

688
Adaptability and Adaptivity in The Generation of Web Applications

dimension which is automatically captured by we propose an AGL which supports the new
GIWA (adaptivity). design elements that we proposed. This AGL is
To generate Web interfaces, the user does not based on Argo_UML. Thus, in this extension of
need all dimensions or sub-dimensions or all in- Argo_UML, we introduce new types of diagrams
formation characterizing a dimension. A profile is to represent the different diagrams of our exten-
thus an instantiation partial of this model accord- sion, namely, WA-UML.
ing to the user’s needs, to the type of application The generation level focuses on the process
and to the execution environment. of Web page generation and describes how the
Different engines and processes are defined in generator GIWA dynamically adjusts to varying
GIWA to instantiate these models and to gener- user preferences into chosen implementation
ate adaptable interfaces Web. The architecture of platform (HTML, WML, SMIL, etc.).
GIWA is detailed in the following section.
Semantic Level of GIWA

Architecture of GIWA Figure 7 depicts different steps of GIWA in the


semantic level. For a designer modelling an adap-
GIWA‘s target is to facilitate the automatic execu- tive Web application using GIWA consists firstly,
tion of the design and the automatic generation of in instantiating the functionalities’ model.
adaptive Web interfaces. It should be possible to Once instantiated, this model is translated into
program the Web applications in such a way that it XML files to be stored in the data repository 
can automatically execute the process specified by and the system execute the algorithm of generation
the design. Figure 6 depicts three different activi- of audience classes  which built the audience
ties of the proposed generator GIWA: Semantic model. Then this model is validating by a specific
level, Conceptual level and Generation level. process called PVMA. The last is also translated
The semantic level instantiates specific data into XML file  which contains the functional
contents of the Web application defined by dif- space for each audience class. At this stage, the
ferent semantic model us functionality model, content is adapted to each audience class. But
audience model and profile model. to adapt the user’s presentation preferences, the
The conceptual level focuses on the creation designer is invited to instantiate the profile model
of diagrams in WA-UML. In fact, in this level using specific interfaces offered by GIWA that
treat respectively by following axes:

Figure 6. Architecture of GIWA

689
Adaptability and Adaptivity in The Generation of Web Applications

Figure 7. The semantic level of GIWA

• Personal information about the user like shortly demonstrate how the functional space
name, age, sex, language. for each user, is modeling. In fact, diagram and
• Users’ presentation preferences defined meta-model of WA-UML are detailed in our work
in by two charters called composition and in (Ben Djemaa, 2008).
graphical charters (defined in the model of In this article we propose an AGL which sup-
media adaptation (cf. Figure 4)). ports the new design elements that we proposed
in WA-UML. The last is based on Argo_UML
After instantiation, the profile model is ex- because it permits to guide the user in the use of the
ported in XSL files in the data repository. At UML notation through a mechanism of critiques
this stage GIWA treat the aspect of adaptability and help messages. In addition, the source code
which appears through the choice of a graphical of Argo_UML is available on the Web making it
charter and through the composition of page after possible to analyze its inner workings. Thus, in
generated users’ application. Figure 8 presents this extension of Argo_UML, we introduce new
some interfaces of GIWA to instantiated the types of diagrams to represent the different dia-
functionality model, the audience model and the grams of our extension. In Figure 9 we present the
profile model. different steps of the conceptual level in GIWA.
The XML files generated by the semantic
Conceptual Level level are extracted from the data repository 
to be imported into the AGL supporting the new
In the Conceptual level, the functional space design elements that we proposed.
for each audience class is represented using dif- This AGL is based on Argo_UML. So, in our
ferent conceptual modeling (use case diagram, use of Argo_UML, we introduce new types of
sequence diagram, class diagram, etc) which are diagrams to represent the new diagrams of our
represented in WA-UML (Ben Djemaa 2006c; extension WA-UML (Ben Djemaa, 2008). In these
Djemaa, 2008). While this article is not solely diagrams, the user can add, displace and copy the
devoted to the conceptual issue, we do want to different design elements as well as replace faces

690
Adaptability and Adaptivity in The Generation of Web Applications

Figure 8. Interfaces of the semantic level in GIWA

The instantiation of the functionality model

The application of the PVMA

Composition charter of GIWA

continued on following page

691
Adaptability and Adaptivity in The Generation of Web Applications

Figure 8. continued

XML file of the audience model

Figure 9. Conceptual level of GIWA

and publish their properties as used in Argo_UML. Generation Level


Figure 10 shows some examples of new icons of
WA-UML presented by Argo_UML. The previous sections dealt with the engineering
All conceptual diagram of WA-UML can be de- process of GIWA. This section focuses on the
scribed with the new AGL based on Argo_UML. process of generation of adaptive Web applica-
The last diagram is translated in XML files . tions and describes how the system is dynamically
In Figure 11 we present some diagrams of adjusted to varying audience classes.
GIWA at the conceptual level. Us an example The generation level focuses on the process of
we present conceptual diagrams related to E- Web page generation and describes how the gen-
commerce application. erator GIWA dynamically adjusts to varying user

692
Adaptability and Adaptivity in The Generation of Web Applications

Figure 10. New icons in WA-UML

(a) New actors in WA-UML

(b) New use cases in WA-UML

(c) New classes in WA-UML

preferences into chosen implementation platform devices user (PC, PDA, cell phone or desktop
(HTML, WML, SMIL, etc.). The target of this browse) .
step is to facilitate the automatic generation of In adaptation media rules (Abdallah, 2008),
adaptive Web interfaces. It should be possible to we use the media properties defined in profile
program the Web applications in such a way that model, e.g. we can test if the dimensions of a
it can automatically execute the process specified particular image fit the size of the screen. Note
by the design. The tool is based on a collection that the adaptation based on user preferences can
of engines, which interpret the models provided be treated in the same way as the adaptation based
by the designer during the generation process. on device capabilities.
(cf. Figure 12). We proposed several adaptation rules for dif-
According to the user/devices profile (refers ferent Medias (text, image, sound and video).
aspect of adaptivity) is captured by the GIWA R1 is an example of an adaptation rule for
using data from logs files to be stored on the text media. This rule dimensions the size of the
server according to a RDF vocabulary (W3C, screen to adapt the size of the text to the hardware
2002)  and then to instantiate the profile model device or to user who is partially-sighted person
(user/devices profile) by specific capabilities (e.g. or blind person. In appendix 1 we present the rest
bandwidth, display resolution,…). Finally, XML, of adaptation rules for text media.
RDF and XSL files are extracted from the data R1: ClientPage.Media.Texte.size_caracter >
repository  and they are sent to the PARSER
in order to apply some adaptation rules for each
height _ screen * width _ screen )
media (text, image, sound and video) and finally length _ text
to publish the HTML page corresponding to the ∧ (¬ UserProfile. partially-sighted person

693
Adaptability and Adaptivity in The Generation of Web Applications

Figure 11. Examples of WA-UML diagrams

An example of WA-UML use case diagram

An example of WA-UML class diagram

continued on following page

694
Adaptability and Adaptivity in The Generation of Web Applications

Figure 11. continued

An example of WA-UML PHP class diagram

Figure 12. Generation level of GIWA

∨ ¬ UserProfile. blind person )


∧(UserProfile.Up_Context.Up_Context_En- R 2 presents an example of adaptation rules
vironmental.DeviceCaracteristic.Up_Hardware. for an image media. This rule adapts the size of
support_text)) the image to be displayed by the device. In ap-
→t1  : Resize_text(ClientPage.Media.Text. pendix 2 we present the rest of adaptation rules
size_caracter, E( for image media.
R2: ClientPage.Media.image.Hauteur > User-
height _ screen * width _ screen )) Profile.Up_Context.Up_Hardware.hauteur ∨ Cli-
length _ text

695
Adaptability and Adaptivity in The Generation of Web Applications

entPage.Media.image.largeur > UserProfile.Up_ rest of adaptation rules for video media.


Context.DeviceCharacteristic.Up_Hardware. r18:¬UserProfile.Up _Context.Up _Con-
Largeur →t2:Resize_image(ClientPage.Media. text_Environmental.DeviceCharacteristic
image,UserProfile.Up_Context.Up_Context_En- .Up_Hardware.support_Video ∧UserProfile.
vironmental.Up_Hardware.hauteur,UserProfile. Up_Context.Up_Context_Environmental.Up_
Up_Context.Up_Context_Environmental.De- Hardware.support_image →t18Convert_video_
viceCharacteristic .Up_Hardware.largeur) to_image(ClientPage.Media.video,UserProfile.
R 3 presents an example of adaptation rules Up_ preference.Up_ presentation.Charte.Mo-
for sound media. This rule deletes the sound in dalite.Visual.graphic.format,number_image:1
the Web page where the device doesn’t support it default)
or where the user is a deaf person. In appendix 3 In Figure 13 we present the simulation of GIWA
we present the rest of adaptation rules for sound for a user who connected with a PC.
media. In Figure 14 we present other interfaces of
R3: UserProfile.deaf ∨ ¬ UserProfile. GIWA related to user who connected to the system
Up_Context.Up_Context_Environmental.De- using cell phone (Nokia SDK 5100).
viceCharacteristic .Up_Hardware.support_sound
∨ ClientPage.Media.audio.sound.jingle →t3 de-
lete_sound(ClientPage.Media.audio). Experimentation and
R 4 presents an example of adaptation rules evaluation
for video media. This rule is applied to substitute
a video by an image if the device is enabling to Evaluating systems is a difficult task, and it
display this video. In appendix 4 we present the

Figure 13. Simulation of GIWA using PC

696
Adaptability and Adaptivity in The Generation of Web Applications

becomes even more difficult when the system is Subjects were videotaped, and the computer
adaptive. It is of crucial importance to be able to screen was recorded on the same videotape. The
distinguish the adaptive features of the system subjects’ actions were tracked using DRUM
from the general usability of the designed tool. (Diagnostic Recorder for Usability Measurement)
The evaluation of our GIWA system was de- and descriptive statistics of task completion time,
signed to address the goals of the system, namely actions performed inefficient use of the system,
to help users to design and generate adaptive web and others could be easily computed using this
applications. So firstly, we wanted to measure tool. There were 8 subjects in the study, 3 female
the usability of GIWA and testing the ability of and 5 male, all employed at the laboratory that had
the process, during its three levels (i.e. semantic, developed and used the target domain SDP. All
conceptual and generation), to design and gener- had experience with World Wide Web (WWW)
ate adaptive interfaces. Secondly, we were also and hypermedia development. Subjects spent in
interested in subjects’ own evaluation of how well total, approximately three hours in carrying out
the adaptive system worked compared to the non- the steps of the experiment, out of which two
adaptive one, and whether they felt in control of hour was spent testing the three level of GIWA.
the adaptive parts of the system. A last difficulty The rest of the time was used for questions on
in making studies of adaptive systems is in the the subject’s background, a small diagnostic test
procedure of the study. of their understanding of certain concepts in
the on-line manual before using our system and
Design of the Experiment after using our system, and finally answering
some questions about their preferences regarding
The study was done in the laboratory of MIRACL. GIWA and the adaptive versus the non-adaptive

Figure 14. Simulation of GIWA using cell phone (Nokia SDK 5100)

697
Adaptability and Adaptivity in The Generation of Web Applications

system. Details of the steps of the experiment are


presented below. In the third of the three above mentioned steps
Each of the subjects followed three steps in the subject was given the XML file generated by
this experiment, related to each level of GIWA. the second step and asked to use firstly a Pc and
We proposed for them an example of web applica- secondly, a cell phone, to complete the following
tion related to the “teaching gate” of our institute. three tasks:
The two first steps related to the semantic and the
conceptual level that were designed to test the • Task 7: connected to GIWA by taping: http://
explanations provided by the system rather than localhost:8080/GIWA/index.html (for
test the usefulness of the system as such. These PC) http://localhost:8080/GIWA/index.
two steps also served as a means to introduce wml (for cell phone).
subjects to the development of Web applications. • Task 8: create an account and then choose
The third step concerned the generation level that a profile or create a new profile.
was realized for both of the two devices: the PC • Task 9: choose the XML file related to the
and the cell phone. conceptual model of the application “teach-
In the first step, a subject was given a set of ing gate” and generated the results by PC
functionalities related to the “teaching gate” with and cell phone.
different types (FIS, FID, and FM), and was asked
to use the generator to complete the following The design goal of these nine tasks (1 to 9)
three tasks: was to oblige the users to actively use all the com-
mands provided, while at the same time expose
• Task 1: Instantiate the functionality model their own patterns of command sequences when
by adding, for each functionality, the code performing the various tasks.
and the description, and generate the XML
file for this model. Results of Carrying out the
• Task 2: Apply the AGAC (Algorithm of Experiment
Generation of Audience Classes).
• Task 3: Apply the PVMA and generate the Usability of the System
XML file of the audience model.
Our results are divided into those concerning:
In the second step the subject was given the
XML file generated by the first step and asked to • Some remarks concerning the task comple-
use the ArgoWA-UML to complete the following tion time, where we can see a weak tendency
three tasks: that users spend some time to achieve the
first task.
• Task 4: Import the XMI File related to the • The actions in ArgoWA-UML that the sub-
semantic level and created the use cases of jects have to do (clicking on icons, making
the application. menu-choices, clicking on association, etc)
• Task 5: create the sequence and classes dia- to build WA-UML conceptual diagrams
grams with the new icons of WA-UML. are less than the action of building UML
• Task 6: choose a Web language (PHP, JSP, diagrams. Thus, this confirms the result
ASP) and generated the technical class that conceptual level of GIWA is a semi
diagram. automatic one.
• Adaptive Web pages generated by cell

698
Adaptability and Adaptivity in The Generation of Web Applications

telephone are more interesting than those will form the tool that in its whole meets the us-
generated by a PC for most of users. ers’ needs and individual differences.
• The subjects’ satisfaction with the system, There are few studies of adaptive systems in
where they compare the result (page gener- general and even fewer of adaptive hypermedia.
ated) and their preferences described at the In the studies of adaptive hypermedia by (Boyle
semantic level. 1994; Kaplan 1993) the main evaluation criterion
• The passage between the three levels of is task completion time. This should obviously be
GIWA was not too clear by some of users. one important criterion by which some systems
• Combination of design and generation in should be evaluated. In our case, though, the goal
GIWA is approved by users. of the adaptive hypermedia system is to generate
the right needs according to the user’s preferences.
User Satisfaction The time spent in retrieving information is not
as relevant as is the quality of the search and the
After the subjects had used the two variants of result. Apart from task completion time, Boyle and
GIWA, we asked them to provide their viewpoints al. also measured reading comprehension through
on various aspects of the generator. We did this a diagnostic test put to the subjects after having
through ten questions, and they were also asked to used their system. Kaplan and al. measured how
freely comment on various aspects of the system. many nodes the users visited - in their case the
For the ten questions the subjects put a cross on more nodes the users visited the better.
a scale grading from 1 to 7 - the interpretation
of the scales can be seen from the statements left
and right of the Table 1. Conclusion
In the evaluation column of the Table 1 we
present the interpretation of the scales which can The research described in this article targets the
be seen from the statements left and right. The x- support of automated generation Web interfaces
axes represents the number of user for each scale, in the context of adaptive Web application. Spe-
the y-axes represents the different interpretation cifically, for Web applications involving various
of user for each question. functionalities (informational or professional
In Table 1 we see the result of the queries on one), the implementation requires a structured
how the subjects perceived the adaptive system approach.
in using GIWA. As we can see, the subjects pre- This article develops a generator called GIWA
ferred the adaptive system (mean 5.25); they also supporting both design aspect and generation
like the combination of design and generation in one. The architecture of GIWA is based on three
GIWA (mean 5.25); and they felt that the system levels: the semantic level, the conceptual level
made good adaptations to their needs (mean 4.75). and the generation level.
Also, they claimed that they saw when the system The primary focus of the GIWA is to provide
changed the inferred task (mean 4.75). engineering support for adaptive Web applica-
Evaluating adaptive systems is often done tions that automatically generate hypermedia
through comparing a non-adaptive version of the presentations in response for each ad hoc user’s
system to an adaptive system. Our study is no requirements. GIWA guides the designer through
exception from this approach. Still, an adaptive the different steps of the generation process, each
system should preferably be designed in such a of them yielding a specific model that is being
way that the adaptivity is only one instrument in interpreted by the GIWA tools to achieve the
the repertoire of design techniques that together objective of automatic presentation generation.

699
Adaptability and Adaptivity in The Generation of Web Applications

Table 1.

NUM Questions Means Evaluation

How eff iciently


Good, the work
wo u l d yo u b e Badly, the generator
would be very ef-
1 able to Work with 4,87 gets in the way
ficient
GIWA?

No, it is very de-


Did you like us-
manding and un- Yes, I really liked
ing
2 5,37 pleasant using it.
GIWA?
to use.

Do you feel in con- No, it feels as if the Yes, I can make the
trol while using generator controls generator do what
3 4,25
GIWA? me. I want.

Did you easily get I got lost several I knew all along
lost in the informa- times and did not exactly where I
4 4,75
tion space? know where I was. was.

Did you find it No, in the begin- Yes, it is possible


easy ning it was ver y to get started right
5 5,12
to get started? difficult. away.

continued on following page

700
Adaptability and Adaptivity in The Generation of Web Applications

Table 1. Continued

Are the different


Steps of GIWA No, it is difficult to
Yes, they are eas-
easy to u nder- find the right icon
6 4,62 ily understood.
stand and use it.
and use?

No, there are


Did you like the Yes, the interface
too many
combination of de- of GIWA is very
7 5,25 details and it is
sign and genera- appealing.
confusing.
tion in GIWA?

Did you see when Yes, it was ob-


No, I never
the vious when the
saw that the
adaptations hap- generator changed
8 4,75 system
pened task and opened
changed.
in GIWA? new operation.

No, I repeatedly
Did the adaptive
had to change the Yes, it managed to
system
answers I got in get relevant
9 make good adapta- 4,75
order to find the information.
tions
right information.
to your needs?

Did you prefer the


The noadaptive
adaptive or the The adaptive Was
Was definitely bet-
10 nodaptive 5,25 definitely better.
ter.
system?

701
Adaptability and Adaptivity in The Generation of Web Applications

In GIWA, we have distinguished two kinds and the navigation patterns, learning from the
of adaptation: adaptability (implemented at the user’s behaviour.
semantic level) and adaptivity (implemented at
the generation level). Adaptability is based on
information about user preferences presentation Acknowledgment
(a.g. font color, page layout etc.) and user pref-
erences navigation stored in the profile model The authors would like to thank Ben Abdallah
before browsing starts. Adaptivity is considered Fatma, who helped in the design and the develop-
in GIWA to provide a system which is able to ment of a part of media adaptation.
automatically adapt a given presentation to the
user device capabilities (hardware and software
configuration). Information about device capabili- References
ties are captured from Logs Files and stored then
in the profile model. The prototype of GIWA has
Abdallah, F., Amous, I., Ben Djemaa, R., & Ben
been built by different engines and equipments.
Hamadou, A. (2008). Génération d’Interfaces Web
The prototype uses java interfaces to instantiated
Adaptable au Contexte et aux Préférences des
models at the semantic level. At the second level
Utilisateurs. Les huitième journées scientifiques
GIWA uses Argo_UML to create conceptual
des jeunes chercheurs en Génie Electrique et
diagrams which are translate in XML files to be
Informatique (GEI’2008) (pp. 103-112), Sousse,
generated in Web pages at the generation level.
Tunisie.
In the last, XML and RDF files are used to store
the data and XSL files to specify transformations Ben Djemaa, R., Amous I., & Ben Hamadou, A.
between consequent steps. (2005, September). Towards an approach for adap-
We have presented an experiment to evaluate tive Web applications’ requirements engineering.
our generator GIWA. Results showed significant The 7th International conference on Information
performance gains, both in design and in genera- integration and Web-based Application and Ser-
tion step. While our experiment provided several vices (IIWAS05) (pp. 19-21). Kuala Lumpur.
revealing results, there are many issues regarding
Ben Djemaa, R., Amous I., & Ben Hamadou A.
adaptive web generation worthy of further study.
(2006a, May). Design and implementation of
It would be interesting to experiment and evaluate
adaptive Web application. In 8th International
GIWA with blind and sighted users. Also interest-
Conference on Enterprise Information Systems
ing is how the framework generalizes between
ICEIS’06 (pp. 23- 27), Paphos Cyprus.
various users.
The work presented in this article provides Ben Djemaa, R., Amous I., & Ben Hamadou
different opportunities for future research. The A. (2006b, February). GIWA: A generator for
future work includes further developments to adaptive Web applications. In Proceedings of
facilitate adaptation at all levels of the genera- the Advanced International Conference on Tele-
tion process. Firstly we plan the extension of the communications and International Conference
ArgoWA-UML tool that is currently built to sup- on Internet and Web Applications and Services
port the semi automatic transition from design (AICT/ICIW 2006) IEEE (pp. 23-25), Guadeloupe,
models to a running implementation of adaptive French Carabeen.
Web applications. Secondly we plan to extend
adaptivity in GIWA in order to dynamically Ben Djemaa, R., Amous I., & Ben Hamadou A.
elaborate and modify both the functional space (2006c). Use case and meta-scenarios for model-

702
Adaptability and Adaptivity in The Generation of Web Applications

ling Adaptive Web Applications. In proceeding ence on Adaptive Hypermedia and Adaptive
of IEEE 1st International Conference on Digital Web-Based Systems (AH 2002) (pp. 133-142),
Information Management (ICDIM). 06-08 De- Malaga, Spain.
cember, 2006. Christ College, Bangalore, India,
Frasincar, F., Houben, G.-J., & Vdovjak, R. (2001).
pp. 283-288.
An RMM-Based Methodology for Hypermedia
Ben Djemaa, R., Amous I., & Ben Hamadou A. Presentation Design. In Proceedings of the 5th East
(2007, May). Adaptable and adaptive Web appli- European Conference on Advances in Databases
cations: From design to generation. International and Information Systems (ADBIS 2001), LNCS
Review on Computers and Software (IRECOS), 2151 (pp. 323-337), Vilnius, Lithuania.
2(3), 198-207.
Isakowitz, T., Stohr, A., & Balasubramanian,
Ben Djemaa, R., Amous I., & Ben Hamadou, E. (1995). RMM: A methodology for structured
A. (2008, April/June). Extending a conceptual hypermedia design. Communications of the ACM,
modeling language for adaptive Web applications. 38(8), 34-44.
International Journal of Intelligent Information
Isakowitz, T., Kamis, A., & Koufaris, M. (1998).
Technologies (IJIIT), 4(2), 37-56.
The Extended RMM Methodology for Web
Boyle, C., & Antonio, O. (1994). MetaDoc: An Publishing. Center for Research on Information
Adaptive Hypertext Reading System. User Mod- Systems.
els and User Adapted Interaction, (UMUAI), 4,
Kapla, C., Justine, F., & James, C. (1993). Adap-
1-19.
tive Hypertext Navigation Based On User Goals
Brusilovsky, P. (1998). Methods and Techniques and Context. User Modeling and User-Adapted
of Adaptive Hypermedia. In P. Brusilovsky, A. Interaction, 3, 193-220.
Kobsa, & J. Vassileva (Eds.), Adaptive Hypertext
Koch, N. (2000). Software Engineering for
and Hypermedia (pp. 1-43). Kluwer Academic
Adaptative Hypermedia Systems-Reference
Publishers.
Model, Modelling Techniques and Development
Cingil, I., Dogac, A., & Azgin, A. (2000). A Process. Ph.D Thesis, Fakultät der Mathematik
broader approach to personalization. Communi- und Informatik, Ludwig-Maximilians Universität
cation of the ACM, 43(8), 136-141. München.
Fiala, Z., Hinz,, M., Meissner K., & Wehneer, Koch, N. (2001). The Authoring Process of the
F. (2003). A component-based component archi- UMLbased Web Engineering Approach. 1st In-
tecture for adaptive dynamic Web documents. ternational Workshop on Web-Oriented Software
Journal of Web engineering, 2(1), 58-73. Technology. The Tenth International Conference
on the World Wide Web, Hong Kong.
Fons, J., Pelechano, V., Pastor, O., Albert, M., &
Valderas, P. (2003). Extending an OO Method to Rossi, G., Schwabe, D., & Guimaraes, R. (2001)
Develop Web Applications. The Twelfth Interna- Designing Personalized Web applications. The
tional World Wide Web Conference (pp. 20-24), Tenth International Conference on the World
Budapest, Hungary. Wide Web, Hong Kong.
Frasincar, F., & Houben, G.-J. (2002). Hypermedia Schwabe, D., & Rossi, G. (1998). Developing
presentation adaptation on the semantic Web. In hypermedia applications using OOHDM. In Pro-
P. de Bra, P. Brusilovsky, & R. Conejo (Eds.), ceedings of Workshop on Hypermedia develop-
Proceedings of the 2nd International Confer- ment Process, Methods and Models (pp. 85-94).

703
Adaptability and Adaptivity in The Generation of Web Applications

Villanova-Oliver, M., Gensel, J., Martin, H., & April 30, 2002 from http://www.w3.org/TR/rdf-
Erb, C. (2002). Design and generation of adaptable schemadix.
Web information systems with KIWIS. In Proceed-
ings of the 3rd IEEE Conference on Information
Technology TCC-2002, Las Vegas, USA World
Wide Web Consortium.
Argo UML. http://www.tigris.org. eXtensible
Stylesheet Language, XSL Specification – version
1.0, W3C Recommendation. Retrieved October
15, 2001 from http://www.w3.org/TR/xsl/ World
Wide Web Consortium, 2002.
RDF Vocabulary Description Language 1.0:
RDF Schema, W3C Working Draft. Retrieved

704
Adaptability and Adaptivity in The Generation of Web Applications

Appendix

Appendix A: Adaptation rules for Text Media

R5: (ClientPage.Media.Text.color ⊄ UserProfile.Up_preference.Up_presentation.Charte.Modalite.


Visual.colors) ∨ (ClientPage.Media.Text.color∉UserProfile.Up_Context.Up_Context_Environmental.
DeviceCharacteristic.Up_Hardware.palette) → t5: Modify_color(ClientPage.Media.Text.color, color)
R6  :ClientPage.Media.Text.police≠UserProfile.Up_preference.Up_presentation.Charte.Modalite.
Visual.Text.police ∧ (UserProfile.partially-sighted person ∨ UserProfile.blind_man ∨ UserProfile.
Up_Context.Up_Context_Environmental.DeviceCharacteristic.Up_Hardware.support_Text∧UserProfile.
Up_Context.Up_Context_Environmental.DeviceCharacteristic.Up_Hardware.support_image) →t6 :Con-
vert_text_to_image(ClientPage.Media.Text)

Appendix B: Adaptation rules for Image Media

R7: (ClientPage.Media.image.color ⊄ UserProfile.Up_preference.Up_presentation.Charte.Modalite.


Visual.colors) ∨ (ClientPage.Media.image.color ∉ UserProfile.Up_Context.Up_Context_Environmental.
DeviceCharacteristic . Up_Hardware.palette)→t7 : color_image(ClientPage.Media.image.color, color )
R8: UserProfile.Up_Context.Up_Context_Environmental.Protocole_de_transfert=”HTTP” ∧ Client-
Page.Media.image.size_File > UserProfile.Up_Context.Up_Context_Environmental.DeviceCharacter-
istic.Up_Hardware.memory → t8 ∧/∨ t10 ∨/∧ t13 convert_gif (ClientPage.Media.image)
R9: ¬ UserProfile.Up_Context.Up_Context_Environmental.DeviceCharacteristic .Up_Hardware.
support_image ∧ ¬ UserProfile.Up_Context.Up_Context_Environmental.DeviceCharacteristic .Up_
Hardware.support_Text ∧ UserProfile.blind_man →t9 Delete (ClientPage.Media.image

Appendix C: Adaptation rules of Sound Media

R10: UserProfile.mal_entendant ∨ ¬ UserProfile.Up_Context.Up_Context_Environmental.De-


viceCharacteristic .Up_Hardware.support_sound ∨ ∃ClientPage.Media.audio.sound.jingle →t10 De-
lete_sound(ClientPage.Media.audio).
R11: ClientPage.Media.audio.size_File > UserProfile.Up_Context.Up_Context_Environmental.
DeviceCharacteristic .Up_Hardware.memory →t11 Stereo_to_mono(ClientPage.Media..audio) ∧ / ∨
Reduct_sampling(ClientPage.Media..audio) ∧ / ∨ Convert_sound(ClientPage.Media..audio, UserProfile.
Up_preference. Up_presentation. Charte .Modalite.auditive.format)

Appendix D: Adaptation rules of Video Media

R12: ¬ UserProfile.Up_Context.Up_Context_Environmental.DeviceCharacteristic .Up_Hardware.


support_Video∧ ¬ UserProfile.Up_Context.Up_Context_Environmental.DeviceCharacteristic .Up_Hard-
ware.support_image ∧ ¬ UserProfile.Up_Context.Up_Context_Environmental.DeviceCharacteristic.
Up_Hardware.support_sound →t12 Delete_video(ClientPage.Media.video)

This work was previously published in International Journal of Information Technology and Web Engineering, Vol. 4, Issue 2,
edited byG. I. Alkhatib and D. C. Rine, pp. 20-44, copyright 2009 by IGI Publishing (an imprint of IGI Global).

705
706

Chapter 3.3
Migrating Web Services in
Mobile and Wireless
Environments
Myung-Woo Park
Yonsei University, South Korea

Yeon-Seok Kim
Yonsei University, South Korea

Kyong-Ho Lee
Yonsei University, South Korea

abstract the effect of the code splitting on migration was


analyzed. Furthermore, to show the feasibility of
Mobile devices enabled with Web services are the proposed migration method, three application
being considered as equal participants of the Web scenarios were devised and implemented.
services environment. The frequent mobility of
devices and the intermittent disconnection of wire-
less network require migrating or replicating Web INTRODUCTION
services onto adjacent devices appropriately. This
article proposes an efficient method for migrat- Web services (Huhns & Singh 2005; Stal, 2006),
ing and replicating Web services among mobile which are independent from operating systems and
devices through code splitting. Specifically, the programming languages, have gained momentum
proposed method splits the source code of a Web as an enabling technology to realize business
service into subcodes based on users’ preferences processes on distributed network environments
for its constituent operations. The subcode with a such as the Web. Additionally, the technology of
higher preference is migrated earlier than others. mobile devices is continually developing and thus
The proposed method also replicates a Web service allows for a new form of Web services, that is,
to other devices to enhance its performance by mobile Web services (Schall, Aiello, & Dustdar,
considering context information such as network 2006; Sirirama, Jarke, & Prinz, 2006). However,
traffic or the parameter size of its operations. To it is difficult to provide Web services on mobile
evaluate the performance of the proposed method, devices seamlessly, since wireless and mobile

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Migrating Web Services in Mobile and Wireless Environments

environments still involve unstable connectivity, It involves developing the context model and
unlike the typical client-server environment. strategies or policies relevant to the migration
If Web services autonomously migrate among of Web services. If a migration of a service is
mobile devices in this unstable wireless environ- requested, the proposed framework collects the
ment, seamless provisioning of services would context information of neighboring devices based
be possible. When a service cannot be provided on the migration policy of the service. It computes
during movement of a device, it can be migrated the suitability values of candidate devices and
to an adjacent mobile device and provide its func- determines a target host. To establish a migra-
tionality continuously. Additionally, requests can tion policy, the proposed method is based on our
be distributed by replicating the service to other previous approach (Kim & Lee, 2007), which is
devices when the requests are concentrated on based on WS-Policy (2006). In this article, we
one device. Moreover, in the case of a client’s do not discuss the context model and migration
request for a service that takes large parameters strategies, but focus on describing the migration
such as bitmap image files as input, the service method itself.
itself can be replicated and executed on the client Meanwhile, the process of identifying when
side, resulting in saving resources. and where to migrate services is also an important
Recently, research on Web service migration issue. The migration of a service may be caused
has been performed. However, most of the research by context changes such as the battery shortage
targets desktop and wired environments or does and location change of a device. For the seamless
not consider constraints such as low bandwidth provisioning of a service, we have to determine
of wireless and mobile environments. Therefore, which device is the most suitable target host.
the research approaches might take much longer This process involves a mechanism to describe
time to migrate. context models and migration strategies, which
To resolve this issue, this article proposes a are relevant to the migration of Web services in
method for migrating Web services through code mobile environments. For example, a service
splitting. Specifically, an original code, which provider should be able to specify that if the CPU
implements the functionality of a service, is split usage-ratio of an origin host is over 80%, a service
into subcodes based on users’ preferences to its should be migrated to a new device, which has
constituent operations. The subcodes of higher enough processing power and supports J2ME. For
preference are migrated earlier to minimize the this purpose, we propose a method to establish the
latency of the operations of high priority and raise context model and migration policy to determine
the efficiency of Web services migration and rep- when and where to migrate services in mobile
lication in wireless and mobile environments. To environments (Kim & Lee, 2007). The method
evaluate the performance of the proposed method, determines a target host based on the migration
the effect of the code splitting on migration was policy of a service as well as the information
analyzed. Furthermore, to show the feasibility of collected from devices in the neighborhood of
the proposed migration method, three application the origin host that is hosting the service. In this
scenarios were devised and implemented. article, we do not discuss the process of determin-
Meanwhile, how to determine when and ing when to migrate a service to a specific target
where to migrate services is an important issue. host, but focus on how to migrate a service to a
The migration of a service may be carried out by target host efficiently.
the request of a service provider or the change The organization of this article is as follows.
of context information, such as the shortage of First, a brief survey of previous works is presented.
battery level and the location change of a device. Next, the methods of splitting Web services

707
Migrating Web Services in Mobile and Wireless Environments

codes and migrating and replicating them are shortcoming of this method is that the execution
described. Next, the effect of splitting codes on order of agent functions in each host must be
migration is analyzed through experiments, and fixed. The middleware for an ad-hoc network
three application scenarios are implemented to proposed by Bellavista, Corradi, and Magistretti
show the feasibility of the proposed migration (2004) can receive and execute a list of binary files
method. Finally, we summarize the conclusions from neighboring devices. Montanari, Lupu, and
and further studies. Stefanelli (2004) propose a middleware frame-
work which can implement migration policies
application independently.
RELATED WORK Previous works are mostly based on desktop
and wired environments, and therefore the latency
We summarize the conventional methods about of Web services are too long to apply to wireless
Web service migration and also discuss previous and mobile environments. Therefore, it is difficult
works concerning code mobility, which do not to deal with a service migration efficiently in mo-
target Web services but are related with migra- bile and wireless environments. In this article, we
tion. Pratitha and Zaslavsky (2004) propose a aim at presenting how to migrate services quickly
framework and strategies for migrating Web and to reduce the latency of service requests dur-
services. Based on context information such as ing migration.
network bandwidth, their method selects a target
server, to which a service should be migrated.
Moreover, their method defines service modules SPLITTING WEB SERVICE CODES
in advance according to context information and
selects a proper module to the corresponding We describe a splitting method of Web service
context. codes. For the splitting and migration method,
Hammerschmidt and Linnemann (2005) the framework of Figure 1 is proposed.
propose a migration framework, which supports The code splitter splits a Web service code and
Web services based on Tomcat-Axis. Particularly, saves them in a code repository in a component
since instances are supported, connections do not form. The migration manager encapsulates com-
need to be restarted after migration. Ishikawa, ponents and instances in a transportable form, and
Yoshioka, Tahara, and Honiden (2004) propose unpacks the encapsulated form and saves it at the
the migration of mobile Web services, which are resource space. The service execution manager
defined as composite Web services. Particularly, manages the execution of a service, which stops
agents that execute several services in a composite and restarts the service execution. The context
Web service can migrate to an appropriate host. manager collects context information from the
An endpoint of each service in a composite Web current and adjacent mobile devices and decides
service can be changed; however, the endpoint of when and where to migrate services. The channel
a composite Web service does not change. The manager takes charge of the establishment of net-
method of Mifsud and Stanski (2002) monitors work connections. The logger records the execu-
context information of available target hosts and tion history of the modules of the framework.
migrates Web services to a specific target host. How to split a Web service code by the code
Meanwhile, Kern, Braun, Fensch, and Rossak splitter in the proposed framework is shown in
(2004) propose a more rapid execution of mobile Figure 2. In our method, it is assumed that the
agents through splitting and arranging codes original source codes of Web services are pro-
which implement the functionality of agents. A portionate to their compiled codes.

708
Migrating Web Services in Mobile and Wireless Environments

Figure 1. The proposed framework

Resource Space

Code repository Instance manager


Component Instance

Component Instance

Component Instance

Service Migration Service Execution


Code Splitter
Manager Manager

Logger Context Manager Channel Manager

Figure 2. The process of splitting codes

Making a call graph Makes a call graph,


which describes function dependency

Analyzing a call graph Describes what functions each operation


depends upon through analyzing a call graph

Evaluates the costs of operation


Evaluating operation combinations
combinations

Splitting a Web service Split the code into sub-codes of minimum


code cost

Making and Analyzing a Call Graph computing the dependency among functions is
based on the method of Grove and Chambers
A code that implements a Web service is analyzed (2001). The functions exposed as operations in
and represented with a call graph, and the depen- a WSDL interface are recorded particularly and
dency among functions (or methods) is derived are used in the next analyzing step. After making
from the call graph. The proposed method for the call graph, it is analyzed in order to determine
functions, upon which each operation depends.

709
Migrating Web Services in Mobile and Wireless Environments

The analysis progresses following nodes, to which termined for efficient code splitting. Alternately,
each operation node is connected. Figure 3 illus- an original code is split based on which operations
trates the making and analyzing of a call graph. will exist in a class, and then the combination of
operations can decide which functions need to
Evaluating Operation be copied. 
Combinations and Splitting a Web The selection of the optimal code split is related
Service Code to both the total size of the codes split and users’
preferences to operations. If a code segment has
A condition required in code splitting is that each not arrived in a target server, it is impossible to
split code should contain at least one operation and invoke its operation. In this case, users’ dissat-
exists in a class form so that it can be compiled. The isfaction with the service would be raised. The
code split must have all the functions upon which proposed method transmits a code segment of
its operations depend. However, if operations higher priority earlier to minimize its service
depend on functions in common, code splitting discontinuance. So, the dissatisfaction rate of a
is impossible or inefficient. Copying appropriate code segment at a certain time is proportionate to
codes of functions can solve this problem. In the the time to migrate the total codes that include it
call graph of Figure 3, Operations 1, 2, and 3 and the code segments with a higher priority. The
commonly depend on Function E, and therefore dissatisfaction would be also proportionate to user
disjoint code splitting is impossible. However, preferences. User preferences can be statistically
if Function E is copied, it is possible to split the calculated from users’ previous requests by the
code, as shown in Figure 4. logger. In this article, the dissatisfaction with a
It is possible but somewhat inefficient to copy service migration is formulated and computed by
all the function codes that are used in common. a cost function as follows:
In Figure 4, Functions B, C, D, and E need to be n i
copied in order to split Operations 2 and 3, and Cost = ∑ ( pi × ∑S j ) (1)
those four functions are almost a half of all the i =1 j =1

function code, including operations. Therefore,


whether functions would be copied should be de-

Figure 3. An example of making and analyzing a call graph

Oper 1( ) Func B( ) 1 2 3
{ {
A( ); C( );
} } Dependent Functions:
F 1 A, E
A B
Func A( ) Func C( ) Dependent Functions:
{ …
Making Analyzing 2
a call graph a call graph B, C, D, E
E( );
} …
Functions C D Dependent Functions:
Oper 3( ) 3
B, C, D, E, F
Oper 2( ) {
{ B( );
B( ); F( ); E
} }

The code of a service

710
Migrating Web Services in Mobile and Wireless Environments

Figure 4. An example of copying functions

1 2 3
Dependent Functions:
1 A, E
F Dependent Functions:
A B 2
B, C, D, E

3 Dependent Functions:
C D B, C, D, E, F

E’ E

where i < j → pi ≥ pj, pi = pj → Si ≤ Sj, n is the When the migration of a service occurs, its
number of partitions, pi (0 ≤ pi ≤ 1) is the sum of endpoint should be changed as the corresponding
user preferences to operations in the ith parti- endpoint of the target host. In the case of repli-
cation, a new endpoint is added to the existing
tion, and Sj (0 ≤ Sj ≤ 1) is the total code size of
list of endpoints of an origin server. The context
jth partition. manager selects between migration and replication
The proposed method computes the costs of depending on context information.
every possible combination of code split. It finds
the combination of minimum cost that corresponds
to the optimal split of operations. The original
Migrating Web Services
service code is split according to the optimal
combination. Each partition of operations and their When the battery level of an origin server becomes
functions is compiled and saved as a component very low or a service cannot be provided any
in the code repository. more due to the location change of its device into
a nonservice area, the service should be migrated
to an appropriate target device. If a target server
MIGRATING AND accepts the migration request, the components and
REPLICATING WEB SERVICES instances in a transportable form are transmit-
ted, as shown in Figure 5. Specifically, existing
The proposed method handles both migration and requests on the execution queue of the service
replication. The former occurs when an origin execution manager are processed on the server,
server cannot provide a service any more because selected between the origin server and the target
of some reasons (e.g., battery shortage and a non- server for faster execution. This is basically the
service area). The replication of a service is done same with the case of a service request during
when the service needs to be copied to a different migration. At the same time, instances and their
server temporarily to maintain its quality. Example associated components on the resource space are
cases are as follows: 1) when network bandwidth transmitted in order of the user preference.
becomes more crowded and traffic slows down; Whenever a component and/or its instance ar-
2) when the number of requests from a particular rive, the service execution manager of the target
location grows rapidly; and 3) when a client wants server deploys the operations of the component.
to use a service with large parameters.

711
Migrating Web Services in Mobile and Wireless Environments

Figure 5. An example procedure of Web services migration


Before On After
migraion migraion migraion

Origin server

Target server

Component Instance

As soon as all the components and instances have In the case where the code has not been trans-
arrived at the target server, they are deleted on mitted yet, t’o corresponds to the time required
the origin server. If the components and instances to respond to the client request and transmit the
of a service have not arrived at the target server instance and component of the operation to the
due to any reason, the process of migration halts. target server. t’’n includes the time to forward the
Additionally, the origin server searches again component and client request to the target server
for a new target server. The former target server and execute the operation on the target server.
maintains the components and instances already It also contains the time for the origin server to
arrived for a certain period of time. After the time notify a client that the result should be received
period, the target server deletes them. from the target server and for a client to receive
If a service is requested during its migration, the result from the target server. Figure 7(b) il-
the proposed method determines which server lustrates the case of Figure 6.
between the origin and target servers executes the
request faster. Figure 6 describes the control flow Replicating Web Services
of the process. In the case where the component
of the operation called has already transmitted The replication of a Web service is classified into
to the target server, to is the time to receive the two cases: the replications of a service to another
instance from the target server, execute it on the server, and to the client requesting it.
origin server, respond its result to a client, and
transmit the instance updated to the target server, Replication to Another Server
that is, synchronize two instances between the two
servers. t’n includes the time to notify a client that The method of replication is similar to that of
the request should be made to the target server and migration. If a target server accepts a replication
the time for the client to request the operation to request, the corresponding components and in-
the target server and receive its response. Figure stances are transmitted. However, all the instances
7(a) describes the case of Figure 6.

712
Migrating Web Services in Mobile and Wireless Environments

Figure 6. The flow of processing in case of a service call on migration

Notify a client t o call the


operation from the target
Y
server

t n < to Execute the operation on the


Y N
origin s erver and resend t he
instance updated to the
Has transmitted the target server
code of an operation?
Execute the operation on the
origin server and transmit
N N the instance and component
t n <to to the target server

Y Transmit the code and client


request t o the target server
and notify a client to receive
to = (Sres + 2Si)/bo , t o = (Sres + Si + Sc)/bo the result from the target
t n = N/bo + Sreq/bn + Sres/bn , t n = N/bo + (Sreq + Sc)/bo + Sres/bn server
Sreq: the size of a request message, Sres: the size of a result message
Si: the size of an instance, Sc: the size of a component,
N: the size of a notification, bo: the bandwidth of an origin server
bn: the bandwidth of a target server

Figure 7. Service requests during the migration process


Target Server Target Server
3. Migration with
a client’s request
Origin Server Origin Server

3. Re-request 4. Response
2. Notification
2. Notification
4. Response
1. Request
1. Request
Component
Client Client
Instance

(a) A component request after migration (b) A component request before migration

do not need to be transmitted, and after finishing replicated service. If a service is requested during
the replication process the components are not replication, the same processes with the case of
deleted from its origin server. The instances that migration are applied.
have been transmitted to the target server are Figure 8 is an example of a service replication,
deleted on the origin server. When a replication where an origin server with three components and
occurs, the origin server records the target servers five instances communicates with five clients. The
on the list of replications. Additionally, the origin context manager decides which server should be
server may request the target server to delete the selected as the target server and which instances

713
Migrating Web Services in Mobile and Wireless Environments

Figure 8. An example of Web services replication

Component Connection before replication


Target Server
Instance Connection after replication

Origin Server

Client
Client
Client

Client
Client

Figure 9. Component replication

Server

Component
1. Request of
Component Replication Instance
2. Replicaton

3. Instance
Re-replicatoin

Client

should be migrated. For example, after copying To do this, a client should examine the compo-
three components and three instances, the tar- nent information about an operation and compare
get server and three clients would reconfigure the sizes of the parameter and component. If a
their connections. The three instances would component were downloaded, an origin server
be removed from the origin server, resulting in would record the client on the replication list.
reducing traffic on the origin server. Likewise, the origin server may request the client
to delete the component.
Replication to a Client If a replication is requested during migration
or replication, a server, from which the compo-
If a service requires a large message as an input nent would be downloaded more quickly, should
parameter, a client may want to download the be determined, as shown in Figure 10. While t′′o
service code of a smaller size and execute it on indicates the download time of the component and
his or her device, as illustrated in Figure 9. instance from the origin server, t′′′n corresponds

714
Migrating Web Services in Mobile and Wireless Environments

Figure 10. The flowchart for a replication request during migration or replication

Has transmitted the Notify a c lient that t he


code of an operation? Y operation should be called
Y from the target server
N
t n <t o

N
Transmit the component and
instance to a client

t o = (Sc + Si)/bo
t n = N/bo + Sreq/bn + (Sc + Si)/bn
Sc: the size of a component, Si: the size of an instance
N: the size of a notification, Sreq: the size of a request message
bo: the bandwidth of an original server, bn: the bandwidth of a target server

to the time to notify the client that the component Second, concerning the number of functions called
and instance would be downloaded from a new by an operation, the test was done with services,
server and when the client would receive them which contain 10 operations and 20 functions.
from the new server. Here we varied the number of functions, which
are called by an operation, from 2 to 20. The
third test is about the number of operations in
EXPERIMENTAL RESULTS a service. Web services contain operations that
call 10 functions. Here the number of operations
We have experimented evaluating the effect of the was varied from 1 to 10. In our experiments, the
proposed code splitting approach on migration. dependencies between operations and functions
Additionally, to show the feasibility of the pro- were randomly selected. In each experiment,
posed migration and replication, three application the size of the functions also varied from 1 KB
scenarios were implemented and tested. The tests (kilobyte) to 100 KB.
were conducted on devices, which have an Intel(R)
PXA270 processor, a memory of 62.28M, and the Evaluation in Terms of the
Windows Mobile 2003 operating system. Dependency between Operations

Code Splitting If operations in a service share functions in


common, we can say that they depend on each
To evaluate the performance of the proposed code other. We define the dependency of operation a
splitting method, we measured the time to split in a service as:
codes from three different experiments. First, to
investigate how much the method is affected by the operation_dependency(a) = the number of opera-
dependency among operations, we experimented tions sharing functions in common with a
with test data, where Web services consist of 10 / the total number of operations in a service
operations and each operation calls 3 functions. (2)

715
Migrating Web Services in Mobile and Wireless Environments

Figure 11(a) shows the experimental results of was done with services, whose operations and
the relation between the average value of operation functions have the size of about 10 KB.
dependencies and the time taken to split codes. The
figure shows that operation dependency and split- Evaluation in Terms of the Number of
ting time are scarcely related. If a Web service Functions Called by Operations
has many functions shared by its operations, the
possibility of splitting would be low. Otherwise, The experimental result of Figure 12(a) shows that
the possibility of splitting would be high. Since as the number of functions called by operations
only the functions shared by operations are cop- grows, splitting time increases. Nodes in a call
ied, the number of copied functions is limited, graph represent functions or operations. The num-
resulting in increasing a limited amount of time. ber of functions called by operations determines
Consequently, the operation dependency is shown the number of edges among function nodes. As
to be unrelated to the splitting time. Moreover, the number of connections among nodes, we need
Figure 11(b) shows that the dependency is unre- more time to analyze and split the graph.
lated to the total size of the codes split. The test Figure 12(b) shows the relationship between
the number of functions called by an operation

Figure 11. An experimental result on operation dependency

Time (sec) Code size (KB)


2 300

1.8
250
1.6

1.4
200
1.2

1 150

0.8
100
0.6

0.4 50
0.2

0 0
30 34 40 40 44 44 50 52 56 60 30 34 40 40 44 44 50 52 56 60
Average of operation dependency (%) Average of operation dependency (%)

1KB 5KB 10KB 50KB 100KB Before splitting After splitting

(a) Splitting time (b) Total size of the codes split (function size
10KB)
Figure 12. An experimental result on the number of function calls
Time (sec) Code size (KB)
3.5 600

3
500

2.5
400
2
300
1.5

200
1

0.5 100

0 0
2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20
The number of functions called by an operation The number of functions called by an operation

1KB 5KB 10KB 50KB 100KB

(a) Splitting time (b) Total size of the codes split (function size
10KB)

716
Migrating Web Services in Mobile and Wireless Environments

and the size of the codes split. The test was as the number of operations increases. We have
done with services, whose operations and func- a plan to consider this matter to enhance the
tions have the size of about 10 KB. In the cases processing time of the proposed method.
where the numbers of the function called are 2
or 4—since the numbers are relatively small and Migration and Replication
each operation did not share functions—there was
no difference in the size of the codes split before To show the feasibility of the proposed migra-
and after splitting. No regularities were found tion method, we implemented three application
while increasing the number of functions called scenarios.
by an operation until it reached 12. In the case
of 14 or more functions, operations called more Migrating Web Services to a Mobile
than 70% of functions in this test. Splitting did Device
not happen since too many functions needed to
be copied otherwise. The first scenario is about the migration of a Web
service, which provides traffic information about
Evaluation in Terms of the Number of a particular bridge. As described briefly in Figure
Operations 14, we assume that a service offers information
about the traffic and weather situation of bridge
As shown in Figure 13, there is a close correla- continuously by migrating from a car leaving the
tion between the number of operations and the bridge to a car arriving at the bridge. The service
splitting time. An increase in the number of op- should be migrated to a new server because
erations results in an exponential growth of the its origin server is leaving the bridge and so it
splitting time since all possible combinations of cannot be provided any more. Additionally, the
operations need to be considered to determine connections from clients should be reconfigured
the code splitting of minimum cost. to the new car.
As a result, we find that the splitting time of a Figure 15(a) shows the log information of an
service is related with its code size, the number origin server, which hosts the service. When
of functions called by operations, and the num- the car is about to leave the bridge, a client calls
ber of operations, while the dependency among the service to get information on traffic speed
operations seldom affects the splitting time. In (getAveVel) and temperature (getTemperature).
particular, the splitting time grows exponentially The service is being migrated to a new server

Figure 13. An experimental result on the number of operations


Time(sec)
1.4

1.2

1
1KB
0.8 5KB
10KB
0.6 50KB
100KB

0.4

0.2

0
1 2 3 4 5 6 7 8 9 10
The number of operations

717
Migrating Web Services in Mobile and Wireless Environments

Figure 14. A scenario of service migration

The new car The original car

Migration

New
connection

Previous
Connection

Client

Figure 15. Log information of servers

arriving at the bridge. Figure 15(b) shows the log Figure 16 provides a brief description of this
information of a car, to which the service has ar- scenario. Visitors or clients are served from the
rived and its two operations, that is, getWindVel museum server, which collects the information
and getStatusOfRoad, have been requested by and adjacent guiding devices and maintains their
the client list. If the requests to the origin server increase
suddenly, the origin server will an appropriate
Replicating Web Services to a Mobile device, to which its service should be replicated. If
Device a target device accepts the replication request from
the origin server, the service code is replicated
This scenario is about a service replication for to the target device and the connections between
providing art information seamlessly in a museum. the origin server and clients are reestablished in
The museum in the scenario provides a Web ser- order to provide the service seamlessly.
vice, which offers information about artists and While Figures 17(a) and 17(c) illustrate the
their artworks, and if necessary, replicates the interaction between the museum and its visitor
service to guiding devices, resulting in reduced before replication, Figures 17(b) and 17(d) show
network traffic. the interaction between the device of a guide and

718
Migrating Web Services in Mobile and Wireless Environments

Figure 16. A scenario of service replication

Museum server

Guide’s device
Replication

New
Previous Connection
Connection

Visitor

Figure 17. Screenshots of the server and client

(a) Client before replication. (b) Client after replication.

(c) Museum server before replication. (d) Guide device after replication.

719
Migrating Web Services in Mobile and Wireless Environments

the visitor after replication. Once a client calls is due to the fact that the instance of the service
an operation, the service provider maintains the was replicated to the target server.
instance and thus enables the client to call other As described before, an origin server maintains
operations of the same service without parameters. a list of replicated services and may request a tar-
After the client calls operation getYear with pa- get server to delete the service replicated. Figure
rameter Monalisa, he or she could call two opera- 18(a) shows that an origin server replicates a Web
tions, such as getDescriptionOfArtWithoutParam service to a target server and records the service
and getArtistWithoutParam, without parameters. and its target server. As shown in Figure 18(b),
As shown in Figures 17(b) and 17(d), we find after replication, the origin server can request to
that the client could call three operations with the target server to delete the service replicated.
no parameters (e.g., getBornYear, getDeadYear, Figure 18(c) indicates that the service was deleted
and getDescriptionOfArtistWithoutParam). This from the target server according to the request of
the origin server.

Figure 18. Deletion of the service replicated

(a) Before requesting deletion (b) After requesting deletion (c) After deletion (target)

Figure 19. Replicating a component to a client


Origin server

Connection Component replication

Client

Before execution A fter execution

Component

720
Migrating Web Services in Mobile and Wireless Environments

Figure 20. An example of component replication

(a) Log information of a server (b) Log information of a client

Replicating Components to a Client the size of a service increases, it takes a lot of


resources to migrate the service. They do not
There may be a case where if a service requires handle the migration of service instances and
a large amount of input and output messages, it the replication of services. Hammerschmidt
is more efficient to download and execute the and Linnemann (2005) support the migration of
service at a client side. In the scenario of Figure instances to reduce network traffic and resource
19, an origin server provides a service of convert- consumption on an origin server. Among the
ing images and a client wants to resize an image. methods concerning code mobility, Bellavista
The client does not transmit a large image file as et al. (2004) and Montanari et al. (2004) support
a parameter; instead, the client resizes the image the migration of binary codes, but do not offer
by executing the service component replicated the instance migration.
from the origin server. In this article, we present a method for mi-
Figure 20(a) shows the log information of the grating and replicating Web services on mobile
origin server, which provides the image conversion and wireless environments. Based on the context
service. Figure 20(b) is the log information of the information of mobile devices, the proposed
client that uses the service replicated. Likewise, method selects between migration and replica-
the origin server records the service replicated tion. Additionally, service instances are preserved
and the client. It may also request the client to during migration and replication. The proposed
delete the service replicated. framework does not consider the reconfiguration
of services after migration. The reconfiguration
Comparison with Previous Works involves increases in code size since it requires
codes—which are relevant to each context—to
Table 1 shows the comparison result of the pro- be implemented. Therefore, it may be not suitable
posed method with previous works. The method for wireless and mobile environments.
of Pratitha and Zaslavsky (2004) migrates ser-
vices to a target server, which is selected based
on context information. Services are composed
of service modules, which have been defined
in advance depending on context. However, as

721
Migrating Web Services in Mobile and Wireless Environments

Table 1. Comparison with previous works


Features Web Instance Supporting mobile and
Reconfiguration Context –awareness
Methods services preservation wireless environments
Pratitha and Zaslavsky (2004) O* X O X O
Hammerschmidt and Lin-
O O X X X
nemann (2005)
Ishikawa, Yoshioka Tahara,
O X X X X
and Honiden (2004)
Mifsud and Stanski (2002) O X X X X
Kern and Braun (2005) X O X X X
Bellavista, Corradi, and Mag-
X X X O O
istretti (2004)
Montanari, Lupu, and Ste-
X X X X O
fanelli (2004)
The proposed method O O X O O
*
O: supported, X: not supported

CONCLUSION AND FUTURE STUDY it may be necessary to merge and split the codes
split repeatedly since user preferences may change
This article presents an efficient method for rapidly and frequently. Therefore, to ref lect the
migrating and replicating Web services through changing desires and preferences of users
code splitting. Specifically, a service is split dynamically at runtime, we have a plan to
into subcodes based on users’ preferences to its enhance the proposed migration method
constituent operations. The subcodes of higher to make it possible to accommodate user
preference are migrated earlier to minimize the feedback at runtime.
discontinuance of the operations of high priority
and raise the efficiency of Web services migration
and replication in wireless and mobile environ- ACKNOWLEDGMENT
ments. From experimental results, we found that
the proposed splitting method depended on the This work was supported by the Korea Research
size of a Web service and the number of func- Foundation Grant funded by the Korean Govern-
tions called by its operations. In particular, as ment (MOEHRD) (KRF-2005-041-D00690).
the number of operations increased, the splitting
time grew at an exponential rate. Moreover, the
proposed migration method was experimented REFERENCES
under three application scenarios. Experimental
results show that the proposed migration method Bellavista, P., Corradi, A., & Magistretti, E. (2004,
helps services to be provided continuously. September 20-22). Lightweight code mobility for
The proposed method selects an optimal com- proxy-based service rebinding in MANET. In
bination of code split, on which the source code Proceedings of the 1st International Symposium
is split and compiled into components. However, on Wireless Communication Systems, Port Louis,

722
Migrating Web Services in Mobile and Wireless Environments

Mauritius (pp. 208-214). on Ontologies, Databases and Application of


Semantics (ODBASE’04), Agia Napa, Cyprus
Grove, D., & Chambers, C. (2001). A framework
(pp. 1358-1375).
for call graph construction algorithms. ACM
Transactions on Programming Languages and Kim, Y.-S., & Lee, K.-H. (2007). An efficient
Systems, 23(6), 685-746. policy establishment scheme for Web services
migration. In Proceedings of the International
Grove, D., DeFouw, G., Dean, J., & Chambers,
Conference on Convergence Information Tech-
C. (1997, October 5-9). Call graph construction in
nology (ICCIT’07) (pp. 595-600).
object-oriented languages. In Proceedings of the
ACM SIGPLAN Conference on Object-Oriented Mifsud, T., & Stanski, P. (2002, October 8-11).
Programming Systems Languages and Applica- Measuring performance of dynamic Web ser-
tions, Atlanta (pp. 108-124). vice migration using LAMS. In Proceedings of
the 10th International Conference on Software,
Hammerschmidt, B. C., & Linnemann, V. (2005,
Telecommunications and Computer Networks
February 22-25). Migrating stateful Web services
(SoftCOM), Venice, Italy (pp. 214-218).
using Apache AXIS and P2P. In Proceedings of the
IADIS International Conference on Applied Com- Montanari, R., Lupu, E., & Stefanelli, C. (2004).
puting, Algarve, Portugal (pp. 433-441). Policy-based dynamic reconfiguration of mobile-
code applications. Computer, 37(7), 73-80.
Huhns, M. N., & Singh, M. P. (2005). Service-
oriented computing: Key concepts and principles. Pratistha, I. M., & Zaslavsky, A. B. (2004, March
IEEE Internet Computing, 9(1), 75-81. 14-17). Fluid: Supporting a transportable and
adaptive Web service. In Proceedings of the
Ishikawa, F., Yoshioka, N., Tahara, Y., & Honiden,
2004 ACM Symposium on Applied Computing,
S. (2004, July 19-20). Toward synthesis of Web
Nicosia, Cyprus (pp. 1600-1606).
services and mobile agents. In Proceedings of
AAMAS Workshop on Web Services and Agent- Schall, D., Aiello, M., & Dustdar, S. (2006). Web
based Engineering (WSABE), New York (pp. services on embedded devices. Web Information
227-245). System, 1(1), 1-6.
Kern, S., & Braun, P. (2005, September 20-22). Sirirama, S. N., Jarke, M., & Prinz, W. (2006,
Towards adaptive migration strategies for mo- February 19-22). Mobile Web services provi-
bile agents. In Proceedings of the Second GSFC/ sioning. In Proceedings of the Advanced Inter-
IEEE Workshop on Radical Agent Concepts national Conference on Telecommunicatons and
(WRAC) (pp. 334-345). NASA Goddard Space International Conference on Internet and Web
Flight Center Visitor’s Center Greenbelt, MD. Application and Services, Guadeloupe, France
(pp. 120-126).
Kern, S., Braun, P., Fensch, C., & Rossak, W.
(2004, October 25-29). Class splitting as a method Stal, M. (2006). Using architectural patterns and
to reduce migration overhead of mobile agents. blueprints for service-oriented architecture. IEEE
In Proceedings of the International Conference Software, 23(2), 54-61.

This work was previously published in International Journal of Web Services Research, Vol. 6, Issue 2, edited byL. Zhang, pp.
1-19, copyright 2009 by IGI Publishing (an imprint of IGI Global).

723
724

Chapter 3.4
Applying Web-Based
Collaborative Decision-
Making in Reverse Logistics:
The Case of Mobile Phones
Giannis T. Tsoulfas
University of Piraeus, Greece

Costas P. Pappis
University of Piraeus, Greece

Nikos I. Karacapilidis
University of Patras, Greece

Abstract INTRODUCTION

The increasing environmental concerns and the tech- The increasing environmental concerns and the
nological advances have boosted the post-use treat- technological advances have boosted the post-use
ment of nearly all kinds of products and a new area treatment of nearly all kinds of products, regard-
for research and application has emerged described less of their size, composition, and initial value.
by the term “reverse logistics.” In this chapter, pa- Relevant legislative frameworks have been enforced
rameters that may affect reverse logistics operation in developed countries aiming at apportioning the
are discussed from a decision-making perspective, so responsibilities related to the recovery of end-of-
that alternative design options may be proposed and life products. In addition, specific targets regarding
evaluated. In particular, these parameters are used product design and recovery rates are set, networks’
for the qualitative evaluation of the reverse supply requirements are suggested and, last but not least,
chain of mobile phones in Greece. For this purpose, voluntary schemes are applauded. As a result, further
we present an illustrative application of a Web-based extensions in research and applications of supply
decision support tool that may assist collaborative chain management have emerged described by
decision-making in conflicting environments, where the term “reverse logistics.” De Brito and Dekker
diverse views, perspectives, and priorities shared (2004) defined reverse logistics as “the process of
among stakeholders have to be considered. planning, implementing and controlling backward
flows of raw materials, in process inventory pack-
aging and finished goods, from a manufacturing,
DOI: 10.4018/978-1-60566-114-8.ch018

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

distribution or use point, to a point of recovery From this perspective, it is important to iden-
or point of proper disposal.” In this definition tify the parameters that may affect reverse logistics
both economic and environmental dimensions operation so that alternative design options are
of reverse logistics are implied, indicating the proposed and evaluated. In Tsoulfas, Dasaklis, and
potential benefits that companies would have by Pappis (2007), a first attempt to define and catego-
adopting such practices. rize them is presented. Given these parameters,
Reverse logistics is a multidisciplinary area in this chapter we discuss a qualitative evaluation
of research. For example, operations research, of the reverse supply chain of mobile phones in
environmental analysis, marketing, and informat- Greece, as presented by Pappis, Tsoulfas, and
ics have all a significant role to play in order to Dasaklis (2006). For this purpose, we make use
assist decision-making regarding the design and of a Web-based decision support tool that may
operation of reverse supply chains. Moreover, assist collaborative decision-making (CDM) in
reverse logistics is often regarded in conjunction conflicting environments, where diverse views,
with forward logistics, since they are interrelated. perspectives, and priorities shared among stake-
However, the distinguishing characteristics of holders have to be considered.
reverse supply chains introduce new dimensions The remainder of the chapter is structured as
in decision-making aspects. In particular, the follows: First, the parameters affecting reverse
main differences between forward and reverse logistics operation are discussed. Then, the reverse
supply chains, as stated by Fleischmann, Krikke, supply chain of mobile phones in Greece is briefly
Dekker, and Flapper (1999) and Krikke, Pappis, presented. Next, the CDM tool is presented, fol-
Tsoulfas, and Bloemhof-Ruwaard (2002), are lowed by its illustrative application regarding the
the following: reverse supply chain of mobile phones in Greece.
Finally, some concluding remarks are outlined.
• In contrast to forward supply chains, in
reverse supply chains there are a lot of
sources of “raw materials” (used prod- PARAMETERS AFFECTING
ucts), which may enter the reverse flow REVERSE LOGISTICS OPERATION
at low or no cost at all, and significantly
fewer “customers” (recyclers, remanufac- Three major categories of parameters that may
turers, etc.). affect reverse logistics operation are identified:
• The economic value of inputs in reverse product-dependant, organizational, and social.
supply chains is lower than the one in the These parameters, which cannot be addressed
case of forward supply chains. independently since they may interact with each
• In the case of reverse supply chains, offer other, may form a nonexhaustive basis for analysis
does not follow demand. in the following decision-making situations:
• The economic efficiency of reverse sup-
ply chains is precarious, since it is not sure a. When assessing the current situation re-
that there will be markets to exploit their garding the operation of reverse supply
outputs. chains;
• Reverse supply chains are characterized b. When exploring alternative options for
by higher uncertainty regarding issues like the reverse supply chain activities, as
quality, volumes, and composition of re- well as their interaction with the external
verse flows. environment.

725
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

Figure 1. The various flows between the two major sessions of reverse supply activities

Generally speaking, reverse supply chains may criterion for the development of reverse lo-
be considered as the conjunction of two major gistics activities, since several operations,
sessions of activities: acquisition and exploitation. such as collection, storage, and transporta-
The first one refers to the activities that aim at tion, may be affected. Generally speaking,
the physical transportation of used products and large products may require special machin-
the second one includes the activities targeting ery and equipment for handling, transpor-
final value extraction or environmentally sound tation, and so forth, whereas small products
management. Although exploitation follows ac- may call for big quantities to be collected
quisition when the materials’ flows are regarded, before being transported.
it may be considered as a necessary condition for • The composition and the technical charac-
the acquisition in the causal chain. To be more teristics of used products may be another
specific, the ability to exploit used products may decisive issue, since they determine the
trigger their acquisition. Otherwise, it would be ways used products should be treated in
purposeless to acquire used products without order to preserve their value and to prevent
having in mind how to treat them. In addition, them from harming the environment.
information exchange between these sessions • The way used products are replaced by new
is bidirectional and repeated. The relationships ones is another important parameter for the
between the two major sessions of reverse supply implementation of reverse logistics. It is
activities are illustrated in Figure 1, where the obvious that replacements, which occur in
parameters affecting these operations are regarded the same place or using the same distribu-
as influential factors. tion means for the return of used products,
have positive effects on the reverse supply
Product-Dependent Parameters chain operation, both from an economic
and an environmental perspective.
Product-dependant parameters refer to the par- • The remaining value of used products is
ticular characteristics of products that determine considered to be a very significant issue for
their post-use treatment from a technical and an consumers and manufacturers, since they
economic point of view. In particular: can both benefit from the post-use treat-
ment of used products. Consumers may
• The weight or the volume of used products achieve reduced prices for new products
and the infrastructure needed is a decisive replacing used ones, while manufacturers

726
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

can extract value from used products by re- with whom they can easier communicate.
furbishing, reusing, or recycling them. • Marketing is a very important criterion for
• Direct reuse or reuse after minor treatment the implementation of reverse logistics.
is a situation commonly perceived in the Companies may participate in campaigns
case of packaging materials and may of- for promoting collection of used prod-
fer significant benefits to companies and ucts and they can indicate recovery op-
the environment, as the useful life cycle of tions in the products themselves or in their
used products is extended and the produc- packages.
tion of new ones is avoided. • Economic motivation is a means used in
• The capability to change the usage of used many industrial sectors in order to in-
products or to provide them to different volve consumers in the recovery activi-
markets (e.g., second hand) may be another ties. Usually, consumers prepay a certain
important criterion. Generally, in such situ- amount of money as a deposit and they get
ations no special treatments are necessary it back when they return the product or the
and the life cycle of products is extended. package to collection facilities.

Organizational Parameters Social Parameters

Organizational parameters refer to issues regard- Social parameters involve attitudes and values
ing the stakeholders involved in the recovery of prevailing in societies that may determine prac-
used products. In particular: tices regarding recovery of used products. In
particular:
• The recovery networks structure is a deci-
sive element for reverse logistics activities. • Social habits may significantly affect the
Generally speaking, companies have two results of reverse logistics activities, since
options: either they will handle the recov- individual attitudes are often affected by
ery processes themselves (even by using mainstreams. Recovery of used products
outsourcing practices), or they will partici- seems to find a more fertile ground in big
pate in wider networks, usually by finan- cities rather than in small communities.
cially supporting them. The first option is • Legislation is a decisive parameter affect-
commonly adopted when companies can ing the recovery of used products. In partic-
achieve significant return rates, whereas ular, the principle of shared responsibility
the second scheme is preferred, especially and the “cradle to grave” perspective have
when used products are widely dispersed. been elevated in legislative frameworks
• Asset control policies that are adopted by around the world. Furthermore, explicit
some companies can contribute to the ef- targets are posed and certain benefits are
fective operation of reverse logistics. Such offered in some cases (e.g., tax relieves,
cases are often met, for example, in the au- improved financial eligibility, etc.).
tomotive sector and in the electronics in- • Social awareness is a critical issue regard-
dustry. By using such practices, companies ing reverse logistics practices, especially
actually sell services rather than the prod- with respect to consumers’ attitudes. In
uct itself. As a result, they can have im- developed countries, the environmental
proved control of their products and, at the standards stemming from social demand
same time, fulfill their customers’ needs, are higher. Education is of particular

727
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

importance regarding this parameter, not S.A., since they are the only authorized take-back
solely in schools, but also in corporate organisation in Greece right now. The program
environments. relies on in-store collection and special bins have
been installed in retail stores. In addition, several
THE REVERSE SUPPLY CHAIN OF bins have been put on central city spots, as the
MOBILE PHONES IN GREECE result of cooperation with local municipalities.
Recycling Appliances S.A. aimed at covering 67%
Facts of Greek mobile phones population by the end of
2006, while their corresponding target for 2008
In 2002, the total number of mobile phones in is 90% (http://www.electrocycle.gr). The used
use worldwide exceeded the number of land-lines mobile phones collected have been destined only
(Donner, 2005). According to the International abroad for further treatment, since no appropriate
Telecommunication Union the mobile subscribers facilities exist in Greece. In the recycling process,
in 2006 were more than 2.5 billions (International the plastic parts of mobile phones are incinerated
Telecommunication Union, 2007). According to and utilised as a fuel to melt the metal mixture.
the same source, the subscribers in Greece were Then metals are separated using electrolytic refin-
around 11 million. ing and mechanical (e.g., magnetic segregation)
Typically, mobile phones are used for only procedures.
1½ years before being replaced (Fishbein, 2002). The possible routes of used mobile phones and
These obsolete mobile phones are mainly replaced the affected activities in the forward supply chain
due to fashion trends and the rapid technological (grey color) are illustrated in Figure 2.
improvements, as new features are added in mobile
phones. Other reasons for replacement are the in- Major Concerns
compatibility with a new provider, or the fact that
they no longer function. Less than 1% of mobile Mobile phones contain a great number of metals
phones retired and discarded annually are recycled such as copper, aluminium, iron, nickel, silicon,
and the majority is accumulated in consumers’desk lead, antimony, beryllium, arsenic, silver, tanta-
drawers, store rooms, or other storage, awaiting lum, and zinc. Some of these metals are toxic and
disposal (Most, 2003). Of this small percentage hazardous for mankind and the environment. This
recovered, some are refurbished and put into use or variety of valuable metals raises very significant
used for replacement parts. If these options are not issues regarding the gradual exhaustion of natu-
possible, some metals are recycled. The refurbish- ral resources. In addition, the side effects of this
ment process can significantly aid to the prolonga- exhaustion are also important. For example, the
tion of a mobile phone’s life cycle and therefore mining of tantalum has been identified as a seri-
prevent it from early entry into the waste stream. ous threat to gorillas clinging to survival in the
The recycling process keeps discarded phones out Democratic Republic of Congo (Macey, 2005).
of disposal facilities and reduces the need for raw Apart from metals, mobile phones contain also
materials used to make new products. brominated flame retardants, which are used in
In the case of Greece, Appliances Recycling the plastic parts and cables in order to reduce the
S.A. is the authorized collective take-back and risk of fire. When burned in incinerators, these
recycling organisation for all electrical and elec- substances have the potential to pollute the air
tronic waste in Greece (Pappis et al., 2006). Actu- and to pose threats for the workers in recycling
ally, all service providers and importers are obliged facilities, since dioxins and furans can be formed.
by law to cooperate with Appliances Recycling When buried in landfills, they may leach into soil

728
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

Figure 2. The possible routes of used mobile phones and the affected activities in the forward supply
chain (grey color)

and drinking water. get more complicated, due to the involvement of


The environmental impact of the substances diverse parties. For example, in the case of mobile
mentioned above is of great concern because phones, manufacturers, distributors, service pro-
some of them, like flame retardants and lead, are viders, recovery operators, and recyclers would
considered to be persistent, bioaccumulative, be the participants of such a decision-making
and suspected carcinogens. Relative legislation situation. Moreover, even different departments
enforcement in the European Union aims at the of these stakeholders might have diverse views
restriction of the use of certain hazardous sub- of the situation.
stances in electrical and electronic equipment, such
as mobile phones (RoHS Directive) (European The Decision-Making Situation
Union, 2003a). In addition, the WEEE Directive
draws the frame regarding the post-use treatment From an OEMs’ perspective, reverse logistics
of electrical and electronic equipment (European implementation in the case of mobile phones
Union, 2003b). is determined by their interaction with several
Apart from the environmental concerns related stakeholders, as shown in Figure 3.
to the treatment of used mobile phones, there are Thus, reverse logistics managers are respon-
some important economic issues as well. Indeed, sible for taking into account and coordinating all
many substances contained in mobile phones stakeholders’ requirements. Correspondingly,
are valuable as it is relatively more expensive similar actions are necessary among the different
to acquire them as primary raw materials (e.g., departments within a company. Consequently,
lead, zinc). a conflicting decision-making environment is
It is obvious that decision-making procedures formed, where the factors “place” and “time”
regarding the operation of reverse supply chains may pose restrictions. Such decision-making

729
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

Figure 3. OEMs and their interactions with stakeholders in reverse logistics activities

situations may be dealt with Web-based CDM computer-based knowledge management systems
tools and a corresponding approach is presented (KMS) focus on providing a corporate memory,
in the sequel. that is, an explicit, disembodied, and persistent
representation of the knowledge and information
in an organization, as well as mechanisms that
A WEB-BASED TOOL improve the sharing and dissemination of knowl-
FOR COLLABORATIVE edge by facilitating interaction and collaboration
DECISION-MAKING among the parties involved (Bolloju, Khalifa, &
Turban, 2002).
Approaching Conflicting CDM may provide a means for a well-struc-
Decision-Making Situations tured decision-making process. Usually, CDM
is performed through debates and negotiations
Choices in decision-making cannot generally be among the parties involved. Conflicts of interest
addressed by individuals working alone or even are inevitable and support for achieving consensus
by several people working separately and then and compromise is required. Decision-makers
merging their pieces of work. Instead, they have may adopt and suggest their own strategy that
to be addressed through collaborative work among fulfils some goals at a specific level and may
stakeholders with diverse views, perspectives, have arguments supporting or against alternative
and priorities. solutions. In addition, they have to confront the
Information and Communication Technology existence of insufficient information. Generally
(ICT) infrastructure to support people working speaking, efficient and effective use of informa-
in teams has been the subject of interest for quite tion technology in the collection and dissemina-
a long time (Fjermestad & Hiltz, 2000). Such tion of information and knowledge produced
systems aim at facilitating group decision-making by diverse sources, the evaluation of alternative
processes by providing forums for expression of schemes, the construction of shared meanings,
opinions, as well as qualitative and quantitative and the associated feedback learning process are
tools for aggregating proposals and evaluating critical factors for the decision-making process
their impact on the issue in hand. They may (Clases & Wehner, 2002).
exploit intranet or Internet technologies to con-
nect decision-makers in a way that encourages
dialogue and, at the same time, stimulates the
exchange of knowledge. Accordingly, recent

730
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

The Web-Based Tool which participants get connected with the system
via Internet (by using a standard Web browser;
Given the above issues, a Web-based tool has there is no need of installation of any specific
been implemented that supports the collabora- software in order to use the tool). Exploitation
tion conducted in decision-making situations, of the Web platform renders, among others, low
by facilitating the creation, leveraging, and uti- operational cost and easy access to the system.
lization of the relevant knowledge. This tool is The knowledge base of the system maintains all
based on an argumentative reasoning approach, the above items (messages and documents), which
where discourses about complex problems are may be considered, appropriately processed and
considered as social processes and they may result transformed, or even re-used in future discussions.
in the formation of groups whose knowledge is Storage of documents and messages being as-
clustered around specific views of the problem serted in an ongoing discussion takes place in an
(Karacapilidis, Adamides, & Pappis, 2004). In automatic way, upon their insertion in the discus-
addition to providing a platform for group reflec- sion. On the other hand, retrieval of knowledge is
tion and capturing of organizational memory, this performed through appropriate interfaces, which
approach augments teamwork in terms of knowl- aid participants in exploring the contents of the
edge elicitation, sharing, and construction, thus knowledge base and exploit previously stored
enhancing the quality of the overall process. This or generated knowledge for their current needs.
is due to its structured language for conversation In such a way, our approach builds a “collective
and its mechanism for evaluation of alternatives. memory” of a particular community. On the other
Taking into account the input provided by the hand, the collaborative decision-making module
individual experts, the proposed tool constructs is responsible for the reasoning and evaluation
an illustrative discourse-based knowledge graph purposes of the system. Alternative mechanisms
that is composed of the ideas expressed so far, as for these purposes can be invoked each time, upon
well as their supporting documents. Moreover, the participants’ wish and context under consid-
through the integrated decision support mecha- eration. These mechanisms follow well-defined
nisms, experts are continuously informed about and broadly accepted algorithms (based on diverse
the status of each discourse item asserted so far decision making approaches, such as multi-criteria
and reflect further on them according to their decision-making, argumentation-based reasoning,
beliefs and interests on the outcome of the discus- utility theory, risk assessment, etc.), which are
sion. In addition, the overall approach aids group stored in the tool’s model base.
sense-making and mutual understanding through The basic discourse elements in the proposed
the collaborative identification and evaluation of tool are issues, alternatives, positions, and prefer-
diverse opinions. ences. In particular, issues correspond to problems
The proposed tool builds on a server-client to be solved, decisions to be made, or goals to be
network architecture. It is composed of two basic achieved. They are brought up by users and are
components, namely the collaboration visualiza- open to dispute (the root entity of a discourse-
tion module and the collaborative decision making based knowledge graph has to be an issue). For
module. The former provides a shared Web-based each issue, the users may propose alternatives (i.e.,
workspace for storing and retrieving the mes- solutions to the problem under consideration) that
sages and documents deployed by the discussion correspond to potential choices. Nested issues, in
participants, using the widely accepted XML cases where some alternatives need to be grouped
document format (http://www.w3.org/XML). This together, are also allowed. Positions are asserted in
module actually provides the interfaces through order to support the selection of a specific course

731
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

of action (alternative), or avert the users’ interest external environment. The parameters that affect
from it by expressing some objection. A position reverse logistics operations are used as a basis
may also refer to another (previously asserted) for the discourse. The decision-making process
position, thus arguing in favor or against it. Finally, may reveal flaws of current practices as well as
preferences provide individuals with a qualitative improvement potentials and areas to focus on.
way to weigh reasons for and against the selection Figures 4 and 5 correspond to instances of
of a certain course of action. A preference is a tuple collaboration concerning the “Recovery network
of the form (position, relation, position), where structure,” and “Marketing,” respectively. In
the relation can be “more important than” or “of these instances, the stakeholders participate in an
equal importance to” or “less important than.” argumentation-based decision-making process.
The use of preferences results in the assignment More specifically, in the instance shown in Figure
of various levels of importance to the alternatives 4, the issue under consideration is “Priorities in
in hand. Like the other discourse elements, they improving the recovery network’s structure,” while
are subject to further argumentative discussion. three alternatives, namely “Extended coopera-
These four types of elements enable users tion with local municipalities,” “Collection bins
to contribute their knowledge on the particular in super markets,” and “Collection programs
problem (by entering issues, alternatives, and in schools,” have been proposed so far (by C,
positions) and also to express their relevant values, A, and B, respectively). The three stakeholders
interests and expectations (by entering positions have argued about them by expressing positions
and preferences). In such a way, the tool supports speaking in favor or against them.
both the rationality-related dimension and the For instance, “People visit super markets at
social dimension of the underlying collaborative least once a week” is a position (asserted by A)
decision-making process. Moreover, the tool con- that argues in favor of the second alternative, while
tinuously processes the elements entered by the “It is time-consuming” is a position (asserted by
users (by triggering its reasoning mechanisms each B) that argues against the first alternative. As also
time a new element is entered in the graph), thus shown in Figures 4 and 5, argumentation can be
enabling users to become aware of the elements for conducted in multiple levels. Furthermore, us-
which there is (or there is not) sufficient (positive ers may also assert preferences about positions
or negative) evidence, and accordingly conduct already expressed. As shown in Figure 5, user C
the discussion in order to reach consensus. has expressed a preference concerning the rela-
tive importance between the positions “It is very
expensive to initiate a nationwide campaign”
ASSESSING THE OPERATION and “The timing is excellent,” arguing that the
OF THE REVERSE SUPPLY first position is of bigger importance for him.
CHAIN OF MOBILE PHONES Users may also express their arguments in favor
IN THE CASE OF GREECE or against a preference.
When clicking on a discourse item, detailed
An illustrative application of the Web-based tool information about it is provided in a dedicated
presented earlier is conducted regarding the quali- window of the basic interface of the tool. More
tative assessment of the operation of the reverse specifically, this part contains information about
supply chain of mobile phones in the case of the user who submitted the selected discussion
Greece. In this application, decision-makers A, B, element, its submission date, any comments that
and C explore interventions in the operation of the the user may have inserted, as well as links to
chain as well as their possible interaction with the related Web pages and multimedia documents that

732
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

Figure 4. Instances of collaboration concerning “Priorities in improving the recovery network’s struc-
ture”

Figure 5. Instances of collaboration concerning “Marketing interventions”

733
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

the user may have uploaded to the tool in order to CONCLUSION


justify this element and aid his/her peers in their
contemplation. The introduction of reverse logistics in supply
Further to the argumentation-based structur- chain management has created new decision-
ing of a discourse, the tool integrates a reasoning making dimensions. Consequently, parameters
mechanism that determines the status of each that may affect the operation of reverse supply
discourse item in order to keep users aware of the chains should be evaluated. In this chapter, a
discourse outcome. More specifically, alternatives, qualitative approach has been discussed with
positions, and preferences of a graph have an ac- respect to such parameters, aiming at facilitat-
tivation label (it can be “active” or “inactive”) ing and augmenting decision-making in reverse
indicating their current status (inactive entries are supply chains. In such cases, several stakeholders
indicated with a red “x”). This label is calculated get involved, including governments, producers,
according to the argumentation underneath and distributors, and customers. As a result, decision-
the type of evidence specified for them. In the in- making procedures get more complicated due to
stance of Figure 4, the position “People visit super increased levels of conflicts of interests but also
markets at least once a week” is inactive because, due to practical reasons. For example, it is not
according to the argumentation rule holding for always easy to get all stakeholders together in
this specific discussion, it has been defeated by a round table. As it has been illustrated in this
the position “The retailers’stores network is wider chapter, ICT may support decision-making pro-
than this of super markets.” Activation in the tool cedures in conflicting environments by provid-
is a recursive procedure; a change of the activation ing the means to structure dialogue, disseminate
label of an element is propagated upwards in the information, and last but not least, facilitate the
discourse graph. Depending on the status of posi- associated reasoning process.
tions and preferences, the mechanism goes through
a scoring procedure for the alternatives of the issue.
A detailed presentation of more technical details FUTURE RESEARCH DIRECTIONS
concerning the argumentation-based reasoning
and scoring mechanisms of the tool can be found The parameters affecting reverse logistics opera-
in Karacapilidis and Papadias (2001). tion may guide decision-makers towards identify-
At each discourse instance, the tool informs ing possible modifications in supply chain activi-
users about what is the most prominent (accord- ties as well as in other corporate issues, such as
ing to the underlying argumentation) alternative marketing and supplier selection. Further research
solution (this is shown by a green “tick” sign). should be devoted to explore the interactions
In the instance shown in Figure 4, “Extended among these parameters and the ways they affect
cooperation with local municipalities” and “Col- the reverse supply chains’ operation and success.
lection programs in schools” are equally justified In addition, research efforts should also focus
as best solutions, while in the instance shown in on how reverse supply chains may interact with
Figure 5“Promotion in retailers’ stores” is the forward supply chains and on relevant expedient
better justified solution so far. However, this may strategies that aim at making the extended supply
change upon the type of the future argumenta- chains more efficient. Moreover, the qualitative
tion; each time an alternative is affected during evaluation of the reverse supply chains of different
the discussion, the issue it belongs to is updated, products may reveal the determinant parameters
since another alternative solution may be indicated for each case, helping to create a body of knowl-
by the tool. edge based on thorough observations. In particu-

734
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

lar, it is important to identify the circumstances REFERENCES


under which reverse supply chains are impeded
and the options to improve their operation. Such Bolloju, N., Khalifa, M., & Turban, E. (2002). In-
knowledge may be exploited in developing organi- tegrating knowledge management into enterprise
zational memory, a process which, beyond storing environments for the next generation decision sup-
individual and collective knowledge, is related port. Decision Support Systems, 33(2), 163–176.
to organizational learning, decision-making, and doi:10.1016/S0167-9236(01)00142-7
competitive capability issues. Clases, C., & Wehner, T. (2002). Steps across
Apart from economic criteria, additional cri- the border: Cooperation, knowledge produc-
teria (environmental, social, etc.) get increasingly tion and systems design. Computer Sup-
involved in decision-making problems, leading ported Cooperative Work, 11(1/2), 39–54.
to more complex decision-making situations. doi:10.1023/A:1015207530896
Moreover, such problems may not be usually
addressed by formal models or methodologies. de Brito, M. P., & Dekker, R. (2004). A framework
Instead, an argumentative practical reasoning for reverse logistics. In R. Dekker, K. Inderfurth,
approach seems to offer a more convenient solu- L. N. van Wassenhove, & M. Fleischmann (Eds.),
tion. Thus, decision-making tools should focus on Quantitative approaches to reverse logistics (pp.
facilitating the cooperation of the different parties 3-27). Berlin, Germany: Springer-Verlag.
involved towards well-structured decision-making
Donner, J. (2005, June 7-8). Research approaches
processes. The corresponding technologies should
to mobile use in the developing world: A review of
further exploit the advances in ICT in order to
the literature. Paper presented at the International
deliver applications of enhanced performance to
Conference on Mobile Communication and Asian
decision-makers, while efficiently and effectively
Modernities, Hong Kong.
addressing communication and collaboration re-
quirements. In particular, it is important to develop European Union. (2003a). Directive 2002/95/EC
a more human-centric view of the problem, which of the European Parliament and of the Council
appropriately structures and manages the underly- of 27 January 2003 on the restriction of the use
ing human interaction. The CDM tool discussed of certain hazardous substances in electrical and
previously in this chapter may be exploited in order electronic equipment. Official Journal of the
to retrieve useful information and knowledge, as European Union.
well as to reason according to previous cases or
European Union. (2003b). Directive 2002/96/EC
predefined rules. The proposed tool may be en-
of the European Parliament and of the Council
hanced with intelligent agent technologies, which
of 27 January 2003 on waste electrical and elec-
are able to facilitate a variety of decision-makers’
tronic equipment (WEEE). Official Journal of the
tasks and actions by acting on their behalf, as well
European Union.
as to automate system’s processes.
Fishbein, B. K. (2002). Waste in the wireless
world: The challenge of cell phones. INFORM
ACKNOWLEDGMENT Inc., New York, USA.
Fjermestad, J., & Hiltz, S. R. (2000). Group sup-
The project is co-funded by the European Social
port systems: A descriptive evaluation of case and
Fund and national resources - EPEAEK II.
field studies. Journal of Management Information
Systems, 17(3), 115–159.

735
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

Fleischmann, M., Krikke, H. R., Dekker, R., & Pappis, C. P., Tsoulfas, G. T., & Dasaklis, T. (2006,
Flapper, S. D. P. (1999). Logistics network (re)- July 2-5). Assessing alternative reverse logistics
design for product recovery and reuse (Manage- policies for the recovery of used mobile phones.
ment Report 17). Rotterdam, The Netherlands: Paper presented at the EURO XXI Conference,
Erasmus University of Rotterdam. Reykjavik, Iceland.
International Telecommunication Union. (2007). Tsoulfas, G. T., Dasaklis, T., & Pappis, C. P.
Basic indicators: Population, GDP, total telephone (2007, May 9-11). Identifying success factors
subscribers and total telephone subscribers per for reverse logistics activities. In Proceedings of
100 people. the 2nd International Conference ECO-X 2007,
Vienna, Austria (pp. 39-46).
Karacapilidis, N., Adamides, E., & Pappis, C.
P. (2004, September 20-25). An IS framework
to support the collaborative design of supply
chains. In M. Negoita, R. J. Howlett, & L. C. Jain ADDITIONAL READING
(Eds.), Knowledge-based intelligent information
and engineering systems: Proceedings of the 8th Adamides, E., & Karacapilidis, N. (2006). A
International Conference, KES 2004, Wellington, knowledge centred framework for collabora-
New Zealand, Part II (pp. 62-70). Berlin/Heidel- tive business process modelling. Business Pro-
berg, Germany: Springer-Verlag. cess Management Journal, 12(5), 557–575.
doi:10.1108/14637150610690993
Karacapilidis, N., & Papadias, D. (2001). Com-
puter supported argumentation and collaborative Ahn, H. J., Lee, H. J., Chob, K., & Park, S. J.
decision making: The HERMES system. Infor- (2005). Utilizing knowledge context in virtual
mation Systems, 26(4), 259–277. doi:10.1016/ collaborative work. Decision Support Systems,
S0306-4379(01)00020-5 39(4), 563–582. doi:10.1016/j.dss.2004.03.005

Krikke, H. R., Pappis, C. P., Tsoulfas, G. T., & Álvarez-Gil, M. J., Berrone, P., Husillos, F. J., &
Bloemhof-Ruwaard, J. (2002). Extended design Lado, N. (2007). Reverse logistics, stakeholders’
principles for closed loop supply chains: Optimis- influence, organizational slack, and managers’
ing economic, logistic and environmental perfor- posture. Journal of Business Research, 60(5),
mance. In A. Klose, M. G. Speranza, & L. N. Van 463–473. doi:10.1016/j.jbusres.2006.12.004
Wassenhove (Eds.), Quantitative approaches to Angell, L. C., & Klassen, R. D. (1999). Integrat-
distribution logistics and supply chain manage- ing environmental issues into the mainstream:
ment (pp. 61-74). Berlin/Heidelberg, Germany: An agenda for research in operations manage-
Springer-Verlag. ment. Journal of Operations Management, 17(5),
Macey, R. (2005, March 10). Mobile phones kill 575–598. doi:10.1016/S0272-6963(99)00006-6
great apes. Sydney Morning Herald. Boons, F. (2002). Greening products: A frame-
Most, E. (2003). Calling all cell phones: Collec- work for product chain management. Journal of
tion, reuse, and recycling programs in the U.S. Cleaner Production, 10(5), 495–505. doi:10.1016/
INFORM Inc., New York, USA. S0959-6526(02)00017-3

736
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

Bowen, F. E., Cousins, P. D., Lamming, R. C., & Georgiadis, P., & Vlachos, D. (2004). The ef-
Faruk, A. C. (2001). The role of supply manage- fect of environmental parameters on product
ment capabilities in green supply. Production and recovery. European Journal of Operational
Operations Management, 10(2), 174–189. Research, 157(2), 449–464. doi:10.1016/S0377-
2217(03)00203-0
Chouinard, M., D’Amours, S., & Aït-Kadi, D.
(2005). Integration of reverse logistics activities Gungor, A., & Gupta, S. M. (1999). Issues in
within a supply chain information system. Com- environmentally conscious manufacturing and
puters in Industry, 56(1), 105–124. doi:10.1016/j. product recovery: A survey. Computers & Indus-
compind.2004.07.005 trial Engineering, 36(4), 811–853. doi:10.1016/
S0360-8352(99)00167-9
Courtney, J. (2001). Decision making and knowl-
edge management in inquiring organizations: Handfield, R., Sroufe, R., & Walton, S. (2005).
Toward a new decision-making paradigm for Integrating environmental management and sup-
DSS. Decision Support Systems, 31(1), 17–38. ply chain strategies. Business Strategy and the
doi:10.1016/S0167-9236(00)00117-2 Environment, 14(1), 1–19. doi:10.1002/bse.422
Dowlatshahi, S. (2000). Developing a theory Karacapilidis, N. (2006). An overview of future
of reverse logistics. Interfaces, 30(3), 143–155. challenges of decision support technologies. In J.
doi:10.1287/inte.30.3.143.11670 Gupta, G. Forgionne, & M. Mora (Eds.), Intelligent
decision-making support systems: Foundations,
Eichner, T., & Pethig, R. (2001). Product design
applications and challenges (pp. 385-399). Lon-
and efficient management of recycling and waste
don: Springer-Verlag.
treatment. Journal of Environmental Economics
and Management, 41(1), 109–134. doi:10.1006/ Khanna, M., & Anton, W. Q. (2002). Corporate
jeem.2000.1126 environmental management: Regulatory and
market-based pressures. Land Economics, 78(4),
Evangelou, C., Karacapilidis, N., & Tzagarakis,
539–558. doi:10.2307/3146852
M. (2006). On the development of knowledge
management services for collaborative decision Liebowitz, J., & Megbolugbe, I. (2003). A set of
making. Journal of Computers, 1(6), 19–28. frameworks to aid the project manager in concep-
tualizing and implementing knowledge manage-
Fleischmann, M., Krikke, H. R., Dekker, R., &
ment initiatives. International Journal of Project
Flapper, S. D. P. (2000). A characterization of logis-
Management, 21(3), 189–198. doi:10.1016/
tics networks for product recovery. Omega, 28(6),
S0263-7863(02)00093-5
653–666. doi:10.1016/S0305-0483(00)00022-0
Malhotra, Y. (2005). Integrating knowledge
Forgionne, G., Gupta, J., & Mora, M. (2002).
management technologies in organizational
Decision making support systems: Achievements,
business processes: Getting real time enter-
challenges and opportunities. In M. Mora, G.
prises to deliver real business performance.
Forgionne, & J. Gupta (Eds.), Decision making
Journal of Knowledge Management, 9(1), 7–28.
support systems: Achievements and challenges
doi:10.1108/13673270510582938
for the new decade (pp. 392-403). Hershey, PA:
Idea Group.

737
Applying Web-Based Collaborative Decision-Making in Reverse Logistics

Mora, M., Forgionne, G., Gupta, J., Cervantes, Rogers, D. S., & Tibben-Lembke, R. S. (1999).
F., & Gelman, O. (2003, September 3-5). A Going backwards: Reverse logistics trends and
framework to assess intelligent decision-making practices. Pittsburgh: Reverse Logistics Execu-
support systems. In V. Palade, R. J. Howlett, & tive Council.
L. C. Jain (Eds.), Knowledge-based intelligent
Shapiro, J. (2001). Modeling and IT per-
information and engineering systems: Proceed-
spectives on supply chain integration. Infor-
ings of the 7th International Conference, KES
mation Systems Frontiers, 3(4), 455–464.
2003, Oxford, UK, Part II (pp. 59-65). Berlin/
doi:10.1023/A:1012876804965
Heidelberg: Springer-Verlag.
Shim, J. P., Warkentin, M., Courtney, J. F., Power,
Muckstadt, J., Murray, D., Rappold, J., & Col-
D. J., Sharda, R., & Carlsson, C. (2002). Past,
lins, D. (2001). Guidelines for collaborative
present and future of decision support technol-
supply chain system design and operation.
ogy. Decision Support Systems, 33(2), 111–126.
Information Systems Frontiers, 3(4), 427–453.
doi:10.1016/S0167-9236(01)00139-7
doi:10.1023/A:1012824820895
Tibben-Lembke, R. S., & Rogers, D. S.
Nonaka, I. (1994). A dynamic theory of organiza-
(2002). Differences between forward and
tional knowledge creation. Organization Science,
reverse logistics in a retail environment.
5(1), 14–37. doi:10.1287/orsc.5.1.14
Supply Chain Management, 7(5), 271–282.
Provis, C. (2004). Negotiation, persuasion doi:10.1108/13598540210447719
and argument. Argumentation, 18(1), 95–112.
Tsoulfas, G. T., & Pappis, C. P. (2006). En-
doi:10.1023/B:ARGU.0000014868.08915.2a
vironmental principles applicable to supply
Ravi, V., & Shankar, R. (2005). Analysis of interac- chains design and operation. Journal of Cleaner
tions among the barriers of reverse logistics. Tech- Production, 14(18), 1593–1602. doi:10.1016/j.
nological Forecasting and Social Change, 72(8), jclepro.2005.05.021
1011–1029. doi:10.1016/j.techfore.2004.07.002
Richey, R. G., Chen, H., Genchev, S. E., &
Daugherty, P. J. (2005). Developing effective
reverse logistics programs. Industrial Marketing
Management, 34(8), 830–840. doi:10.1016/j.
indmarman.2005.01.003

This work was previously published in Web-Based Green Products Life Cycle Management Systems: Reverse Supply Chain
Utilization, edited by H. Wang, pp. 401-415, copyright 2005 by Information Science Reference (an imprint of IGI Global).

738
739

Chapter 3.5
WSBen:
A Web Services Discovery and
Composition Benchmark Toolkit1

Seog-Chan Oh
General Motors R&D Center, USA

Dongwon Lee
The Pennsylvania State University, USA

ABSTRACT the performance of web services discovery and


composition algorithms. The WSBen toolkit is
In this article, a novel benchmark toolkit, WSBen, available at: http://pike.psu.edu/sw/wsben/.
for testing web services discovery and composition
algorithms is presented. The WSBen includes: (1) a
collection of synthetically generated web services Introduction
files in WSDL format with diverse data and model
characteristics; (2) queries for testing discovery A Web Service is a set of related functionalities
and composition algorithms; (3) auxiliary files to that can be loosely coupled with other services
do statistical analysis on the WSDL test sets; (4) programmatically through the Web. Examples
converted WSDL test sets that conventional AI of web applications using Web services include
planners can read; and (5) a graphical interface weather forecasting, credit check, and travel
to control all these behaviors. Users can fine- agency programs. As a growing number of Web
tune the generated WSDL test files by varying services are available on the Web and in organi-
underlying network models. To illustrate the ap- zations, finding and composing the right set of
plication of the WSBen, in addition, we present Web services become ever more important. As
case studies from three domains: (1) web service a result, in recent years, a plethora of research
composition; (2) AI planning; and (3) the laws of work and products on Web-service discovery
networks in Physics community. It is our hope that and composition problems have appeared2. In
WSBen will provide useful insights in evaluating addition, the Web service research community

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
WSBen

has hosted open competition programs (e.g., WSDL files at operation level, yielding 569 valid
ICEBE053, EEE064) to solicit algorithms and WSDL files. Finally, we conducted type flatten-
software to discover pertinent Web services and ing and data cleaning processes subsequently.
compose them to make value-added functional- The type flattening process is to extract atomic
ity. Despite all this attention, however, there have types from user-defined complex types using type
been very few test environments available for hierarchy of XML schema. This process helps find
evaluating such algorithms and software. The more compatible parameter faster. Details are
lack of such a testing environment with flexible found in (Kil, Oh, & Lee, 2006). The final step
features hinders the development of new compo- is the data cleansing to improve the quality of
sition algorithms and validation of the proposed parameters. For instance, substantial number of
ones. Therefore, the need for a benchmark arises output parameters (16%) was named “return”,
naturally to evaluate and compare algorithms “result”, or “response” which is too ambiguous
and software for the Web-service discovery and for clients. However, often, their more precise un-
composition problems. As desiderata for such a derline meaning can be derived from contexts. For
benchmark, it must have (a large number of) web instance, if the output parameter named “result”
services in the standard-based WSDL files and belongs to the operation named “getAddress'”,
test queries that can represent diverse scenarios then the “result” is in fact “Address”. In addition,
and situations that emphasize different aspects of often, naming follows apparent pattern such as
various Web-service application domains. Often, getFooFromBar or searchFooByBar. Therefore,
however, test environments used in research and to replace names of parameters or operations by
evaluation have skewed test cases that do not more meaningful ones, we removed spam tokens
necessarily capture real scenarios. Consider the like “get” or “by” as much as we could.
following example. We measured how many distinct parameters
each WSDL file contained. Suppose that given
Example 1 (Motivating) Let us use the follow- a parameter p ∈ P, we denote the number of
ing notations: A Web service w ∈ W, specified occurrences of p.name as #(p.name). That is,
in a WSDL file, can be viewed as a collection #("pwd") indicates the number of occurrences
of operations, each of which in turn consists of of the parameter with name of “pwd”. Figure
input and output parameters. When an operation 1 illustrates #(p.name) distributions of PUB06
op has input parameters opi = {p1,...,pn} and and the ICEBE05 test set, where the x-axis is #(p.
output parameters opo = {q1,...,qn}, we denote name) and the y-axis is the number of parameters
the operation by op(opi, opo). Furthermore, each with the same #(p.name) value. The distribution
parameter is viewed as a pair of (name, type). of PUB06 has no humps. We also plotted a power-
We denote the name and type of a parameter p by function, over the #(p.name) distribution, and
p.name and p.type, respectively. For the motivat- found that the exponent is 1.1394. Although 1.1394
ing observation, we first downloaded 1,544 raw does not suffice the requirement to be the power
WSDL files that Fan and Kambhampati (2005) law (Denning, 2004), the distribution is skewed
gathered from real-world Web services registries enough to be seen as the Zipf-like distribution.
such as XMethods or BindingPoint. We refer to Indeed, the parameters such as “license key”,
the data set as PUB06. For the purpose of pre- “start date”, “end date,” or “password” have
processing PUB06, first, we conducted WSDL a large #(p.name) value, while most parameters
validation according to WSDL standard, where appear just once. This observation also implies
874 invalid WSDL files are removed and 670 files the existence of hub parameters, which appear in
are left out. Second, we removed 101 duplicated Web services frequently, and serve important roles

740
WSBen

on the inter-connections between Web services. our best knowledge, there have been no publicly
On the contrary, the distribution of ICEBE05 test available benchmark tools. To address these needs
set has four humps equally at around 1, 100, 200, and shortcomings, therefore, we developed the
and 800 with the highest value at third hump. WSBen - a Web-Service discovery and composi-
This distribution shape differs considerably from tion Benchmark tool. The main contributions of
PUB06, the real public Web services. This implies WSBen is to provide diverse Web service test sets
that the test environments of ICEBE05 do not based on three network models such as “random”,
necessarily capture characteristics of real Web “small-world”, and “scale-free” types. These
services. three network models have been shown to model
many real-world networks sufficiently (Albert &
In conclusion, as demonstrated in the example, Barabasi, 2002). We also present three use cases
our claim is that any Web-services discovery and in different communities to demonstrate the ap-
composition solutions must be evaluated under plication of WSBen. In addition, we propose a
diverse configurations of Web services networks flexible framework, by which we can study real
including two cases of Figure 1. However, to Web service networks, and establish the design

Figure 1. #(p.name) distributions. (left) PUB05. (right) ICEBE05.

Table 1. Summary of notation


Notation Meaning
w, W Web service, set of Web services
p, P Parameter, set of parameters
r, r , r
i o
Request, initial and goal parameter sets of r
Gp(Vp, Ep) Parameter node network
Gop(Vop, Eop) Operation node network
Gsw(Vsw, Esw) Web service node network
f
G (V , E )
op
f
op
f
op
Full-matching operation node network
Gc1(Vc1, Ec1) Parameter cluster network
g ri ( p) Minimum cost of achieving p ∈ P from ri in Gp

xTS WSBen’s 5-tuple input framework (e.g., baTS, erTS and nwsTS are instances)

741
WSBen

foundation of WSBen. As a whole, this article is of projecting Web-service networks is that it can
based on two of our earlier works (Kil et al., 2006; allow for straightforward analysis on referred
Oh, Kil, Lee, & Kumara, 2006). Extended from network’s characteristics. Throughout this ar-
the previous works, this article describes how ticle, we will use our conceptual Web-service
WSBen is designed and works to generate test network concept in order to analyze real public
sets from the software architecture perspective, Web-service networks as well as WSDL test file
and furthermore introduces three use cases to sets generated by WSBen.
highlight the practical benefits of WSBen. Table
1 summarizes important notations used in this Web-Service Discovery and
article. Composition
This article is organized as follows. First, in
the background section, we review concepts and Suppose that a Web service w has one opera-
techniques required for the WSBen development, tion so that a Web service can be considered as
especially focusing on the complex network an operation, and input and output parameter
theory. Second, in the related works section, we sets of w are denoted by wi and wo, respectively.
discuss related studies in the literature as well as When one has a request r that has initial input
surveying existing world-wide challenges with parameters ri and desired output parameters ro,
regard to Web services and Semantic Web. Third, one needs to find a Web service w that can fulfill
in the overview of WSBen section, we present such that (1) ri ⊇ wi and (2) ro ⊆ wo. Finding a Web
WSBen with its design concept, test set generation service that can fulfill r alone is referred to as
mechanism, key functions and characteristics. Web-service discovery (WSD) problem. When it
Fourth, in the use cases of WSBen section, we is impossible for one Web service to fully satisfy
illustrate how WSBen can be exploited to obtain r, on the other hand, one has to compose multiple
research benefits, especially by demonstrating Web services {w1, w2,...,wn}, such that (1) for all
three use cases. We expect three use cases enough wk ∈ {w1, w2,...,wn}, wki can be applicable when wki
to provide vigorous experiments and evaluation of is required at a particular stage in composition,
our WSBen. Finally, conclusions are provided. and (2) (ri ∪ w1o ∪ w2o∪...∪ wno) ⊇ ro. This problem
is often called as Web-service composition (WSC)
problem. In addition, one can also consider
Background different matching schemes from the operation
perspective – “partial” and “full” matching. In
In this section, we review prerequisite techniques general, given w1 and w2, if w1 can be invoked at
and concepts required to build WSBen. First, the current information state and w1o ⊇ w2i , then w1
we revisit the definition and complexity of Web- can “fully” match w2. On the other hand, if w1
service discovery and composition problems. cannot fully match w2, but w1 can match a subset
Second, we introduce three complex network of w2, then w1 can “partially” match w2. When
topologies based on which WSBen is designed only full matching is considered in the WSC
to populate WSDL test files. Finally, we explain problem, it can be seen as a single-source shortest
our conceptual methodology to project a bipartite path problem whose computational complexity is
Web-service network consisting of three distinct known as polynomial (Bertsekas, 2000). On the
nodes (parameter, operation, and Web service) other hand, when both full and partial matching
and heterogeneous arc types into three distinct must be considered concurrently, the problem
Web-service networks, each of which consists of becomes a decision problem to determine the
single node and uniform arc. The main benefit existence of a solution of k operators or less for

742
WSBen

propositional STRIPS planning, with restrictions (Watts & Strogatz, 1998) with probability
on negation in pre- and post-conditions (Nilsson, p. Lrandom is identical to L(1).
2001). Its computational complexity is proved • C: The average clustering coefficient. Sup-
to be NP-complete (Bylander, 1994). Therefore, pose that for a node i with vi neighbor,
when the number of Web services to search is not
2 Ei
small, finding an optimal solution to the WSC Ci = ,
vi (vi - 1)
problem (i.e., a chain of Web services to invoke)
is prohibitively expensive, leading to approximate where Ei is the number of edges between vi
algorithms instead. neighbors of i. C is the average clustering
coefficient Ci for a network. C(p) is defined
Complex Network Models as C of the randomly rewired Watts-Strogatz
graph with probability p. Crandom is identical
There are many empirical systems to form com- to C(1).
plex networks such as the scale-free network and
the small-world network, in which nodes signify Definition 3 (Small-world network) Small-world
the elements of the system and edges represent networks are characterized by a highly clustered
the interactions between them. topology like regular lattices and the small network
diameter, where the network diameter suggests
Definition 1 (Random network) A network is the longest shortest distance between nodes.
defined as the random network on N nodes, if each Specifically, small-world networks are C  Crandom
pair of nodes is connected with probability p. As a and L ≈ Lrandom (Delgado, 2002).
result, edges are randomly placed among a fixed set
of nodes. The random network can be constructed By using the Watts-Strogatz model (Watts,
by means of the Erdos-Renyi's random-graph 1999; Watts & Strogatz, 1998), we can construct
model (Erdos, Graham, & Nesetril, 1996). networks that have the small-world properties. The
model depends on two parameters, connectivity (k)
Definition 2 (Regular network) Rg(N, k) is defined and randomness (p), given the desired size of the
as the regular network on N nodes, if node i is graph (N). The Watts-Strogatz model starts with
adjacent to nodes [(i + j)mod N] and [(i - j)mod a Rg(N, k) and then every edge is rewired at random
N] for 1 ≤ j ≤ k, where k is the number of valid with probability p; for every edge (i, j), we decide
edge of each node. If k = N - 1, Rg(N, k) becomes whether we change j node (the destination node
the complete N-nodes graph, where every node of (i, j)) with probability p. The Watts-Strogatz
is adjacent to all the other N - 1 nodes. model leads to different graphs according to the
different p as follows:
We can define some metrics to quantify the
characteristic properties of the complex networks • When p = 0, an Rg(N, k) is built.
as follows: • When p = 1, a completely random network
is built.
• L: The average shortest distance between
reachable pairs of nodes, where the distance Otherwise, with 0 < p < 1, each edge (i, j) is
between nodes refers to the number of hops reconnected with probability p to a new node k that
between the nodes. L(p) is defined as L of is chosen at random (no self-links allowed). If the
the randomly rewired Watts-Strogatz graph new edge (i, k) is added, the (i, j) is removed from
the graph. The long-range connections (short-cuts)

743
WSBen

generated by this process decrease the distance to the probabilities of Equation (1), and the
between the nodes. For intermediate values of new link liz is added.
p, there is the “small-world” region, where the • With the probability of 1 - p - q, a new node
graph is highly clustered yet has a small average with m links is added. These new links con-
path length. nect the new node to m other nodes chosen
according to the probabilities of Equation
Definition 4 (Scale-free network) Networks (1).
are called scale-free networks if the number of • Once the desired number N nodes are
nodes that have v number of neighbor nodes is obtained, the algorithm stops. The graphs
proportional to Pw(v)∝v(-g), where g is typically generated by this algorithm are scale-free
greater than two with no humps. graphs, and the edges of the graphs are
constructed such that the correlations among
Barabasi and Albert provided several extended edges do not form. When p = q, the algo-
models (Albert, Jeong, & Barabasi, 1999; Delgado, rithm results in a graph, whose connectivity
2002) to provide the scale-free properties. The distribution can be approximated by
extended model uses an algorithm to build graphs
2 m (1- p ) +1- 2 p
that depend on four parameters: m0 (initial number P(v) ∝ (v + 1)
-
m
+1
(2)
of nodes), m(number of links added and/or rewired
at every step of the algorithm), p (probability of where v is the number of edges.
adding links), q (probability of edge rewiring).
The procedure starts with m0 isolated nodes and Diverse Web Service Network
performs one of the following three actions at Models
every step:
A set of Web services form a network (or graph).
• With the probability of p, m(≤ m0) new Depending on the policy to determine nodes and
links are added. The two nodes are picked edges of the network, there are varieties: Web
randomly. The starting point of the link is service level (i.e., coarse granularity), operation
chosen uniformly, and the end point of the level, and parameter level (i.e., fine granularity)
new link is chosen according to the follow- models. The graph at the middle of Figure 2 has a
ing probability distribution: bipartite graph structure and consists of three dis-
tinct kinds of vertices (i.e., parameter, operation,
vi + 1
Πi = (1) and web-service node) and directed arcs between
∑ j (v j + 1) bipartite nodes (i.e., operation nodes and parameter
nodes). An edge incident from a parameter node
where ∏i is the probability of selecting the i to an operation node means that the parameter is
node, and vi is the number of edges of node one of the inputs of the corresponding operation.
i. Reversely, an edge incident from an operation node
The process is repeated m times. to a parameter means that the parameter is one of
• With the probability of q, m edges are re- the outputs of the corresponding operation. The
wired. For this purpose, i node and its link graph has two Web services, labeled WS1 and WS2.
lij are chosen at random. The link is deleted. WS1 has two operations Op11 and Op12, and WS2 has
Instead, another node z is selected according one operation, Op21, respectively. There are seven
parameters, labeled P1 to P7. According to the node
granularity, we can project the upper graph into

744
WSBen

three different Web service networks. Op12 which, in turn, fully matches Op21 in
the upper graph. Therefore, Op11→Op12→Op21
• Parameter-Node Network: A graph Gp(Vp, can be shown in the operation node network.
Ep), where Vp is a set of all parameter nodes In particular, the fully matching operation
and Ep is a set of edges. An edge (pi, pj) is node network, Gopf has only Op12→Op21.
directly incident from input parameters pi ∈ • Web-service Node Network: A graph
Vp to output parameters pj ∈ Vp, where there Gws(Vws, Ews), where Vws is a set of all web-
is an operation that has an input parameter service nodes, and Ews is a set of edges. An
matching pi and an output parameter match- edge (wsi, wsj) is incident from web-service
ing pj. For example, P1→Op11→P3 in the upper node wsi ∈ Vws, to wsj ∈ Vws, where there is
graph is projected into P1→P3 in the parameter at least one edge between any operation in
node network. Figure 3 shows the parameter wsi and any operation in wsj. For example,
node network for PUB06 and the ICEBE05 since WS1 possesses Op12 and WS2 possesses
test set. Op21 in the upper graph, WS1→WS2 appears
• Operation-Node Network: A graph Gop(Vop, in the Web service node network.
Eop), where Vop is a set of all operation nodes,
and Eop is a set of edges. An edge (opi, opj) is
incident from operation opi ∈ Vp to operation Related works
opj ∈ Vp, here opi can fully or partially match
opj. For example, Op11 partially matches Constantinescu, Faltings, and Binder (2004)

Figure 2. Web services networks: (a) WSDLs, (b) Conceptual Web service network, (c) Web service
networks from diverse models, (d) Parameter node network, Gp, (e) Operation node network, Gop, (f)
Fully invocable operation node network, Gopf , and (g) Web service node network, Gws

745
WSBen

Figure 3. Diverse parameter networks. (left) PUB05. (right) ICEBE05

proposed a scalable syntactic test bed where Web obtained in the reasonable time window due to
services are generated as transformation between the problem’s inherently high complexity.
sets of terms in two application domains. For doing There are three unique challenges that have
that, they first defined parameter sets correspond- been established to investigate research issues with
ing to application domains and then, connected regard to Web services and Semantic Web. First is
those parameter sets randomly and constructed the Semantic Web Services Challenge5. This venue
a service graph which structure (i.e., nodes and invites application submissions for demonstrating
arcs) is similar to the parameter cluster network practical progress towards achieving the vision
of WSBen. However, WSBen takes a significant of the Semantic Web. According to the event, it
different approach to construct its parameter has the overall goal to advance our understanding
cluster networks in that WSBen does not simply of how semantic technologies can be exploited to
connect parameter sets at random but simulates produce useful applications for the Web. Second
topologies of real Web service networks. WSBen is is the Web Services Challenge6. This venue
inspired by extensive studies on real Web services, solicits approaches, methods, and algorithms
and therefore is designed to support various Web in the domain of Web-service discovery and
service network topologies and distributions. As composition. This event evaluates participants’
a result, WSBen can present more realistic testing approaches based on their quantitative and
situation for researchers who want to test their qualitative performance results on discovery
Web service discovery or composition algorithms and composition problems. The Web Services
than that of Constantinescu et al. (2004). Challenge is more driven by common problems,
XMark (XMark, 2006) is an XML benchmark but the Semantic Web Challenge concentrates
suite that can help identify the list of functions more on the environment. As such, the Semantic
which an ideal benchmark should support. WSBen Web Challenge places more focus on semantics
uses XMark as a reference model to identify nec- while the Web Services Challenges favors applied
essary functions to simplify the testing process. and short-term solutions (Brian, William, Michael,
One feature that is offered by XMark but not by & Andreas, 2007). Third is the Service Oriented
WSBen is the provision of solutions to queries. Architecture Contest7 which asks participants
In other words, XMark provides queries and to openly choose particular domain-specific
their corresponding solutions but WSBen gives problems and show their best approaches for
requests only because the optimal solution to a them. There are unique characteristics for each
Web service composition problem may not be venue so that they have undoubtedly contributed

746
WSBen

to advance the state-of-art technologies in Web tool. It is because METEOR-S consists of three
services and Semantic Web. Among these modules such as process designer, configuration
challenges, WSBen can be exploited especially module, and execution environment, where the
for the Web Services Challenge to provide execution environment requires executable Web
various benchmark environments, discovery and services but WSBen can generate only WSDL
composition problems by varying Web-service files without real implementation.
network topologies.
As for WSC, there are two main approaches, Overview of WSBen
depending on the use of domain knowledge.
First, the template-workflow based approach is In this section, we present a novel benchmark tool
to use software programs and domain experts to titled WSBen, which provides a set of functions
bind manually-generated workflows to the cor- to simplify the generation of test environments
responding concrete Web services. METEOR-S for WSD and WSC problems.
(Sivashanmugam, Verma, Sheth, & Miller, 2003)
is an example of this approach. Second, various AI Overview of WSBen
planning techniques have been applied to the WSC
problem, ranging from simple classical STRIPS- At a higher level, a Web service can be assumed
style planning to an extended estimated regression as a transformation between two different ap-
planning (McDermott, 2002). We believe that our plication domains, and each can be represented
WSBen is complementary for AI Planning based by a cluster. This assumption was the basis in
tools for the WSC problem. In fact, we demonstrate developing WSBen. From the perspective of
how WSBen can be used to compare the perfor- graph theory, WSBen builds Parameter Cluster
mance of AI planners for the WSC problems in Network, which consists of clusters and directed
the illustrative use-cases section. In this article, edges connecting two different clusters. These
meanwhile, we do not propose how METEOR-S directed edges become Web service templates
can make use of WSBen for a test case generation from which WSBen generates Web services as
users specify. Formally, the parameter cluster

Figure 4. Overview of WSBen

747
WSBen

network is defined as follows: STRIPS format, enabling concurrent com-


parison with state-of-the-art AI planners.
Definition 5 (Parameter Cluster Network) A
directed graph Gcl(Vcl, Ecl), where Vcl is a set of WSBen input framework: xTS
clusters and Ecl is a set of directed edges that are WSBen input framework, xTS consists of five
incident from input clusters i ∈ Vcl to output clusters tuples, <|J|, Gr, h, Gp,|W|>. In detail:
j ∈ Vcl. Here, cluster i and j contain a set of non-
overlapping parameters denoted by Pai and Paj, 1. |J| is the total number of parameter clus-
respectively, where Pai ∩ Paj = f. Each directed ters.
edge is also called a Web service template, from 2. Gr denotes a graph model to specify the
which WSDL files are generated. underlying topology of a parameter cluster
network. Gr can be on of the following
Figure 4 shows the overview of WSBen. In three models discussed in the Background
detail, WSBen consists of the following func- section:
tionalities: • Erdos-Renyi(|J|, p): This model has such a
simple generation approach that it chooses
• Input framework: Users can specify and each of the possible
control the generated synthetic WSDL files | J | (| J | -1)
and their characteristics. For this purpose, 2
WSBen provides an input framework xTS
= <|J|, Gr, h, Gp,|W|>. xTS applies existing edges in the graph with |J| nodes with
complex and random network models to probability p. The resulting graph be-
specify Gr. Each element of xTS will be comes the same as the binomial graph.
discussed in more detail below. • Newman-Watts-Strogatz(|J|, k, p): The
• Parameter cluster network, Gcl(Vcl, Ecl): If initialization is a regular ring graph
xTS is given by users, based on the first with k neighbors. During the generation
four elements, WSBen generates Gcl. Each process, new edges (shortcuts) are added
cluster of Gcl is filled with some number of randomly with probability p for each edge.
atomic parameters. In this network, Web Note that no edges are removed, differing
services are defined as transformations be- from Watts-Strogatz model.
tween two different clusters. That is, <i, j> • Barabasi-Albert(|J|, m): This graph model
∈ Ecl becomes Web service templates. The is generated by adding new nodes with m
role of Web service templates in the test set edges that are preferentially attached to
generation will be illustrated. existing nodes with a high degree. The
• Test set and sample requests: By randomly initialization is a graph with m nodes and
selecting the Web service templates (arcs of no edges. Note that the current implemen-
the parameter cluster network), WSDL files tation of WSBen is limited because it can
are generated. Once a test set is generated, only generate the simplified version of
users can generate sample test requests r = the extended Barabai-Albert model, by
<ri, ro >. The generation process of test sets setting p = q = 0 and m0 = m, resulting
and test requests will be illustrated. in graphs with g = 2.0 ± 0.1, where g is
• Test and evaluation: WSBen can export both the exponent of a power function Pw(v)
the Web service WSDL files and test requests defined over connectivity v range in the
into files in PDDL (McDermott, 1996) and form of Pw(v) ∝ v-g.

748
WSBen

3. h denotes the parameter condense rate. With


h, users can control the density of partial- 1. A Web service template <i, j> is chosen at
matching cases in produced Web services. random from Ecl.
4. Mp denotes the minimum number of param- 2. WSBen generates a WSDL file, in which each
eters a cluster can contain. In other words, input parameter is selected from i cluster with
clusters may have a different number of probability ∆i, and each output parameter is
parameters but all clusters must have at least selected from j cluster with probability ∆j.
Mp number of parameters.
5. |W| denotes the total number of Web services Figure 5 illustrates how WSBen builds Gcl and
of a test set. generates WSDL files based on the Gcl. Suppose
that xTS = <8, Barabasi - Albert (8, 2), 0.8, 1.5,
With |J| and Gr, the first two input elements 100> is given. Then, the generation steps are as
of xTS, we can build Gcl with each empty cluster. follows:
Thus, we need a procedure to fill each empty
cluster with parameters. For this purpose, WSBen 1. WSBen generates a graph of Barabasi-
uses the following procedure: Albert(8,2). The direction of each edge is
determined at random.
1. A parameter cluster network Gcl with empty 2. ∆j and |Paj| are specified. For example,
clusters is built by specifying |J| and Gr. Cluster 5 has nine parameters as shown in
2. Co-occurrence probability of each cluster Figure 5. That is, |Pa5| =9, as
is measured by the following probability:
kj 1
kj ∆5 = = × 0.8 = 0.16,
∆j = max j∈Vcl k j 5
max j∈Vcl k j (3)
resulting in
where ∆j is the co-occurrence probability of
Mp
cluster j, and kj is the edge degree of cluster | Pa5 |= ≈ 9.
j. ∆5
h is the parameter condense rate which is 3. Paj is specified. For example Pa5={17,18,1
given by users. 9,20,21,22,23,24,25} as shown in Figure 5
3. |Paj| is measured based on the following because |Pa5| = 9 and for ∀i, j ∈ Vcl, Pai ∩
equation. Paj = f. Note that the parameter names are
Mp automatically generated, and thus do not
| Pa j |= (4) contain any semantics.
∆j
4. Finally, Gcl is built and WSBen generates
where Paj is the set of parameters contained |W| Web services. For example, in Figure
in cluster j. 5, WS1 is instanced from a Web service
4. For each j cluster, atomic parameters are template <3, 1> ∈ Ecl. Note that ∆1 = 0.16
generated up to |Paj|, with duplicated pa- and ∆3 = 0.8. ∆1 = 0.16 suggests that the
rameters forbidden (i.e., ∀i, j ∈ Vcl, Pai ∩ occurrence probability of each parameter
Paj = f). in Cluster 1 has 0.16. Due to the low prob-
ability, only “1” and “9” are selected from
Once a complete parameter cluster network, Cluster 1. Similarly, ∆3 = 0.8 means that the
Gcl(Vcl, Ecl) is built, WSBen repeats the following occurrence probability of each parameter in
procedure until |W| number of Web services are Cluster 3 has 0.8. Due to the high probabil-
generated:

749
WSBen

Figure 5. Test set generation with <8, Barabasi - Albert (8, 2), 0.8, 1.5, 100>

Algorithm 1. Forward searching algorithm of


ity, all parameters in Cluster 3 that are “13” WSBen
and “14” are selected. In the case where no
parameter is generated, dummy parameters Test Request Generation
“S” and “T” are filled in the input and output Input : ri
Output: g r i(p) for all p reachable from ri
parameters, respectively. 1: s = ri; C = f; d = 1;
2: while (d ≠ f) do
3: d = {w | w ∈ Ω(s), w ∉ C};
The state, s ∈ S is a collection of parameters in 4: for p in wo(w ∈ d) do
|P|. Therefore, ri and ro are states. The test request 5: if g r i(p) = ∞ then
6: g r i(p) = d; s = s ∪ {p};
r is constructed such that ro is farthest away from 7: C = C ∪ d; d++;
ri in a parameter space in terms of g r (p) - the cost
i

of achieving p ∈ P from a state ri. To obtain g r i

(p), we propose following Forward Searching w ∈ W and is assumed to be 1. Ow(p) is a set of


algorithm. Web services: Ow(p) = {w ∈ W | p ∈ wo}. At
first, g r (p) is initialized to 0 if p ∈ ri, and to ∞
i

Forward Searching: g r (p) can be character-


i otherwise. Then, the current information state
ized by the solution of a recursive equation as s is set to ri (Line 1 in Algorithm 1). We denote
follows: Ω(s) by a set of Web services w ∈ W such that
wi ⊆ s. That is, w can be invoked or applicable in
g r i ( p ) = min [c( w) + maxi g r i ( p ')] (5) the state s.
w∈Ow ( p ) p '∈w
Every time for ∀w ∈ Ω(s), each parameter p
where c(w) is an invocation cost of a Web service, ∈ wo is added to s, and g r (p) is updated until Ω(s)
i

750
WSBen

stops to increase, meaning that this process ends 1. WSBen selects a Cluster j ∈ Gcl at ran-
with finding g r (p) for all parameters reachable
i dom.
from ri (Lines 2-6 in Algorithm 1). 2. WSBen copies all parameters in the Cluster
We can use Equation (5) to drive the lower j (i.e., Paj) into ri, and then ro is constructed
bound of the optimal cost of WSC solutions. Note so that it consists of the first five largest
that the invocation cost of a Web service is as- parameters in terms of g r (p). Consequently,
i

sumed to be 1. Thus, the optimal cost of a WSC parameters in ro are farthest away from
problem coincides with the minimum number of parameters in ri in a parameter space.
Web services required to solve the WSC problem.
For a set of parameters A, we can regard the fol- As a default, WSBen repeats the above pro-
lowing cost function: cedure five times, generating five request sets for
each test set.
g rmax
i ( A) = max g r i ( p ) (6)
p∈ A
Implementation
The cost of achieving a set of parameters
cannot be lower than the cost of achieving each As shown in Figure 6, WSBen provides user inter-
of the parameters in the set. Thus, g rmax ( A) is the
i faces to specify xTS and several parameters, which
lower bound of the optimal cost of achieving ro are required to form a parameter cluster network
from ri. and generate WSDL files. WSBen is implemented
Based on the forward searching algorithm, in Python, and run on Python 2.3 or later. It runs
WSBen create a test request r, as follows: on Unix and Windows. For the creation, ma-
nipulation, and functions of complex networks,
we used a Python package called NetworkX8.

Figure 6. WSBen user interface

751
WSBen

Current implementation of WSBen is limited as terms of Gp and outgoing edge degree distribu-
follows: (1) it supports only the exact matching tion.
without type compatibility check, and (2) each
Web service contains only one operation so that
a Web service can be viewed as equivalent to an Illustrative use cases of
operation. Therefore, wi and wo indicate the input WSBen
and output parameter set of a Web service, w.
Figure 6 also shows three sample parameter In this section, we present three use cases to
cluster networks, where each circular node rep- demonstrate the application of WSBen: (1)
resents a cluster and edges with heads denote the evaluating Web-service composition algorithms;
Web service template, from which Web services (2) comparing performance of AI planners;
are instanced. The size of node is proportional to and (3) estimating the size of giant component.
the number of parameters in the node, while the These use cases are prepared to provide vigorous
transparency level of a node's color is inversely experiments and evaluation for assessing the usage
proportional to the degree of the node. For ex- of WSBen. For each use case, we will provide three
ample, in the left cluster network, Cluster 18 can Web-service test sets by varying xTS with three
be considered a hub cluster in that it has the high parameter cluster networks such as “random”,
degree. Therefore, it is presented by a small circle “small-world”, and “scale-free” types. Note that
with denser color. these three network models have the expression
Following the mechanism of WSBen explained power enough to model many real-world net-
so far, we can build three illustrative test set works sufficiently (Albert & Barabasi, 2002).
frameworks by specifying xTS as follows: This implies that our generated test cases can be
appropriate for representing diverse real-world
1. baTS = <100, Barabasi - Albert (100, 6), Web-service networks. Furthermore, these three
0.8, 5, |W|> Web-service test sets are significantly distinctive
2. nwsTS = <100, Newman - Watts - from each other in terms of their Web-service
Strogatz(100, 6, 0.1), 0.8, 5, |W|> network topologies and degree distributions as
3. erTS = <100, Erdos - Renyi(100, 0.006), 0.8, we have shown in the previous section. This
5, |W|> indicates that we have sufficient reason to analyze

Figures 7 and 8 show that there are distinctive


differences between baTS, nwsTS, and erTS in
Figure 7. Gp of baTS, erTS, and nwsTS when |W| = 1,000

752
WSBen

Figure 8. Outgoing edge degree of baTS, erTS, and nwsTS when |W| = 1,000

Figure 9. Composed services using WSPR for three test sets. (left) baTS with |W| = 5,000. (center) erTS
with |W| = 5,000. (right) nwsTS with |W| = 5,000

how different network topologies can affect the are shown in Figure 9, where WSPR addressed a
performance of Web-service applications or request for each of the three test sets. Note that
environments. WSBen can automatically create sample requests
for a given test set. In the graph, each composed
Evaluating Web-Service solution has nodes such as “Ri” and “Ro”, which
Composition Algorithms represent the initial condition and goal state, re-
spectively. Other nodes represent Web services.
Recently, many WSC researches have been re- The directed arcs indicate the invocation flow,
ported in the Web service community. As such, where a solid edge means full-matching invoca-
the EEE06 Web-service composition contest holds tion and a dotted edge represents partial-matching
as many participants as 11. Among the 11 WSC invocation. From the experiments based on diverse
algorithms, we choose a WSC algorithm named test sets such as baTS, erTS, and nwsTS, we can
WSPR (Oh, Lee, & Kumara, 2007), which was understand how different network models of Gcl
proved effective and efficient in the contest, in influences the performance of WSC algorithms.
order to demonstrate the application of WSBen. In general, given the same number of clusters, the
In this case, we use three test sets generated by Barabasi-Albert model generates Gcl with a greater
WSBen: (1) baTS with |W| = 5,000; (2) erTS with number of parameters, and a larger variance of the
|W| = 5,000; and (3) nwsTS with |W| = 5,000. The number of parameters between clusters than the
resultant composed services generated by WSPR Newman-Watts-Strogatz and Erdos-Renyi models

753
WSBen

do. Due to the greater number of parameters and it can run even in large number of operators. For
larger variance, baTS needs more partial-matching comparing the performance of three planners, we
Web services to fulfill the given requests than use two evaluation metrics as follows:
others. The increasing need for partial-matching
Web services leads to increasing number of Web 1. t(Time): It measures how long an algorithm
services in the composed service. This is the takes to find a right solution, in seconds. This
reason that the baTS case has more Web services is a measure of computational efficiency.
to create a resultant composed service as shown 2. #W: The number of Web services in a right
in Figure 9 (left). solution. This is a measure of effective-
ness.
Comparing Performance of AI
Planners All AI planners run with their default options,
except that the maximum number of nodes for
We demonstrate how WSBen can be used to Blackbox and Graphplan was set to 32,768 and
compare the performance of AI planners. For this 10,000, respectively. Commonly, the time to
purpose, we choose three prominent AI planners – read operator and fact files is not included in t.
Graphplan (Blum & Furst, 1997), Blackbox (Kautz Blackbox and IPP accept only the PDDL format,
& Selman,1996), and IPP9. Blackbox and IPP are while Graphplan accepts only the STRIPS for-
extended planning systems that originated from mat for their operator and fact files. Note that an
Graphplan. In particular, Blackbox is extended to operator file corresponds to a test set, and a fact
be able to map a plan graph into a set of clauses for file corresponds to a test request file. Also note
checking the satisfiability problem. Consequently, that WSBen provides a function to convert test

Table 2. Results over baTS with |W| = 3,000


BlackBox GraphPlan IPP
Requests
#W t #W t #W t
r1 61 478.69 - - - -
r2 - - - - - -
r3 5 5 5 0.09 5 26.22
r4 9 27.78 9 0.11 9 28.56
r5 4 1.4 4 0.04 4 23.97

Table 3. Results over erTS with |W| = 3,000


BlackBox GraphPlan IPP
Requests
#W t #W t #W t
r1 75 38.09 - - - -
r2 50 16.02 - - - -
r3 22 18.68 - - 22 24.78
r4 23 4.38 - - 23 21.06
r5 38 4.01 - - 38 21

754
WSBen

sets and requests into PDDL and STRIPS files suggested the theoretical framework in order to
automatically. The experiments were performed estimate the giant component size in networks
on Linux with three Intel Xeon CPU, running by using random graph theory. In order to see
at 2.4GHz with 8GB RAM. if their theoretical framework works, we gener-
Tables 2, 3, and 4 shows the results of the five ated the g opf with different network size for each
test requests for each of baTS, erTS, and nwsTS of following cases:
with |W| = 3,000. Graphplan ran out of memory
in many cases. IPP also failed to solve the some 1. Random model: <50, Erdos - Renyi(100,
requests. As a whole, Blackbox showed better 0.6), 0.8, 5, |W|>
performance than others, meaning that it can solve 2. Scale-free model: <50, Barabasi - Al-
more requests than others. It is because Blackbox bert(100, 6), 0.8, 5, |W|>
uses the local-search SAT solver, Walksat, for 3. NWS model: <50, Newman - Watts -
the satisfiability problem, so that Blackbox can Strogatz(100, 6, 0.1), 0.8, 5, |W|>
run relatively well even with a large number of
operators. For each of these networks, we measured the
We can estimate the size of giant component real size of giant components. Then, we calculated
in a service network using random graph theory. the theoretical size of giant components according
Often it is believed to be important to have a to the estimation model of Newman et al. (2001).
large and dense giant component in a service The comparisons between real and theoretical
network. Otherwise, the isolated services will sizes are summarized in Figure 10. For g opf based
never have a chance to provide any services to on the random parameter cluster network in Figure
clients. Newman, Strogatz, and Watts (2001) 10(A), the theoretical value of the giant component

Table 4. Results over nwsTS with |W| = 3,000


BlackBox GraphPlan IPP
Requests
#W t #W t #W t
r1 48 571.63 - - 48 29.52
r2 35 114.67 - - 35 28.57
r3 24 192.99 - - 24 30.19
r4 26 11.88 - - 26 28.39
r5 31 111.21 - - - -

Figure 10. Comparison of real and theoretical size of giant components: (A) random, (B) scale-free,
and (C) NWS models.

Estimating the size of giant component

755
WSBen

size is very close to the measured one for each ACKNOWLEdGMENT


synthetic network. This implies that even a simple
random model may be very helpful to estimate Authors would like to thank Hyunyoung Kil and
the inter-operable portion of such networks with Ergin Elmacioglu for their help on the design and
random topology without even analyzing the implementation of WSBen, and Professor Soundar
available network beyond its degree distribution. R. T. Kumara at Penn State for his helpful com-
However, Figure 10(B) shows that the estimation ments on the earlier draft of this article.
model is not a good model for Scale-free type.
There is a considerable gap between theory and
real value for many of the synthetic networks in REFERENCES
this type. The deviation between theory and actual
networks becomes more dramatic for the NWS Albert, R. & Barabasi, A. (2002). Statistical me-
(small world phenomenon and highly clustered chanics of complex networks. Reviews of Modern
property) type shown in Figure 10(C). The results Physics, 74(1), pp. 47-95.
show that the random network model might be
good generative model for such Web services Albert, R. & Barabasi, A. (2000). Topology of evolv-
networks if these networks are entirely random, ing networks. Phys. Rev.Lett., 85, pp. 5234-5237.
which is also in accordance with the basic as- Albert, R., Jeong, H. & Barabasi, A. (1999). The
sumption by Newman et al. (2001). diameter of the world wide web. Nature, 401, pp.
130-131.

Conclusion Bertsekas, D. (2000). Dynamic programming


and optimal control, volume 1, 2nd ed. Athena
A novel Web-service benchmark toolkit, WSBen, Scientific, Boston.
is presented with three use cases in different Blum, A. & Furst, M. (1997). Fast planning through
application domains. The WSBen development is planning graph analysis. Artificial Intelligence, 90,
inspired by the study on real-world Web services, pp. 281-300.
and is designed to provide diverse scenarios and
configurations which users can fine-tune easily. Brian, B., William, C., Michael, J., & Andreas,
As a result, using WSBen, users can conduct W. (2007). WSC-07: Evolving the web services
extensive experimental validation on their Web- challenge. In Proc. of 9th IEEE International
service discovery and composition algorithms. Conference on E-Commerce Technology and 4th
It is our hope that WSBen will provide useful IEEE International Conference on Enterprise
insights to the design and development of Web- Computing, E-Commerce, and E-Services (CEC-
services discovery and composition solutions and EEE 2007), pp. 505-508, Tokyo, Japan.
software. The latest version of WSBen is available Bylander, T. (1994). The computational complex-
at: http://pike.psu.edu/sw/wsben/. Further research ity of propositional STRIPS planning. Artificial
is needed to extend WSBen to support approximate Intelligence, 69(1-2), pp. 165-204.
and semantic matching among Web services. Also,
we plan to discover additional applications where Constantinescu, I., Faltings, B. & Binder, W. (2004).
the WSBen can be used. Large scale testbed for type compatible service
composition. In Proc. of 14th International Con-
ference on Automated Planning and Scheduling
(ICAPS), pp. 23-28, Whistler, British Columbia,
Canada.

756
WSBen

Delgado, J. (2002). Emergence of social conven- Oh, S., Kil, H., Lee, D., & Kumara, S. (2006). WS-
tions in complex networks. Artificial Intelligence, Ben: a Web-services discovery and composition
141, pp. 171-185. benchmark. In Proc. of the fourth International
IEEE Conference on Web service (ICWS), pp.
Den ni ng, J. P. (2004). Net work laws.
239-246, Chicago.
Communications of the ACM, 47(11), pp. 15-20.
Oh, S., Lee, D., & Kumara, S. (2007). WSPR: a
Erdos, P, Graham, R., & Nesetril, J. (1996). The
heuristic algorithm for Web-service composition.
mathematics of Paul Erdos. Springer-Verlag,
Int’l J. of Web services Research (IJWSR), 4(1),
Berlin, Germany.
pp. 1-22.
Fan, J., & Kambhampati, S. (2005). A snapshot
Sivashanmugam, K., Verma, K., Sheth, A., &
of public Web services. SIGMOD Record, 34(1),
Miller, J. (2003). Adding semantics to Web ser-
pp. 24-32.
vices standards. In Proc. of the first IEEE Inter-
Kautz, H. & Selman, B. (1996). Unifying SAT- national Conference on Web services (ICWS), pp.
based and graph-based planning. In Proc. of the 395-401, Las Vegas, NV.
16th International Joint Conference on Artificial
Watts, D. J. (1999). The dynamics of networks
Intelligence (IJCAI), pp. 318-325, Stockholm,
between order and randomness. Princeton Univ.
Sweden.
Press, Princeton, NJ.
Kil, H., Oh, S., & Lee, D. (2006). On the Topologi-
Watts, D. J., & Strogatz, S. H. (1998). Collective
cal Landscape of Semantic Web services Match-
dynamics of small-world networks. Nature, 393(4),
making. In Proc. of 1st International Workshop on
pp. 440-442.
Semantic Matchmaking and Resource Retreival
(SMR2006), pp. 19-34, Seoul, Korea. XMark. An XML Benchmark Project. Retrieve
on September 17, 2006 from, http://monetdb.cwi.
McDermott, D. (1996). A heuristic estimator for
nl/xml/.
means-ends analysis in planning. In Proc. of the
third International Conference on Artificial In-
telligence Planning Systems (AIPS), pp. 142-149,
Edinburgh, Scotland. Endnotes
McDermott, D. (2002). Estimated-regression

1
This article is a substantial extension from
planning for interactions with Web services. In
the short version that appeared in the
Proc. of the sixth International Conference on
proceedings of the 4th International IEEE
Artificial Intelligence Planning and Scheduling
Conference on Web Services (ICWS), held
Systems (AIPS), pp. 67-73, Toulouse, France.
in Chicago, USA, 2006. The work of Seog-
Newman, M. E. J., Strogatz, S. H., & Watts, D. Chan Oh was done while the author was
J. (2001). Random graphs with arbitrary degree with the Pennsylvania State University.
distributions and their applications. Phys. Rev.
2
As of August 2007, according to the esti-
E, 64(026118). mation of Google Scholar, there are about
2,360 scholarly articles mentioning “Web
Nilsson, J. (2001). Artificial Intelligence: a new
Services Composition”.
synthesis. Morgan Kaufmann, San Francisco,

3
http://www.comp.hkbu.edu.hk/~ctr/wschal-
CA.
lenge/

4
http://insel.flp.cs.tu-berlin.de/wsc06/

757
WSBen

5
The Semantic Web Services Challenge 6
The Web Services Challenge (2007): http://
(2007): http://sws-challenge.org/wiki/index. www.wschallenge.org/wsc07/
php/Main_Page

This work was previously published in International Journal of Web Services Research, Vol. 6, Issue 1, edited by J.N.D. Gupta,
S.K. Sharma, and J. Hsu, pp. 1-19, copyright 2009 by IGI Publishing (an imprint of IGI Global).

758
759

Chapter 3.6
Architecture of the
Organic.Edunet Web Portal
Nikos Manouselis
Greek Research & Technology Network (GRNET S.A.), Greece

Kostas Kastrantas
Greek Research & Technology Network (GRNET S.A.), Greece
Salvador Sanchez-Alonso
University of Alcalá, Spain

Jesús Cáceres
University of Alcalá, Spain

Hannes Ebner
Royal Institute of Technology (KTH), Sweden

Matthias Palmer
Royal Institute of Technology (KTH), Sweden

Ambjorn Naeve
Royal Institute of Technology (KTH), Sweden

Abstract cifically, it describes the architecture behind the


Organic.Edunet Web portal that aims to provide
The use of Semantic Web technologies in educa- access to a federation of repositories with learn-
tional Web portals has been reported to facilitate ing resources on agricultural topics. The various
users’ search, access, and retrieval of learning components of the architecture are presented and
resources. To achieve this, a number of different the supporting technologies are explained. In ad-
architectural components and services need to be dition, the article focuses on how Semantic Web
harmonically combined and implemented. This technologies are being adopted, specialized, and
article presents how this issue is dealt with in the put in practice in order to facilitate ontology-aided
context of a large-scale case study. More spe- sharing and reusing of learning resources.

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Architecture of the Organic.Edunet Web Portal

Introduction examples include both independent learning re-


sources’ portals such as MERLOT (http://www.
Following their introduction and commercial merlot.org) and Teachers’ Domain (http://www.
growth after 2000, Web portals have lately at- teachersdomain.org/), as well as portals that list or
tracted increased research interest that focuses aggregate learning resources from various other
on a variety of aspects such as their business sources (e.g. other portals or repositories) such as
models, interface design, technical development, OERCommons (http://www.oercommons.org).
or their quality (Mahadevan, 2000; Tatnall, 2005a; Richards et al. (2002) stress that Web portals
Moraga et al., 2006; Tatnall, 2007). The term Web with learning resources may offer a wide variety
portal has been initially used to refer to well- of services based on what they seek to give to the
known Internet search and navigation sites that user community behind them, although the more
provided a starting point for web visitors to explore common are those aimed at facilitating users’
and access information on the World Wide Web search, access, and retrieval of the resources.
(Warner, 1999; Winkler, 2001). A Web portal can For this purpose, they include services that will
be generally viewed as a single, distilled view of facilitate these processes, utilizing different types
information from various sources that integrates of user-related information (such as personal pref-
information, content, and other software services erences) or resource-related information (such as
or applications (Averweg, 2007). Therefore, today the learning resource characteristics). One of the
Web portals can be simply defined as gateways to most recent trends in portal development is the use
information and services from multiple sources, of Semantic Web technologies (Maedche et al.,
and their continuous development has been 2001). Semantic Web is an evolving extension of
highlighted by relevant publications (Tatnall, the World Wide Web (WWW) in which web con-
2005b). tent can be expressed not only in natural language,
A type of Web portals with particular interest but also in a format that can be read and processed
are educational ones (Conceicao et al., 2003; Boff by software systems, thus permitting them to
et al., 2006). Educational Web portals generally find, share and integrate information more eas-
serve as gateways to information and services of ily (Berners-Lee, 1998). Numerous applications
some learning or teaching relevance and may cover and case studies of Semantic Web technologies
a variety of types. They range from institutional (e.g. ontologies for annotating information and
Web portals that provide access to course listings expressing its semantics in a machine-processable
and institutional information (such as Ethridge manner) have been reported during the past few
et al, 2000), to community portals that serve the years. For instance, the World Wide Web Consor-
needs of particular communities of learning and tium (W3C) reports on several systems that have
practice (such as DeSanctis et al., 2001; Luke et been put in production in existing organizations,
al., 2004). One category of educational portals as well as a number of commercial products (http://
that have recently received considerable interest esw.w3.org/topic/CommercialProducts). Yet,
(Neven & Duval, 2002; Richards et al., 2002; the Semantic Web technologies have not so far
Hatala et al., 2004) is that of Web portals that reached the wide public. Some of the experts in
provide access to some organized collection of the field claim that the reason is that large-scale
learning resources. These portals usually facili- applications, serving the needs of large user com-
tate users’ access to the content in one or more munities, have not been delivered yet (Shadbolt
learning repositories – that is, to database systems et al., 2006). To further illustrate their potential
that facilitate the storage, location and retrieval (and especially for Web portals), there is a need
of learning resources (Holden, 2003). Popular for implementing state-of-the-art Semantic Web

760
Architecture of the Organic.Edunet Web Portal

technologies in large-scale applications. In the that would be able to mediate and help. As
context of educational Web portals, this involves Berners-Lee reports, one of the major obstacles
the semantic annotation of big collections of to this has been the fact that most information on
learning resources and their access and use from the Web is designed for human consumption, and
existing communities of users. even if it was derived from a very well specified
This article aims to contribute to this devel- database, the structure of the data is not evident
opment by presenting such a large-scale imple- to an automated software system browsing the
mentation effort. More specifically, it discusses Web. On the contrary, in his vision of the Semantic
how semantic annotation and Semantic Web Web, data recovery for a particular context of use
technologies are being adopted, specialized, should be a routine, automated process. This is the
and put in practice in order to set up a technical reason why the empowering role of Semantic Web
infrastructure that will facilitate sharing and reus- technologies has already been acknowledged in
ing of learning resources for an educational Web various contexts, such as education and training
portal. The case study is the Organic.Edunet Web (Aroyo & Dimitrova, 2006). In this new paradigm,
portal, a portal that serves the needs of learning data would be specifically oriented to machine
and teaching communities of the agricultural consumption by means of formal descriptions
sector, by facilitating their access to a network based on the existence and wide availability of
(also called a federation) of learning repositories ontologies, knowledge models of a given domain.
with learning resources on Organic Agriculture Ontologies, which can be defined as collections
(OA) and Agroecology (AE) topics. of concepts representing domain-specific entities,
The article is structured as it follows. First, a the relationships between those concepts, and
short review of the way Semantic Web approaches the range of admissible values for each concept
are being implemented in similar applications is (Daraselia et al., 2004) are in fact the key ele-
carried out. A description of the Organic.Edunet ment of the Semantic Web. Ontologies serve as
initiative is given, and the rationale for developing knowledge models for each particular domain of
the Organic.Edunet Web portal is outlined. The science, thus allowing to unambiguously repre-
main part of the article focuses on the description sent, refer, and describe entities in that domain,
of the technical architecture of the Web portal, and serving as the basis for interoperability and
and on the way Semantic Web technologies are common understanding under formal and strict
implemented in it. A discussion of perceived semantics.
benefits and potential challenges is later carried Since the Web is becoming a popular educa-
out, to finally provide the main conclusions of tional medium at schools, universities and pro-
this work. fessional training institutions, a prominent new
stream of research on the Educational Semantic
Web has been established. Research studies
Background already report semantic-based annotation and
sharing of learning resources. For example, Forte
Educational Semantic Web et al. (1999) report on the principles underlying
the semantic and pedagogic interoperability
From its initial conception around 1989 (Berners- mechanisms built in the European Knowledge
Lee, 1998), the WWW (or simply, the Web) has Pool System, a distributed repository of learning
been designed as an information space, with the objects developed by the European research proj-
goal that it should be useful not only for human- ect ARIADNE (http://www.ariadne-eu.org).
to-human communication, but also for machines In addition, Soto et al. (2007) designed an
ontology schema capable to bring more flex-

761
Architecture of the Organic.Edunet Web Portal

ibility to the description of the entities stored in institutions, from elementary schools to relevant
semantic learning object repositories and, at the university departments. But apart from raising
same time, to facilitate automated functions and the awareness and education level of consumers,
task delegation to agents. Furthermore, Sicilia agricultural professionals must also be properly
et al. (2005) describe the design of a learning educated. By “agricultural professionals” we
object repository approach to what they called refer to the different types of future agricultural
“semantic lifecycle” and illustrate thus through experts (e.g. natural production experts, veterinary
the concrete architecture of a semantic learning experts, agricultural economists, extension offi-
object repository prototype. cers, etc.), who study in agricultural universities
Moreover, semantic web applications are around Europe, and who should be provided with
becoming more and more usual in education & a wide range of information related to OA and
training contexts. Sancho et al. (2005) for in- AE theories, methods, practices, and economic/
stance, applied these technologies to e-learning environmental impacts.
personalization by combining the information Both groups (pupils and young agricultural stu-
provided by ontologies, and the user profile, to dents) constitute user groups of high importance.
create personalized units of learning. Santos Children constitute tomorrow’s consumers, and
et al. (2006) described an approach to promote they have to be properly approached and educated
interoperability among heterogeneous agents so that their nutritional, as well as their ecologi-
that are part of an educational portal. Their main cal and environmental awareness are developed.
contribution was to provide a means for social Students of agricultural universities constitute
agents to communicate with agents outside its tomorrow’s agricultural professionals. They are
original scope through the use of semantic web expected to guide farmers through the adoption of
technologies. Other implementations of Semantic OA and AE principles, or to serve themselves as
Web technologies in educational portals also exist the next generation of farmers/producers. There-
in the literature (Woukeu et al., 2003; Tane et al., fore, these two user groups have to be carefully
2004; Moreale & Vargas-Vera, 2004; Verdejo et approached through publicly available, quality,
al., 2004; Kotzinos et al., 2005). and multilingual educational content.
To further illustrate the potential of the In this direction, the Organic.Edunet initiative
Educational Semantic Web, there is a need for (http://www.organic-edunet.eu), a European proj-
implementing state-of-the-art Semantic Web ect that is funded by the eContentplus Programme
technologies in large-scale applications that in- and which involves 15 partners from 10 countries,
volve the semantic annotation of big collections aims to facilitate access, usage and exploitation of
of learning resources and their access and use digital educational content related to OA and AE.
from existing communities of users. Organic.Edunet will deploy a multilingual online
federation of learning repositories, populated with
Organic.Edunet quality content from various content producers.
In addition, it will deploy a multilingual online
To further promote the familiarization of con- environment (the Organic.Edunet Web portal) that
sumers with the benefits of OA and AE - for will facilitate end-users’ search, retrieval, access
their own health as well as for the benefits of the and use of the content in the learning repositories.
environment - the most dynamic consumer groups In this way, digital content resources that can be
have to be properly educated. Young people at all used to educate European youth about the benefits
stages of formal education have to be carefully of OA and AE, will become easily accessible,
approached through relevant educational pro- usable and exploitable.
grams in the curricula of all kinds of educational

762
Architecture of the Organic.Edunet Web Portal

To achieve its aims, Organic.Edunet adopts ments for describing semantic Web Services:
state-of-art technologies that have been developed ontologies that provide the terminology used by
and tested in several research initiatives, but have other elements, goals that define the problems that
yet to be proven in a real-life context. A charac- should be solved by Web Services, Web Services
teristic example involves the implementation of descriptions that define various aspects of a Web
Semantic Web technologies that have been pre- Service, and finally mediators which bypass inter-
viously developed in the context of the “LUISA: pretability problems. In the following section, we
Learning Content Management System Using describe the overall architecture of the Organic.
Innovative Semantic Web Services Architecture” Edunet infrastructure, and how technologies such
EU project (http://www.luisa-project.eu/). The as the ones adopted from LUISA are engaged.
main characteristics of the LUISA architecture
are its service-orientation and the built-in capa-
bilities for semantic querying. For this purpose, Organic.Edunet
semantic Web Services are involved, reusing the Infrastructure
EU framework WSMO (http://www.wsmo.org/)
for the brokering of multiple repositories. WSMO Overall Architecture
(which stands for Web Service Modeling Ontol-
ogy) provides ontological specifications for the The overall architecture of Organic.Edunet is il-
core elements of Semantic Web services. Taking lustrated in Figure 1. The main elements of this
the Web Service Modeling Framework (WSMF) architecture are the following:
as reference, WSMO defines four different ele-

Figure 1. Overall architecture of Organic.Edunet


Users

Web Portal Module

Organic.Edunet’s
Learning repositories
semantic services Module

bioAgro

ENOAt
content
Ecologica Learning repositories Management Module Providers

Intute

Public
resources

school

Learning content
Exchange Module

Federation of Learning
content repositories

763
Architecture of the Organic.Edunet Web Portal

• Learning Repository Management Module: queries and the harvesting of metadata.


includes the suite of tools that the Organic. Organic.Edunet is expected to be connected
Edunet content providers will use to create with two external federations; the Learning
a digital collection of learning resources, to Resource Exchange (LRE) of the European
describe resources with appropriate meta- Schoolnet (http://lre.eun.org) and the ARI-
data, and to publish resources in their own ADNE Foundation (http://www.ariadne-eu.
learning repository. Overall, six learning org/).
repositories are expected to be set up by the • Semantic Services Module: it is the core of
Organic.Edunet content providers (namely the Semantic Web technologies’ application
the Bio@gro, ENOAT, ECOLOGICA/COM- in the architecture, and supports the seman-
PASS, Intute, School, and Public Resources tically-enabled services that the Organic.
ones). Edunet Web portal will offer, by reasoning
• Learning Resource Exchange Module: upon a number of integrated ontologies.
concerns the connection of the Organic. • Web Portal Module: refers to the end-user
Edunet federation with other federations of visible parts of the whole infrastructure,
learning repositories, using open standards allowing users (including school teachers
and specifications for the exchange of search and pupils, university teachers and students,

Figure 2. Overview of the learning repository management module components


Learning content
Management Module

scAM repositories

scAM
repositories

repositories
tools
SH AM E Editor
+
C onfolio

content Providers

764
Architecture of the Organic.Edunet Web Portal

researchers etc.) to search, locate, retrieve the associated metadata into a learning repository
and access learning resources on OA and that is called SCAM (Standardized Contextualized
AE throughout the whole Organic.Edunet Access to Metadata), an Open Source Semantic
federation. Web repository solution for learning resources
(Palmér et al., 2004). Figure 3 presents how a
Each module is further detailed in the para- SCAM repository is accessed, using a combina-
graphs that follow. tion of technologies.
The repository backend is resource-oriented
Learning Repository Management and will store its metadata according to a Resource
Module Description Framework (RDF, www.w3.org/
RDF/) representation of the Organic.Edunet
This module deals with the way content producers IEEE LOM application profile. The repository
organize, annotate and publish learning resources provides a range of connection interfaces, al-
and metadata in an Organic.Edunet repository. lowing the most appropriate to be chosen for
As illustrated in Figure 2, each of the Organic. each situation. An interface which exposes the
Edunet content providers is expected to collect repository closest to the internal representation
and annotate its learning resources, according is the REST (Representational State Transfer),
to a multilingual application profile of the IEEE a resource-based software architecture building
Learning Object Metadata (LOM) standard fully on top of well established standards such as
(LTSC, 2002). Two existing software tools are the HTTP protocol (Fielding, 2000). This makes
being adapted and integrated for this purpose: it very easy to build interactive web applications
on top of this interface.
• A configurable metadata editor built upon The Confolio repository front-end builds on
the code-library SHAME (available as top of the REST-based web services exposed
Open Source at http://shame.sourceforge. by the repository and an AJAX (Asynchronous
net). With this code-library application JavaScript and XML) toolkit, which enables
programmers can develop flexible and eas- cross-browser compatibility and operating system
ily extensible annotation tools for Semantic independent application. The basic operations of
Web-based metadata. SHAME implements Confolio can be separated in two groups: adminis-
the Annotation Profile Model (Palmér et al., trative (e.g. creation of new portfolios of learning
2007a; Palmér et al., 2007b). This model is resources) and end-user (e.g. creation of folders
a configuration mechanism for the annota- and description of resources using the SHAME
tion of metadata and leaves the question metadata editor).
of metadata standard compliance up to a Using the Learning Repository Management
metadata expert and not to the application Module, Organic.Edunet content providers may
developer. collect resources, annotate them using metadata
• The electronic portfolio system Confolio conforming to the developed application profile,
(http://www.confolio.org) that allows the reviewing and approving resources, and then
flexible management of folder-based reposi- releasing resources for publication. Then, the
tory interfaces. metadata of the resources stored in a particular
Organic.Edunet repository are (a) made available
The content providers will use the integration for harvesting from the Semantic Services Mod-
of the SHAME editor and the Confolio tool in ule and (b) made available for harvesting and/or
order to upload (if desired) their resources and search federation to external federations.

765
Architecture of the Organic.Edunet Web Portal

Figure 3. Overview of technology layers in a and RDF Query Language - a W3C standardized
SCAM repository query language and protocol for accessing RDF
data (http://www.w3.org/TR/rdf-sparql-query/).
scAM repository

Struts2/ Spring/ REST Semantic Services Module

DAO The Semantic Services Module is the core engine


behind the Organic.Edunet Web portal that allows
JENA/ NG4J offering users with semantic search capabilities.
To support this, it is based on a semantic represen-
tation of the learning resources’ metadata, as well
Learning Object as a number of ontologies that are engaged during
repository search queries to provide reasoning capabilities.
More specifically, metadata is transformed into
an ontological representation inside a sub-module
called LOMR.
The LOMR (standing for Learning Object
Learning Resources Exchange Metadata Repository) is not itself a metadata
Module repository but rather a framework which provides
Web Service interfaces to any given, “real” learn-
The Learning Resources Exchange Module allows ing object repository. LOMR instances allow
for the communication of the Organic.Edunet developers to select the best repository imple-
repositories with external federations. Organic. mentation for a given application need, enabling
Edunet will aim at the connection with the LRE specialized components, such as custom query
and ARIADNE federations by adopting two resolvers and result composers, to benefit from
widely used protocols and specifications: the availability of different, heterogeneous LOMR
instances. LOMR main features include the stor-
• For communicating with ARIADNE: the age of learning object metadata in semantic format,
Open Archives Initiative Protocol for Meta- the provision of a service-oriented interface and
data Harvesting (OAI-PHM, http://www. the import of metadata in non-semantic formats,
openarchives.org/OAI/ openarchivesproto- among others.
col.html) for making metadata available for
harvesting from the ARIADNE services. In addition, LOMR offers semantic services to
• For communicating with LRE: the Simple the Web Portal Module, following WSMO. It uses
Query Interface (SQI, http://www.prolearn- the Web Services Modeling Language (WSML,
project.org/lori) for serving/exchanging http://www.wsmo.org/wsml/) in order to provide
queries with the LRE services. formal syntax and semantics for WSMO, since it
is richer in reasoning capabilities than the OWL
Metadata is transformed from its RDF repre- Web ontology language (http://www.w3.org/
sentation into an XML representation, in order TR/owl-features/) recommended by the W3C.
to be available for the external federations. Ad- Interoperability can be easily achieved through
ditional possibilities also exist for further inter- translating WSML to OWL through open source
connecting the Organic.Edunet repositories, due tools that are publicly available, such as the Web
to their SCAM basis, e.g. the SPARQL Protocol Service Modeling Toolkit (WSMT, http://wsmt.

766
Architecture of the Organic.Edunet Web Portal

Figure 4. Illustration of the way the Learning Resources Exchange Module operates
Organic.Edunet’s
Learning repositories

bioAgro

ENOAt content
Providers
Ecologica Learning repositories Management Module

Intute

Public
resources

school
Learning content Learning content
Exchange Module Exchange Module

XM L XM L
T ransform ation T ransform ation

OAI -PM H SQI API

Ariadne
Federation LrE
Federation

sourceforge.net) and WSMO Studio (WSMO4J, languages that have been indicated as related to
http://wsmo4j.sourceforge.net). the particular region, even if this has not been
As a starting point, three ontologies are indicated in the initial search query. The third
expected to be used by the Semantic Services ontology will be representing IEEE LOM. It is
Module. The first ontology will represent the expected to allow reasoning related to semantics
domain area (OA and AE). It will serve all sub- of the LOM structure itself, such as searching for
ject classification purposes, as well as allow for information in other elements than the ones that
reasoning related to the semantics of the OA and a user has initially indicated.
AE concepts themselves. For example, searching In LOMR, metadata will be harvested from
for resources that have been classified using some the individual Organic.Edunet repositories us-
concepts or terms related to the ones that the user ing an appropriate harvesting mechanism. As
has initially indicated. The popular AGROVOC Figure 5 shows, the RDF representations stored
(http://www.fao.org/aims/ag_intro.htm) ontology in the SCAM repositories will be converted to
of FAO will be used as a basis for the construction the WSML representation that LOMR requires.
of this ontology. The second ontology will be a Once all the metadata information is stored in the
geographical one. It will help reasoning related LOMR repository in the formal, ontology-based
to the geographical origin and/or coverage of format, LOMR will be able to expose various
resources and their associated languages. For functionalities through semantic Web Services
instance, it may allow users from a particular (described according to WSMO), allowing a wide
geographical region to search for resources in variety of interactions with the Organic.Edunet
Web portal.

767
Architecture of the Organic.Edunet Web Portal

Figure 5. Overview of the Semantic services module


Organic.Edunet Web Portal Module

semantic services Module

W eb Service Query
Query F ram ew ork
Interface

uses
W SM L L OM R

Ontologies
LOMr Module

LOM2WsML
LOM 2W SM L translator
C onverter Module

H arverst
M echanism

Learning repository Management Module

Web Portal Module Discussion

The final module of the Organic.Edunet architec- Benefits


ture is the Web Portal one. It actually comprises
the online environment that will interact with As it has been described in the presentation of
the various user roles (school teachers & pupils, the Organic.Edunet portal architecture, there is
university teachers and students). For this purpose, a number of benefits expected from the adoption
it entails a role-filtering mechanism that will of Semantic Web technologies. The following
allow each user category to be presented with a paragraphs will go into more detail on these
user interface tailored to its specific needs. Apart benefits.
from allowing users to semantically search and The use of ontologies for the classification
retrieve learning resources using the Semantic of learning resources will allow the refinement
Services Module, the Web Portal Module will also and expansion of queries. Users currently have
provide the users with the option of evaluating/ to rely on keyword-based searches: for instance,
rating learning resources. Multi-dimensional nu- a teacher looking for learning resources on the
merical evaluations will be stored in appropriately advantages of the use of organic fertilizers might
defined evaluative metadata (Vuorikari et al., try something like “advantages organic fertilizers”
2008). Then, they will give input to a collabora- in Google (or in any other keyword-based search
tive filtering mechanism that will recommend engine). A search on these keywords would return
users to look at resources that other users with results containing either the terms “organic” or
similar preferences liked in the past (Manouselis “fertilizer” or “advantage” or a combination of
& Costopoulou, 2007). them, but many of those would be seen as non ap-

768
Architecture of the Organic.Edunet Web Portal

Figure 6. Elements of the Web portal module

U niv School Users


Students Pupils
T eachers T eachers

Web Portal Module


role Filtering Mechanism
collaborative Filtering Module

C ollabor ative
Web POrtAL F ilter ing Eval LR
M echanism

semantic services Module

propriate for most users. An example on the kind resources explaining the advantages of the use of
of (inadequate) resources that might be retrieved organic fertilizers and not criticizing them, and
with this method –traditional keyword-based would allow users avoid suggestions for learning
searches– would be the following: resources on other topics.
In addition, the use of ontologies will further
• Commercial information on products by enhance the search and browsing services of-
companies selling organic fertilizers. fered to the users. More specifically, users will
• Resources on the advantages of non-organic be able to browse through learning resources by
fertilizers (matching the three keywords used selecting concepts of the ontologies used, together
for the search). with an expression of their relationships. In a
• Resources criticizing the use of organic learning objects portal on organic farming, these
fertilizers and discouraging users on its technologies would help to easily access similar
application due to their low efficiency and materials to a given one, as the relationships in
high prices. the ontology would provide the ability to navigate
• Resources on the elaboration of organic from one instance to another. An example would
yoghurt (as they would match to at least one be a search on learning objects about organic pest
of the keywords provided). control, which would return e.g. a case study on
the use of several types of insecticide-fungicide
Contrary to these examples, the use of ontolo- dust for use on fruit trees. Thus, portal users could
gies for the description of materials would force navigate from the relationship from this object to
the search engine to stick to strict-matching fruit trees, and find e.g. learning objects on the
criteria to unambiguous definitions. It would commercialization of organic apples, or even
also allow to search only educational-oriented lectures on organic fungicides applicable only
materials explicitly annotated with the predicates to specific geographical regions.
such as “IsAbout” or “Provides BackgroundOn”, Learning resources in current public reposi-
which are related to e.g. organic fertilizers. This tories often have a high variability in their char-
would even allow users to find just those learning acterizations: from anything in digital format

769
Architecture of the Organic.Edunet Web Portal

to well-defined educational oriented learning (2) check if the mapping is consistent with
materials including metadata conformant to the the rest of the subsumers inside the upper
IEEE LOM standard (McGreal, 2004). The de- ontology, (3) provide appropriate predicates
scription of all the knowledge about the domain to characterize the new category, and (4) edit
of learning objects in the form of an ontology, and it in an ontology editor to come up with the
the use of this ontology as the basis for a learning final formal version.
object portal on organic farming, would provide • Even though semantic web technologies are
the portal with the flexibility necessary to seam- attractive and promise many benefits, they
lessly accommodate different conceptualizations. are not, unfortunately, ready for production
It would also provide the ability to interact with use yet. Ontology-management systems
external systems, even if each of these systems hardly support the large ontologies needed
have a different understanding of what a learning for most production environments, and thus
object is, how their metadata should look like, should be preferably used for research and
etc. This model would eventually provide the experimentation purposes. A good example
users with a number of different functionalities, of the lack of maturity of these technologies
adapted to each particular concept of learning are the APIs for ontology persistence in Java
object, and not necessarily restricted by only one available today: Jena, Sesame and Protégé’s
of these conceptualizations, applying technologies persistence APIs find many difficulties in
already in practice (Soto et al., 2007). managing medium to large sized ontolo-
gies.
Challenges • Although the Organic.Edunet portal is based
on a distributed architecture, the semantic
Apart from the benefits, a number of challenges module calls for a centralized repository
have also to be dealt with during the implementa- that harvests the data from all repositories
tion of the Organic.Edunet architecture: in the federation. Even though the harvest-
ing tests carried out so far have shown good
• The process of selecting, developing, and results, scalability, size and performance
specifying the ontologies to be used (espe- issues might arise as the project progresses
cially as far as domain-dependent ones such and have an impact in the development.
as the OA & AE ontology are concerned) is
demanding and time-consuming, and needs
the help of a number of experts from different Conclusion
disciplines. To make all the experts reach
agreements is not always straight-forward To further illustrate the potential of Semantic
(Sánchez-Alonso et al., 2008.). Web technologies in Web portal applications,
• The process of engineering a new ontology experiences from large-scale implementations are
often implies checking the new knowledge required. Especially in the case of Web portals
against the commonsense knowledge and that provide access to learning resources, the
general terms in an upper ontology. This pro- implementation of Web portals with services
cess, which has to be carefully carried out, that are based on Semantic Web technologies
can be summarized in four iterative phases that will be tested semantically annotating large
as described by Sánchez-Alonso & García collections of learning resources, and by being
(2006): (1) find one or several terms that accessed and used from communities of users
subsume the category under consideration, with numerous members. In this direction, this

770
Architecture of the Organic.Edunet Web Portal

article presented a large-scale implementation Conceicao S., Sherry L., Gibson D., Amenta-Shin,
effort that engages Semantic Web in order to set G.,”Managing Digital Resources for an Urban
up a technical infrastructure that will facilitate Education Portal”. in Proc. of World Conference
sharing and reusing of learning resources for the on Educational Multimedia, Hypermedia and
agricultural domain. The next steps of this work Telecommunications (ED-MEDIA 2003), Ho-
concern reporting the results from the actual nolulu, Hawaii, June 23-28, 2003.
implementation, deployment, and initial testing of
Daraselia, N., Yuryev, A., Egorov, S., Novichkova,
the technologies that are integrated in the Organic.
S., Nikitin, A., & Mazo, I. (2004). Extracting
Edunet Web portal.
human protein interactions from MEDLINE us-
ing a full-sentence parser, 20, 604-611. Oxford
Univ Press.
Acknowledgment
DeSanctis G., Wright M., Jiang L., “Building A
The work presented in this article has been funded Global Learning Community”, Communications
with support by the European Commission, and of the ACM, 44 (12), 80 – 82, December 2001.
more specifically the project No ECP-2006-
EC (2004), European Action Plan for Organic
EDU-410012 “Organic.Edunet: A Multilingual
Food and Farming, COM 415, Brussels 10 June
Federation of Learning Repositories with Qual-
2004.
ity Content for the Awareness and Education of
European Youth about Organic Agriculture and Ethridge R.R., Hadden C.M., Smith M.P.,
Agroecology” of the eContentplus Programme. “Building a Personalized Education Portal: Get
a Behind-the-Scenes Look at LSU’s Award-
Winning System”, Educause Quarterly, 23(3),
References 12-19, 2000.
Fielding, R. T. (2000). Chapter 5: Representational
Aroyo, L., Denaux, R., Dimitrova, V., & Pye, M.
State Transfer (REST), Architectural Styles and
(2006). Interactive Ontology-Based User Knowl-
the Design of Network-based Software Architec-
edge Acquisition: A Case Study. Lecture notes in
tures, Dissertation.
computer science, 4011, 560.
Forte, E., Haenni, F., Warkentyne, K., Duval, E.,
Averweg U. (2007), Portal Technologies and
Cardinaels, K., Vervaet, E., et al. (1999). Semantic
Executive Information Systems Implementation,
and pedagogic interoperability mechanisms in
in A. Tatnall (Ed.), The Encyclopedia of Portal
the ARIADNE educational repository. SIGMOD
Technology and Applications, Hershey, PA: Idea
Rec., 28(1), 20-25.
Group Publishing.
Hatala M., Richards G., Eap T., Willms J., “The
Berners-Lee, T. (1998). Semantic Web Road Map.
interoperability of learning object repositories and
W3C Design Issues. URL http://www. w3. org/
services: standards, implementations and lessons
DesignIssues/Semantic. html, Oct.
learned”, in Proc. of the 13th International World
Boff E., Rizzon Santos E., Vicari R.M., “Social Wide Web Conference, New York, NY, USA,
Agents to Improve Collaboration on an Education- 19-27, 2004.
al Portal,” Sixth IEEE International Conference
Holden, C. (2003). From Local Challenges to a
on Advanced Learning Technologies (ICALT’06),
Global Community: Learning Repositories and
IEEE Computer Press, 896-900, 2006.

771
Architecture of the Organic.Edunet Web Portal

the Global Learning Repositories Summit. Version Moreale E., Vargas-Vera M., “Semantic Services
1.0, Academic ADL Co-Lab, November 11. in e-Learning: an Argumentation Case Study”,
Educational Technology & Society, 7 (4), 112-
Kotzinos D., Pediaditaki S., Apostolidis A., Atha-
128, 2004.
nasis N., Christophides V., “Online curriculum
on the semantic Web: the CSD-UoC portal for Neven F., Duval E., “Reusable learning objects: a
peer-to-peer e-learning”, in Proc. of the 14th survey of LOM-based repositories”, in Proc. of the
International Conference on World Wide Web, 10th ACM International Conference on Multime-
Chiba, Japan, 307 – 314, 2005. dia, Juan-les-Pins, France, 291 – 294, 2002.
LTSC (2002). IEEE Standard for Learning Object Palmér M, Enoksson F, Nilsson M, Naeve A
Metadata, 1484.12.1-2002. IEEE Learning Tech- (2007a). Annotation Profiles: Configuring forms
nology Standards Committee, 2002. to edit RDF. In: Proceedings of the Dublin Core
Metadata Conference. DCMI Conference Papers,
Luke R., Clement A., Terada R., Bortolussi D.,
United States.
Booth C., Brooks D., Christ D., “The promise and
perils of a participatory approach to developing Palmér, M., Enoksson, F., Naeve, A (2007b).
an open source community learning network”, LUISA deliverable 3.2: Annotation Profile Speci-
in Proc. of the 8th Conference on Participatory fication,.
Design on “Artful integration: interweaving me-
Palmér, M., Naeve, A., Paulsson, F., (2004), The
dia, materials and practices”, 11 – 19, Toronto,
SCAM-framework – helping applications to
Ontario, Canada, 2004.
store and access metadata on the semantic web,
Maedche, A., Staab, S., Stojanovic, N., Studer, Proceedings of the First European Semantic Web
R., & Sure, Y. (2001). SEAL - A framework for Symposium (ESWS 2004), Heraklion, Greece,
developing SEmantic Web PortALs. Lecture Notes May, 2004, Springer, ISBN 3-540-21999-4.
in Computer Science, 2097(1-22), 46.
Richards, G., McGreal, R., Hatala, M., & Friesen,
Mahadevan B., “Business Models for Internet N. (2002). The Evolution of Learning Object Re-
based E-Commerce: An Anatomy”, California pository Technologies: Portals for On-line Objects
Management Review, 42(4), 55-69, 2000. for Learning. Journal of Distance Education,
17(3), 67-79.
Manouselis N., Costopoulou C. (2007). Experi-
mental Analysis of Design Choices in Multi-Attri- Sánchez-Alonso S., Cáceres J., Holm A.S., Lieb-
bute Utility Collaborative Filtering, International lein G., Breland T.A., Mills R.A., Manouselis N.,
Journal of Pattern Recognition and Artificial “Engineering an ontology on organic agriculture
Intelligence (IJPRAI), Special Issue on Person- and agroecology: the case of the Organic.Edunet
alization Techniques for Recommender Systems project”, in Proc. of the World Conference on
and Intelligent User Interfaces, 21(2), 311-331. Agricultural Information and IT (IAALD AFITA
WCCA 2008), Tokyo, Japan, 24 - 27 August,
McGreal, R. (2004). Learning Objects: A Practical
2008.
Definition. International Journal of Instructional
Technology and Distance Learning, 1(9), 21-32. Sánchez-Alonso, S. and García, E. (2006) Making
use of upper ontologies to foster interoperability
Moraga Α., Calero C., Piattini Μ., “Comparing
between SKOS concept schemes. Online Infor-
different quality models for portals”, Online In-
mation Review, 30(3), pp. 263-277.
formation Review, 30 (5), 555-568, 2006.

772
Architecture of the Organic.Edunet Web Portal

Sancho, P., Martínez, I., & Fernández-Manjón, B. Tatnall, A., Ed. (2005a). Web Portals: the New
(2005). Semantic Web Technologies Applied to Gateways to Internet Information and Services.
e-learning Personalization in< e-aula>. Journal of Hershey, PA, Idea Group Publishing.
Universal Computer Science, 11(9), 1470-1481.
Tatnall, A., Ed. (2007). Encyclopaedia of Portal
Santos, E. R., Boff, E., & Vicari, R. M. (2006). Technology and Applications. Hershey, PA, In-
Semantic web technologies applied to interop- formation Science Reference.
erability on an educational portal. In Proc. of
Verdejo M.F., Barros B., Mayorga J.I., Read T.,
Intelligent Tutoring Systems, 8th International
“Designing a Semantic Portal for Collaborative
Conference, 4053, 308-317.
Learning Communities”, in Selected Papers from
Shadbolt N., Berners-Lee T., Hall W. (2006). the 10th Conference of the Spanish Association
The Semantic Web Revisited, IEEE Intelligent for Artificial Intelligence (CAEPIA03), Current
Systems, 21(3), 96-101, May/June. Topics in Artificial Intelligence, LNCS 3040,
Springer Berlin/Heidelberg, 251-259, 2004.
Sicilia M. A, García E. Sánchez-Alonso S, and
Soto J. (2005). A semantic lifecycle approach to Vuorikari R., Manouselis N., Duval E. (2008).
learning object repositories. In Proceedings of Using Metadata for Storing, Sharing, and Reus-
ELETE 2005 - eLearning on Telecommunications, ing Evaluations in Social Recommendation: the
Lisbon, Portugal. Case of Learning Resources, in Go D.H. & Foo
S. (Eds.) Social Information Retrieval Systems:
Soto, J., García, E. and Sánchez-Alonso, S.
Emerging Technologies and Applications for
(2007). Semantic learning object repositories.
Searching the Web Effectively, Hershey, PA: Idea
International Journal of Continuing Engineer-
Group Publishing.
ing Education and Life-Long Learning, 17(6),
pp. 432-446. Warner, S. (1999). Internet portals, what are they
and how to build a niche Internet portal to enhance
Stokes E., Edge A., West A. (2001). Environmental
the delivery of information services. In Proceed-
education in the educational systems of the Euro-
ings of 8th Asian-Pasific SHLL Conference.
pean Union, Environment Directorate-General,
European Commission, April. Winkler, R. (2001). Portals – The All-In-One Web
Supersites: Features, Functions, Definition, Tax-
Tane J., Schmitz C., Stumme G., “Semantic re-
onomy. SAP Design Guild, Edition 3. Retrieved
source management for the web: an e-learning
May 2, 2008, http://www.sapdesignguild.org/
application”, in Proc. of the 13th International
editions/edition3/portal_definition.asp.
World Wide Web Conference, New York, NY,
USA, 1-10, 2004. Woukeu A., Wills G., Conole G., Carr L., Kampa
S., Hall W., “Ontological Hypermedia in Educa-
Tatnall, A. (2005b), “Portals, Portals Every-
tion: A framework for building web-based edu-
where...”, in Tatnall, A. (Ed.), Web Portals: the New
cational portals”, in Proc. of World Conference
Gateways to Internet Information and Services,
on Educational Multimedia, Hypermedia and
Hershey, PA, Idea Group Publishing, 1-14.
Telecommunications (ED-MEDIA 2003), H
Honolulu, Hawaii, June 23-28, 2003.

This work was previously published in International Journal of Web Portals, Vol. 1, Issue 1, edited by J. Polgar, pp. 71-91,
copyright 2009 by IGI Publishing (an imprint of IGI Global).

773
774

Chapter 3.7
Interactive Whiteboards in
the Web 2.0 Classroom
David Miller
Keele University, UK

Derek Glover
Keele University, UK

Abstract questioning techniques, and increasingly, by


consideration of the use of gestures at the
IWB. While IWBs are not a solution to all
This chapter summarizes the work underway learning problems, it is suggested that they
to chart, critically evaluate, and systematize offers scope for greater student involvement
the introduction of interactive whiteboards and understanding in the learning process.
(IWB) into modern foreign language classrooms
in England. It is suggested that there is a
developmental cycle whereby teachers take Introduction
some time to understand the technology
and become competent in its use. They then The interactive whiteboard (IWB) is part of the
look to its advantages in presentation and growing variety of equipment used in conjunction
the motivation of students before becoming with a computer and data projector to incorporate
aware of its pedagogical value and develop software, Internet links and data equipment for
a changed classroom practice. This cycle is whole class use. Increasingly schools are equipping
based upon enhanced teacher understanding each subject area, and in many cases every class-
of the nature of interactivity and the potential room, with an interactive whiteboard to supplement
offered by the IWB in meeting a variety of or replace traditional white or blackboards. This is
learning needs. The relationship between IWB happening in many parts of the world, for example
use and Web 2.0 arises from the potential of in Mexico there has been a focus on IWB installation
both to add impetus for teachers to structure and use, wherever possible, to ensure that the full
lesson development and enhance activity. potential of the equipment and associated software
It is supported by teacher understanding of can underpin quality lessons to be taught on the
widest possible scale. This shows a fundamental
DOI: 10.4018/978-1-60566-190-2.ch027 belief that IWB technology and pedagogy can make

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Interactive Whiteboards in the Web 2.0 Classroom

a difference across a range of subjects (Hennessey, The capacity to use the equipment in this way is
Wishart & Whitelock, 2007; Belli, 2005; McFar- dependent upon both technical fluency in the use
lane, 2005). Research shows that this may be true of the equipment and associated software, and
for certain young people and for a period of time pedagogic understanding and flexibility to exploit
but that fundamental changes promoting contin- the possibility of interactivity between teacher
ued educational achievement are only possible and student, and student and student. To achieve
where teachers recognize the significance of the this has much in common with the educational
word “interactive” and develop their approaches development of all ICT and reflects a move,
to teaching to promote this. Such approaches are whether recognized or not, to the use of the Web
concerned with driving student involvement and 2.0 platform (Belshaw, 2007). Web 2.0 is here
increasing understanding. They are based on the understood to be related to a focus on learning
recognition of students’ differing learning needs through concentration on multimedia use, age and
in order to ensure conceptual understanding and ability linked group and individualized learning,
cognitive development (Armstrong et al., 2005; and an awareness of variations in personal learning
Hall & Higgins, 2005; Kent, 2006; Smith et al., styles (Xhakli, 2008). This brings with it a change
2005; Sturcke, 2004; Jones, 2004). of emphasis from the teacher centered transmis-
Glover and Miller (2003) have traced the pat- sive approach to learning to one characterized by
tern of increasing use in terms of the influence interactivity, collaboration, user-generated content
of “missioners, tentatives and luddites” within and immediacy of feedback. This is based on short
schools. More importantly they have demonstrated attention switches from the teacher to the IWB as
that teachers need to be helped through a three- a mediating agency allowing access to other ICT
stage development process so that they can move technology within the classroom.
from traditional to increasingly more interactive In a sense the IWB presents a new meta-lan-
approaches, specified as: guage for classroom use. It certainly has developed
its own vocabulary, which offers new technical
a. Supported didactic, where the teacher makes terms. These become part of the basic language
some use of the IWB but only as a visual from initial training sessions with phrases such
support to the lesson and not as integral to as “calibration,” “drag and drop,” and “hide and
conceptual development. reveal” being early concepts for the user to un-
b. Interactive, where the teacher makes some derstand. With the use of the interactive potential,
use of the potential of the IWB to stimulate phrases such as “virtual manipulatives” (Weiss,
student responses from time to time in the 2005) signify understanding of both process and
lesson and to demonstrate some concepts. pedagogic possibility, and as the integration of
c. Enhanced interactivity, where the teacher technology and pedagogy becomes better under-
develops the materials so that the students stood teachers and learners become aware of as-
focus upon the IWB as a means of prompting, sociated words from subject specific areas such as
explaining, developing and testing concepts “the use of artifacts,” which in both mathematics
for most of the lesson. and modern languages has its own significance
within the IWB focused classroom.
It is only at the third stage that the potential Language, however, is more than vocabulary,
of the board as the focus of learning based upon and IWB users become aware of the use of intona-
a new understanding of the learning process, is tion, whereby the same word or phrase used in a
recognized and realized by the teacher (Miller & different way signifies another meaning. This can
Glover, 2004; Ziolkowski, 2004; Watson, 2006). be illustrated by considering the word “interac-

775
Interactive Whiteboards in the Web 2.0 Classroom

tive” which is seen to operate at three levels — as al., (2007) examine the integration of the IWB
indicating that there is a relationship between with teaching in the lower secondary school. Both
the technology and the user whereby a physical are however, reporting on the need to move from
action leads to changes in the visual content on didactic to more interactive approaches.
the boards; as an instruction to a user when us- Our illustrations are taken from research into
ing the board; or to one planning the sequence of the learning of modern foreign languages within
conceptual developments and seeking a process ten schools in England in 2004. These schools
by which movement on the board can lead to ac- were all at an early stage of technology use and
tion in the brain of the recipient and subsequent the experience of teachers in these schools accords
action in the classroom. with that of all new learners in that they have had to
This argument suggests that teaching and gain both competence and confidence in working
learning is limited by what occurs on the desk with technology in enhancing pedagogy.
or the board but there is an intermediary in the
process. This is the teacher, however defined; who
acts as a mediator in the process of learning and IWB and modern
who, we have noticed, develops a set of gestures language teaching
as the non-verbal aspect of language. While not
all users have the same hand and face gestures The selected schools were known to have previ-
for similar aspects of mediation, research shows ously good OfSTED (national inspection service)
that users make consistent use of the same non- reports and were therefore likely to be showing
verbal expressions as lessons proceed (Miller & good practice. Overall 13 lessons were video-
Glover, 2006). recorded for subsequent analysis according to the
This chapter concentrates on the outcomes following framework:
of research that has been centered on the way in
which IWB users, both teachers and learners, have • The timeline and activity sequence in
developed their use of hardware and software to each lesson. This usually included a revi-
enhance teaching and learning in modern foreign sion starter, and then moved through vo-
language teaching. Our work was based on recent cabulary use to sentence construction and
research and practice publications that highlighted grammatical understanding.
the way in which IWBs could be a support in target • Classroom management issues. These in-
language teaching. Research has also highlighted cluded the way in which the room was set
the role of ICT in language teaching and directed out for the lesson, the nature of the envi-
teachers to the use of the Internet, streaming videos ronment to favor or inhibit IWB use for all
and downloaded resources as a stimulant to inter- students in the room, the integration of the
est in the classroom. Interactivity, however, is a IWB with traditional textbooks and other
feature of the Web 2.0 philosophy and this may resources and the use of student groupings
extend beyond the classroom to include e-mail for learning activities.
correspondence, blogging, and the use of realia. • Enhancement from IWB use was sought
These are shown in the developing shareware within a framework of revision of past
from the Teacher Resource Exchange in England work, establishing new principles and data,
(tre.ngfl.gov.uk/server.php). As yet, though, there sequencing of information and learning, as
is very little modern language and IWB specific well as the demonstration of processes and
research. Glover et al., (2007) deal with the re- reinforcement of learning through recall
search outlined below in more detail and Gray et and the use of examples.

776
Interactive Whiteboards in the Web 2.0 Classroom

• The contribution of IWB use to cognitive Presentation


development was assessed through the es-
tablishment of aims, the use of varied learn- During the lesson observation notes were made
ing styles, stepped learning sequences with of the techniques used in the presentation of
revision as needed, problem solving, and materials. In some lessons teachers used several
recall and discussion as a bridge to further techniques, in others they used just two or three
learning. but exploited them to the full as a further spur to
• The contribution of IWB use to the con- learning. Overall, the frequency of use was as
ceptual development of discrete elements shown in Table 1.
in the lesson through the identification of Observation and interview evidence was also
processes, manipulation of data, and review used to explore the processes by which IWB use
to ensure understanding and application as promoted interactivity as understood by teachers in
part of cognitive development. the lessons. In using techniques the four most com-
• The nature of IWB techniques used within mon methods of securing interactivity were:
the lesson and the way in which these are
perceived by students. • Drag and drop, matching a response to a
• An assessment of the teaching style used in stimulant.
the lesson. • Hide and reveal, opening a hidden response
• Identification of practical and pedagogical when the stimulant was understood.
issues arising from the use of IWB technol- • Matching equivalent terms, e.g. vocabu-
ogy in its contribution to effective learning. lary in different languages.
• Measurement of the percentage of the les- • Movement, to demonstrate principles, e.g.
son when the IWB was the focus of teach- sentence construction.
ing and learning.
Students were also observed writing (and
Structured interviews were also undertaken with replacing) words, e.g. as they explained a story
ten teachers to probe aspects of their understanding in a village mapped from the IWB, and shading
of presentational, motivational and pedagogical is- e.g. to show rooms in a house where one would
sues inherent in technology use. The interviewers watch TV. In all of these the aim of the teachers
attempted to identify the reasons why, and how, was to: “have a number of children working at
teachers felt that the IWB made a difference to the board so that they could gain competence and
learning. Two groups of ten students each were in- confidence and to get others involved especially
terviewed in two schools to gain some triangulation where we were using competitive approaches to
with teacher opinion. There appears to be a run-in keep them all involved” (Male teacher, Spanish
period of between eighteen months and two years lesson).
while teachers develop competence in handling the Teachers made use of superimposition by mov-
technology, in developing fluency in its use and in ing phrases or words and putting them alongside
establishing a battery of basic screens to support vocabulary or in sentences, and considerable use
their teaching. Whilst teachers may have developed of matched verbal and visual representation of vo-
these skills their practice could still be grounded cabulary. They made use of the coloring potential
in older styles of teaching — or these styles may for parts of speech, and shading, to mark parts of
emerge in some lessons but not in others according a sentence as construction developed.
to the needs of the topic and the class context. Nine of the teachers made some comment about
higher standards of presentation as a result of the

777
Interactive Whiteboards in the Web 2.0 Classroom

Table 1. Use of techniques in IWB focused teaching

Techniques Example No of lessons (n=13)


Movement and animation Cycle route on map 11
Drag and drop Vocabulary 10
Overwriting of screen Verb endings 10
Verbal and visual linkage Sounds and objects 10
Superimposition Labeling 10
Hide and reveal Sentence construction 10
Shading Comprehension 5
Imported sound Clip 4
Gap infilling Sentence construction 4
Internet access Life in village 4
Highlighting Parts of speech 4
Automatic responses Vocabulary 4
Applet development Describing actions 3
Tools Connecting lines 2

use of IWB software and in each of the student taching sounds. This incorporation of sound was
groups there were three references to the way in a feature of half the lessons observed in modern
which writing on the board had improved. Ac- languages. Four of the 13 lessons also made use
cording to teachers, it was “sort of professional of passages from the Internet as the basis of a
looking” and “much easier to read than the writing comprehension activity and in two lessons stu-
we used to have.” dents working with laptops were asked to pursue
At the same time both teachers and students this at a higher level while the others in the class
spoke of the problems of “over-writing” where worked at the IWB.
teachers made notes on diagrams on the board, Observation suggests that the use of the presen-
and where “the writing looks odd, sort of angular.” tational aspects of the IWB varies as students get
This is partly due to the level of fluency developed older. Year seven students (aged 11-12) showed
in the use of the pen on the IWB, but also related enthusiasm and interest when filling in missing
to the precision generated by the software. words in a competitive situation — the capacity for
Although increasing, at the time there were few the IWB to have associated sounds for success and
commercial or professional programs specifically failure added to this. By year nine (aged 13-14),
designed for teaching modern languages using however, it appears that students are less willing
the IWB. As a result practitioners speak of the to participate in either volunteering to write on
need to develop their own materials often through the board unless all students are involved, or as a
electronically scanning textbooks, or from down- member of a small group at the board; completing
loaded Internet material. In three of the lessons “hide and reveal” type statements or hazarded
characterized as “supported didactic” a page of answers, and demonstrating verbal relationships
sketches had been scanned from a textbook and to the rest of the class. Indeed, there is some
this lacked the movement, color and vitality of evidence that by this stage students will attempt
comparable material built up by the teacher from to subvert some of the presentational advantages
clipart collections, but given interest through at- through spotting wrong results so that they incur

778
Interactive Whiteboards in the Web 2.0 Classroom

the “noise of failure,” or give the wrong answer verb endings. Over half the lessons observed made
to “appear to be one of the gang.” some use of associated sound, imported picto-
One skilled teacher pointed out, however, that rial material and “real” newspaper or magazine
this does not mean that students have outgrown the extracts as a basis for comprehension work and
board. Rather they expect the teacher to be fluent the application of vocabulary. It was agreed that
in its use and to lead their learning in such a way this was the greatest presentational advantage in
that their consolidation takes place individually that pre-prepared materials could be highlighted,
in their exercise books following teacher use of expanded, developed and analyzed by over-written
IWB materials. Discussion showed that even to comment. In discussion respondents also con-
age 16 students appreciate its value when the IWB sidered the issues of “savability.” All except two
is a source of further material for comprehension, participants had a battery of screens that they used
or when it is used to demonstrate grammatical as they prepared their lessons. The general view
rules in action. was that although it took time to prepare lessons
Consideration of the content and approach for IWB use they could then be stored and used
of the observed lessons indicates that the more in three ways:
didactic teaching was in lessons where there
were fewer activities in the lesson period, where 1. Catalogued by topic and then drawn out as
the pace was more limited and where there were each lesson was prepared.
longer periods of textbook or exercise work. In 2. Catalogued by lesson and then copied if
these lessons there were also fewer techniques the same screen was to be used in another
used and teachers tended to make use of “drag and lesson.
drop” or “hide and reveal” more than in lessons 3. Catalogued by intended year group and then
that used movement, automation (manipulatives) developed with further material if being used
and color changes. In the lessons characterized in a different context.
by enhanced interactivity there was a tendency to
use more activities with several techniques and Teachers were less ready to regularly link their
a combination of commercially or professionally presentation to the printer so that materials could
produced materials with those developed by the be made available for students. In 9 of the 13 ob-
teacher. These lessons had greater pace and tended served lessons there was an element of copying
to use the IWB as the focus of all activity includ- from the IWB at some stage in the lesson. Table
ing board-based exercises and extension work. 2 shows the results of an analysis of the copying
A year nine group learning German followed a used in observed lessons:
three minute revision starter with three activities Some copied activity characterizes all the
building vocabulary through highlighting, drag teaching described as “modified didacticism,”
and drop and hide and reveal; building phrases but also occurs in the other styles of IWB use.
through pair work drawing upon matching of It seems that teachers are less willing to explore
vocabulary, gender and translation, to sentence or use the copying facility than is claimed by the
construction based on an Internet activity. The promoters of IWB technology. The more positive
lesson concluded with revisiting screens and the view emerges from a linguist who commented on
use of color highlighting to identify rules for case the time saved by being able to print off materials
and gender agreement. for those needing extra help.
Teachers commented on, and used, color
highlighting and arrows to indicate movement
and positioning for parts of speech and to indicate

779
Interactive Whiteboards in the Web 2.0 Classroom

Table 2. Analysis of copying activity during observed lessons (multiple activities possible)

Nature of copying activity Number of language lessons (n=9)


Examples for exercises 6
Rules of grammar or process 4
Copied screen as a record 2
Aims of the lesson 1
Homework material 3

Motivation The observed lessons show, however, that older


and more able students gain from the IWB because
In all the discussions with teachers it has been dif- they appreciate the visualization of structures
ficult to sort out the motivational factors from the more readily than through verbally dominated
presentational or pedagogic in the successful use approaches. A German lesson for 15-16 year olds
of the IWB. Seven of the teachers made reference exploited the IWB to build up and then analyze
to the intuitive use of the technology as a feature sentences in terms of constituent vocabulary,
in the everyday lives of students and felt that the constructional frameworks and comprehension
schools should be offering a high level of presen- alongside continuous and enthusiastic encourage-
tation and attractiveness so that “what happens in ment from the member of staff who constantly
school should not be seen as a poor relation to what referred them back to earlier screens. It was not
they see on TV and computer screens.” simply the IWB, but also the way in which it had
Our evidence suggests that the major features become integrated into the teaching method in
that encourage student motivation are as fol- a highly personal way combining visualization
lows: and encouragement of all students, that enhanced
learning.
• The intrinsic stimulation provided by the Another factor in the motivation of students
combination of the visual, kinesthetic and stemmed from the way in which teachers exploited
auditory paths to learning. a “different type of contact with the lesson in
• Those aspects of classroom management the student’s hands.” Good practice obviously
that lead to a focus on the IWB with linked builds upon knowledge of particular groups and
desk activities throughout the lesson. of individuals within the groups and a realistic
• The stepped learning that characterizes assessment was that “the IWB still doesn’t mean
much IWB teaching offering constant chal- that we shall have a lesson where all the students
lenges with frequent assessment of achieve- are paying attention all the time.” Boys, for ex-
ment as a stimulant to further involvement. ample, are generally more ready to demonstrate or
• The particular advantages for slower learn- complete work at the IWB than girls of the same
ing students or those who need reinforce- age. Older boys were more ready to demonstrate
ment through the presentation of data or in part because it provides an opportunity for
processes with more than one learning style them to show their superiority in technological
(i.e. the ability of the board to allow mate- fields when teachers comment upon inadequacies
rial to be presented or represented in a va- of programs or available tools, while girls were
riety of ways. more concerned about “being right” before they
would commit themselves to the board. Evidence

780
Interactive Whiteboards in the Web 2.0 Classroom

from the two student groups showed that they • The availability of games that support
thought that “lessons had less wasted time” and learning, require responses that can be im-
that “they moved with more pace so that they mediately assessed and then linked to a
didn’t want them to come to an end.” If there scoring system with team races or noughts
is one single motivational factor during lessons and crosses.
it appears to be that the immediacy of response • The “fun” arising from the use of pictorial
ensures maintained interest. Seven of the teachers matter and the immediacy of any process-
refer to the enhanced engagement in lessons and ing built into the programs.
four referred to the ways in which the use of the • The opportunity to revisit earlier con-
IWB encouraged participation. cepts and examples in underpinning
Although there was general agreement that understanding.
teachers needed to consider aspects of lighting,
student seating arrangements, sight lines, and the Where lessons have such a dynamism and at-
area of the board in use by students considering traction it is likely that they will offer interest and
their physical characteristics, the observed les- challenge. This supports both revision of earlier
sons highlighted continuing issues. In four of the work and enhanced understanding of new work.
13 lessons tables were organized in such a way Above all as one teacher commented this offers
that students were in rows at right angles to the “credible media for a new age.” Teachers were
board, or at grouped tables where half of the stu- conscious, however, of the time demands for prepa-
dents naturally had their backs to the board. This ration even when using commercial materials, and
problem is not subject specific but is related to the four referred to the problems of technology that
size of the room, access problems and the need could inhibit slick use of the IWB.
for teachers to move around while desk work is These data show that those lessons charac-
in progress. In three classrooms light infiltration terized by enhanced interactivity focused on
rendered vision difficult for those seated at the the board for a greater proportion of the lesson,
near front of the sides of the rooms. Ameliora- while those where the board was a support for
tion was achieved in one school by using laptops more didactic approaches used the board for a
with the same screen program so that vision was significantly more limited period. For linguists
achieved and in another by breaking the lesson more of the lesson may have to take place away
up in such a way that board activity was distinct from the board, e.g. in practicing vocabulary use,
from grouped activity. The latter was dependent constructing sentences, and repeating words and
upon group work using laptops and linked audio phrases. The most interactive lessons were those
material while one group worked with the IWB where these activities were linked to the board.
and then groups moved to different activities in a In four of the 13 lessons this led to a combina-
subsequent lesson. tion of choral reading, repetition of phrases and
When the student groups were asked to iden- word completion using sentences from the board.
tify why lessons were of greater interest than in Overall time on task is greater when the IWB is
traditional teaching they identified: the focus of teaching and learning.
There was considerable concern that there
• The inherent interest of color, shading, dy- could be a novelty value in the use of the new
namics, hide and reveal and demonstration. technology, “but we have to remember that stu-
• The sequential development of ideas and dents are used to this at home” and “that they
exemplars resulting from pre-prepared and think advanced technology now.” One teacher
commercial software. commented “there is now danger that if we don’t

781
Interactive Whiteboards in the Web 2.0 Classroom

use the technology we will be seen as lacking in multiple-choice selection. This is very noticeable.
some way.” All the respondents accept this but it As the teacher you too are working in a commu-
is clear that teachers have developed strategies to nity, where you are visible. It does give a sense of
ensure that there should be a continuing upward competition, of expectation, the idea of can you
progression in learning and attainment. In a year beat it?” (male teacher, Spanish lesson).
seven French lesson the teacher used an intro-
ductory activity based upon naming colors, then
moved to five vocabulary development exercises Pedagogy
and finished with a learning check linked to boys
versus girls scoring to ensure that momentum was It was clear that teachers were using the learning
maintained, that all the students were taking part of concepts as a basis for cognitive understanding.
and that visual stimulation was used to the full As a result in all but two of the 13 lessons there
with a total of ten screens during the 35 minute were discernible cognitive aims and a series of
lesson. That said, the dynamism of the teacher activities to explore, develop, explain and reinforce
was important in supporting continuing learning subsequent understanding. This was summed up
— even broken with a two minute march to the one teacher as follows: “Sustained learner inter-
French alphabet to stimulate renewed activity. est works in a number of different levels. It is not
While it would be easy to claim great ad- just a gimmick ... the interaction is important, like
vantages for the IWB in motivating students at kids coming out to the board, having choices,
all ages it is evident that it is the quality of the e.g. they can decide on the verb ending, find the
teaching that ensures progress. Comparison was stem and match up the right pronoun. It makes
made of two lessons of vocabulary development concrete in their minds how the language works”
with year seven groups. In one there were seven (male teacher, German lesson).
screens used in the course of the lesson but these There was a high level of understanding that
were interspersed with pair work, a brief exercise students learn in different ways. This was seen
and a discussion about rooms in the house. The where a pattern of viewing pictures, learning as-
students were animated throughout. In a compa- sociated vocabulary, repeating its use in sentence
rable lesson, again with seven screens used, the construction, and then undertaking written or
teaching approach was much more didactic, there spoken group work ensured that: “we both enjoy
was little variation in activity from stage to stage teaching and learning more ... you can give clearer
in the lesson and the inter-relationship between examples which are more interesting because of
teacher and learners was authoritarian and defen- access to color and clip art. It’s more aesthetically
sive. In such circumstances the lesson could not pleasing and is good for visual and kinesthetic
have the vigor, and “fun” element shown with a learners and it’s useful in that you can jumble up
different teacher. sentences and get them involved in reconstruc-
But there is another subtle influence noted tion” (male teacher, French lesson).
by four of the respondents. This is because the Although there was some use of commercially
constant progression in an interactive situation developed activities, such as a short color recogni-
absorbs those who might otherwise become fidgety tion program — “we have developed our own
in a traditional classroom situation. They, in turn materials from a number of sources, including
are less “nagged” during the lesson, enjoyment download from the net, magazine and picture
increases and motivation is supported: “It enhances scanning and my own extensive library of clip
collaborative work. This may just take the form of art images” — this was seen to have advantages
kids shouting out, correcting each other, say in a in that what was developed was meeting specific

782
Interactive Whiteboards in the Web 2.0 Classroom

needs. Two teachers, however, expressed reser- as a bridge to the main part of the les-
vations — one about the time taken to produce son, and to revise necessary associated
good professional looking materials, and the learning.
other about “the danger of getting too structured • Proceeding to the main part of the les-
and then unable to work flexibly if a problem son where the IWB is the focus of much
occurs in the learning process for a particular activity being used for illustration, expla-
topic” (female teacher, French lesson). nation, sequenced ideas and the develop-
Teachers were all conscious of the need to ment of main principles. The progression
maximize interactivity between themselves, was through the use of vocabulary and
the students and the learning materials. This is its application in sentences reinforced
achieved through developing the opportunity to by practice and comprehension. During
use “visual manipulation” so that concepts can this section of the work the approach was
be illustrated and worked upon by the students; distinguished by challenged responses
the growth of shared evaluation of resources and with the emphasis on understanding and
the use of shared materials developed within then using language correctly — with
subject areas, and exploitation of immediacy of practice in the completion of sentences
feedback either through programmed software on the IWB reinforced by group activi-
or through the use of presentational tools as ties. In this way, as one teacher com-
with the colors program in French, or with right mented, “you move the students with
and wrong answer symbols. These programs are you.” Interview respondents identified a
most effective as starters or for work with the tension between those who thought that
least able when rapid responses and moving on time taken in managing the students’ use
enhance word manipulation. of the IWB while others were watching
There was also much debate about the place of could be seen as a loss to active learn-
traditional textbooks, exercise books, homework ing but in eight language lessons students
and other data sources in teaching. Over-writing were given tasks alongside the work be-
was seen to offer scope for assisting cognitive ing illustrated on the board so that all the
development by “showing the same thing in students were active.
different ways.” Much of the Internet use was • Concluding with a plenary session in-
to download games and activities that did just volving the use of recall, examples and
this by underpinning learning of vocabulary and previously worked material to ensure un-
phrase development, or even with some audio derstanding and to act as the basis for ex-
links to check pronunciation. Most importantly, tension work. This section of the lesson
however, were the ways in which the IWB was was more usually concerned with revis-
being used to underpin lesson structure and to iting vocabulary and structures and then
enhance cognitive development. Teachers vari- looking at an associated screen requiring
ously appear to use a structure of: comprehension or conversation as a con-
solidation for the lesson.
• Setting objectives with or without revis-
iting earlier IWB slides. Awareness of the need for cognitive develop-
• Using a bright and lively starter includ- ment and the place of concepts within this was
ing “drag and drop,” “hide and reveal” shown in the frequent reference to sequencing of
and multiple answers to stimulate inter- ideas, the availability of a range of pre-prepared
est, to offer a chance for brainstorming examples appropriate “to age and ability,” adapt-

783
Interactive Whiteboards in the Web 2.0 Classroom

ability of materials to allow for “alternative ap- and the understanding of these by students so that
proaches and the use of different ways of learning.” they could use the board to help in their evaluation
This was through vocabulary understanding and of progress. They showed an awareness of what the
pronunciation, and through phrase and sentence IWB could offer and in the two most stimulating
construction to use in verbal and aural comprehen- lessons Web 2.0 approaches were integrated into
sion. Three linguists outlined the use of supportive the teaching. In one lesson there were five groups
materials from the net or other sources, and three working at their own level in differing learning
referred to the need to help students understand situations. These included the use of an interactive
the technology e.g. in the use of pens and pro- software program at the IWB, access to the net by
grams, so that they could become fluent in the a group using a laptop, randomized questioning
interactivity required if whole class participation in pairs with an interactive program on a desktop,
was to be assured. and the preparation of a presentation by a group
There were comments that dependence on working with PowerPoint. It is possible that all
sequenced slides in some pre-prepared materials these approaches can exist individually without
in PowerPoint and Excel, as well as in some of being specifically labeled as Web 2.0 but they are
the commercial materials, could inhibit flexibil- now being used to shift the emphasis from teacher
ity in revisiting ideas and in offering alternative to student, from lecture to learning.
explanations appropriate to “whether they can
learn verbally or not.” This was not seen to be
a problem in the observed lessons because of The Developing agenda
the technological fluency shown in accessing
screens. There was a general view amongst those Arising from the agenda it appears that there are
interviewed that when the staff have the time to two pedagogic areas for further investigation. The
develop their materials and access to appropriate first is the relationship between the teacher, the
technological support it was possible to use the student and the materials involved. For enhanced
IWB to generate faster and more effective learn- interactivity to occur this has to be understood
ing, with tighter planning and the implementation as a chain reaction where the IWB is a means of
of lesson plans according to the need to cover the mediation between learners and learning. There
prepared material. are four elements in this process:
There was frequent reference in the interviews
to the need to match materials to the needs of the a. Teaching approach. Ernest (1994) suggests
students and that some differentiation of task, a simple scale for the approach used by
activity or outcome required teachers to be flex- teachers. At the lowest level the teacher is an
ible, adaptable, and “aware of the ways in which “instructor” concerned with the presentation
consolidation can occur without going back to of concepts as rules followed by practice. At
old fashioned practices such as copying.” This the higher level the teacher is “facilitator”
was illustrated in a comparison between two offering approaches that enhance under-
groups learning and applying clothing vocabulary standing, and at the highest level the teacher
showing that the more able group moved on to is a “mediator” bridging between student
determine the difference between summer and understanding and development. In their use
winter clothing while using similar screens of of the interactive whiteboards the instructor
information. is concerned with elements of presentation.
In pedagogic review the teachers also drew at- Conversely the mediator deals with issues
tention to the clear match of objectives to activities arising from questions and thereby regards

784
Interactive Whiteboards in the Web 2.0 Classroom

the interactive whiteboard as a vehicle for effect in consideration of the accommodation


interaction with students. available at differing costs within a French
b. The use of the interactive whiteboard. In holiday town where the input of so many
both the approaches discussed above, it Euros into a slot machine then produced a
is evident that the interactive whiteboard range of menus for description and selection.
enhances the role of the teacher regard- It is our contention that the use of particular
less of where s/he is on the spectrum. The manipulations might be used effectively
teacher-as-instructor will be working with to support the role of teacher-as-mediator
prepared material, to be presented in a logi- (Miller, Glover, Averis & Door, 2005).
cal sequence, and often with a PowerPoint c. Questioning. Experienced and effective
sequence as the basis of the teaching. The teachers use questioning intuitively. They
material is likely to be focused on statements probably think little about the nature or level
of facts and definitions, headings etc. but of the question but proceed as they think fit.
there will also be examples to be copied and Inexperienced and poor teachers appear not
exercises to be completed. Such material is to have such skills. Much has been written
likely to be organized, clear and monotone. about the nature of questions and the art of
On the other hand, the teacher-as-mediator questioning. Mason (2000), in his commen-
will be concerned with how the IWB can tary on the work of many in this field, clearly
support the features of mediation such as demonstrates the complexities of the process
modeling and coaching in relation to the and relates questioning to both conceptual
topic under consideration. In collaborative and cognitive development. Analysis of the
classrooms, modeling serves to share with video recorded lessons suggests that open
students not only what one is thinking about and closed questions and those focusing
the content to be learned, but also the pro- on product or process are frequently used
cess of communication and collaborative but are only partially helpful in developing
learning. Modelling may involve thinking higher order learning.
aloud (sharing thoughts about something) d. Learning Models. The fourth element in
or demonstrating (showing students how developing interactivity stems from the
to do something in a step-by-step fashion). learning model espoused by the teacher.
Coaching involves giving hints or cues, Observations have been made on the way
providing feedback, redirecting students’ in which teachers use the constructivist and
efforts, and helping them use a strategy. A social-constructivist views of learning as
major principle of coaching is to provide defined by Piaget (Piaget & Inhelder, 1974)
the right amount of help when students need and Vygotsky (1978). Students construct
it — neither too much nor too little so that concepts and meaning, as a solo activity,
students retain as much responsibility as based on their own experience. Associated
possible for their own learning (Tinzmann et with this model is the notion of “cognitive
al., 1990). This can be seen in the selection conflict” whereby children are exposed to
of appropriate adjectives or in the search for something that is different from (conflicts
word meanings. Miller, Glover and Averis with) their currently perceived models.
(2005) have suggested that as competence From Vygotsky, the focus is on the social-
improves teachers become more ready to constructivist view of language and the ex-
develop and use manipulatives as the basis of tent to which it is linked with the formation
interaction. This is seen to particularly good of knowledge. Furthermore, all knowledge

785
Interactive Whiteboards in the Web 2.0 Classroom

is a social construction and based on shared linking students to the IWB but ensuring that the
views and images. In language teaching the pace of the lesson was maintained. In the main
social context of much learning offers scope section of the lesson her movements were slower,
for constructivist learning to be enhanced. often indicating building or process stages, and
The opportunity to call on a vast range of then opened in an invitational way as explanation
Internet resources helps when technological was returned to the students for consolidation.
fluency allows access. There was then a return to quicker, pointing and
sequencing gestures as stages were revisited in the
plenary section of the lesson. When asked about
Gesture the pattern of interaction the teacher referred to
“the need to keep them on their toes, but to feel
In the introduction to this chapter we spoke of the that we were learning together.”
impact of intonation on language understanding Ferscha et al., (2005) attempt to extend the
and we return to this in considering the way in which gesture typology with three families of gestures
teachers, and indeed board-using students, gesture — hand gestures, gestures of an artifact held
while mediating between board and class. permanently (e.g. an IWB pen) and gestures that
There is an increasing awareness that teaching are detached from the hand and manipulated oc-
is a multi-modal activity drawing upon a range of casionally (e.g. change of software). All of these
communicatory activity including verbal, visual convey messages by the way in which they are
and interpersonal communication, as well as as- used. While such a system is of potential value for
sociated technology. Jewett (2004) has shown user interface computer technology development
that knowledge of multi-modal perception and it does not offer the sort of vocabulary of gestures
pedagogy can support both teachers and taught. that match the instinctive activity by teacher and
Abrahamson (2003) outlines the role of artifacts student in the classroom. However, it is the basis
or bridging tools, including gesture in that learning of gesture sensing devices and could well offer
process. Watson and De Geest (2005) outline the an insight into a typology because it may be that
need for consideration of all aspects of communica- students read more into body language, as shown
tion in teaching and learning, and Rasmussen et al when recall of an IWB screen fails and frustra-
(2004) explore the use of consistent gesture as part tion is indicated, or when invitations are issued
of these multi-modal approaches. Goldin-Meadow for students to work at the IWB and they respond
and Wagner (2005) take these patterns of gesture with acceptance gestures.
further and consider the impact of these on both In our analysis of video-recorded lessons it was
learners and their learning environment through possible to ascertain the reliance on gesture by
reflection of the state of knowledge and subsequent both teacher and students and the combination of
change through cognitive understanding. gesture as explanation, indication and invitation.
There is considerable evidence of the way in The IWB both encourages and reinforces learning
which the teachers using enhanced interactive through the use of visual as well as more transitory
approaches were constantly using recognizable gestures that offer shapes in the air. During this
gesture patterns. One female teacher used all- lesson gestures were used and emulated in an often
embracing movements to secure attention at the involuntary manner, in all three areas of gesture;
start of most lessons almost sweeping the students hand, software and artifact. The hand movements
along with her as she summed up her aims and then that mediated technology and learning through
moved towards the IWB. During starter periods her movements were:
hands were used in a quick to and fro movement

786
Interactive Whiteboards in the Web 2.0 Classroom

• Invitational, with the use of movement this can be harnessed to enhance teaching appears
linking students to the IWB, offering the to be related to other factors including the nature
pen for use, showing a step and offering of curriculum development programs, school
an opportunity for participation – often en- technology resource levels, and individual teach-
couraged with IWB software. ers’ planning and reflection. Kennewell (2001)
• Displaying, with hand gestures pointing to suggests that effective evaluation of ICT use will
material on the IWB and then using move- prompt more awareness of, and adaptation to, the
ment, highlighting or overwriting to indi- complexity of influences in the classroom. More
cate content or process. pessimistically, Robertson (2003) argues that
• Blocking, with hand gestures putting a bar- despite the potential impact of ICT on teaching
rier between the students and the IWB as and learning it remains a marginal influence on
a result of mistakes or the need to re-think student attainment. He argues that other signifi-
a process and then followed by an invita- cant changes have been more willingly achieved
tional reinforcement of process and use of in education and that the slow pace of change
drag and drop and over-writing to support in ICT may be related to social, anthropological
this. and cultural aspects of the human and computer
• Sequencing, with the gestures to indicate interaction. Kirschner and Selinger (2003) point
progression and using gestures to pose to the disparate technological competence of
a question and then to work through se- teachers and the children they teach and argue
quences of example questions. that if ICT is to be a core technology then teach-
ers need to recognize not only how to use the
It would seem that students learn not only different technologies but also follow through a
because of the difference in presentation but also five stage development from pre-novice through
because the IWB offers additional modes of ges- novice, apprentice, and practitioner to expert
turing that support verbal and visual explanation. user. The elements of this stage are the ability
It may be that this kinaesthetic quality will meet to reflect, evaluate and adapt both content and
the needs of those who cannot readily learn with approach to address student needs. If this is to
didactic approaches. Our observations suggest that be achieved, then the work of teacher educators
where teachers are using enhanced interactivity takes on a major role extending beyond the “how
with the IWB they are employing considerable to” to the “why” of ICT and the use of interactive
gesturing to great advantage whiteboards (Sturkle, 2004).
This requires understanding of the potential
of Web 2.0 tools in association with IWB use
Achieving interactivity to change the way in which teachers encourage
learning. Interactivity may be a matter of question
The starting point for the effective use of IWB and answer but Web 2.0 approaches in modern
technology has to be in teacher training. Nev- language teaching may open the way for the
ertheless, the move from traditional didactic use of interactive software, as for example in
approaches to changed pedagogy is complex. It vocabulary extension work; for the use of the
has been recognized that although UK student net in developing comprehension; for the use of
teachers are required to have a basic knowledge search engines in preparing presentations, and for
of computer use as a requirement for certification enhanced understanding of the cultural context.
many already have a high degree of computer In this way presentation spurs motivation and
literacy and technological understanding. Whether this, in turn, promotes higher attainment. This

787
Interactive Whiteboards in the Web 2.0 Classroom

is especially so where collaborative group work benefit (Matthew et al., 2002). Mooij (2004)
has been developed to meet differing learning argues that teachers have to be aware not only
styles. Web 2.0 tools provide the means of both of the technical aspects of newer technologies
conceptual and cognitive development. but also of the curricular and instructional gains
However at the time of the investigation it that can be made, and more importantly of the
was not possible to podcast and share videos. way in which technology and pedagogy can be
These technologies offer considerable opportu- integrated to achieve flexible and individually
nities (and threats) for teachers and pupils. The sensitive learning situations. Triggs and John
possibilities will undoubtedly be constrained by (2004) have demonstrated the need for work-
the technical, pedagogical and attitudinal back- ing groups at departmental, whole-school and
grounds of the teachers. Further limiting factors educational service levels, interconnecting for
will be the way in which uses of some Web 2.0 professional growth through the sharing of
technologies are “censored” and restricted by technical and pedagogic experience.
school firewalls. The recurrent theme is one of a discrete way
In language teaching students may find con- of teaching and learning using ICT and Taylor
siderable benefits in using (and creating) products (2004) suggests that this requires a three stage
that may help them with their study. Generally development from personalization to achieve
the technological skills will be within the grasp fluency in using the technology, through peda-
of learners — but the option to demonstrate and gogic sensitivity to its potential, to the develop-
use these skills may be overlooked. ment of contingent thinking to allow responsive
Even at the most basic technological level and reflective use of materials. In the context
this may require fundamental changes in aspects of continuing professional development, this
of initial teacher education. In simple terms, the requires strong support within teacher training
assertion that mentoring teachers should be at institutions and the schools with whom they
least competent in ICT use was found wanting work in partnership. This will then help teachers
by Cuckle and Clarke (2003) who comment on who have been inappropriately, or inadequately,
the considerable variation in student support trained in the pedagogy and do not realize the
between schools. When that competence occurs need to develop interactivity through the use
for Knezek and Christensen (2002) the focus of of a variety of teaching and learning styles,
subsequent change is determined by evidence artefacts and gesture — in short, coping with
that: the affordances of the technology (Conole &
Dyke, 2004). Failure to make a significant peda-
as teachers progress from lower level, simple ap- gogic change will, we suggest, lead to wasted
plications toward full integration of technology in opportunities and the danger that equipment
the classroom in support of higher cognitive func- with the potential to change understanding,
tions, attitudes progress in predictable patterns application and the conceptual development of
along with changes in their needs. (p. 375) learning will be at worst, unused, and at best a
presentational aid.
Once established as teachers and in continu- For this to occur there has to be further
ing professional development there is some consideration of the professional development
evidence that successful one-to-one coaching provided for users. Glazer and Hannafin (2006)
can be achieved where the technologically building on Vygotsky’s social constructivist
adept students are paired with teachers having approaches suggest that this exploration of what
a much wider pedagogic experience to mutual happens in the classroom is best undertaken as a

788
Interactive Whiteboards in the Web 2.0 Classroom

social enterprise where peers rely on the expertise teachers can then overcome the novelty factor. Our
and support of one another to adopt innovative evidence suggests that there is an initial period
practices: “Reciprocal interactions in a community where interest is stimulated by the cleverness of
of practice, where teachers take responsibility the technology, but after a period students are
for each other’s learning and development, may more aware of three great gains:
provide an effective means of supporting situated
professional learning” (p. 179). Contextual work 1. Brighter and clearer presentation of
by Schrum et al. (2005) points to the need for material
departments to continually refine, reassess and 2. Stepped learning and the ability to recall
redevelop their teaching approaches. Eekelen et earlier material
al. (2005) have shown that this process needs to be 3. Rapid responses to interactive examples so
regulated rather than self-regulated and unstruc- that learning is reinforced or revisited
tured — with implications for those responsible
for professional development, and Tearle (2003) Where students have reached this stage, they
shows that this is particularly true of learning in accept the IWB as part of the battery of learning
technology based contexts where the learning resources offered to them and progress beyond
culture is fundamental to teacher involvement novelty to enhanced learning. At this stage any pos-
and shared experience sible behavioral problems are usually overcome
because students are caught up in the sequence
and pace of learning and appear to “take off” in
Conclusion their understanding, achievement and consequent
self-esteem.
There appears to be a learning curve for both There is evidence that language teaching is
teachers and students. The former need time to being transformed by competent and confident
develop their technological fluency, apply peda- teachers but this is not to suggest that the IWB is
gogic principles to the available materials or to the a panacea for all ills. As yet, there is only a limited
development of materials, and then to incorporate shift in classroom practice and student learning
the IWB seamlessly into their teaching. Few teach- and transformation will require markedly changed
ers base all their lesson on the IWB all the time, and teacher understanding. Our evidence suggests
over half those interviewed stressed that the IWB that there is a teacher progression from sup-
has to be seen as part of the equipment available ported didactic to enhanced interactivity in their
but that there was still a need for the use of texts, classroom and pedagogical management. Where
exercises and other media. Teachers then appear to there is still reliance on the copying of material,
become more aware of the nature of interactivity textbook exercises and minimal conceptualization
and its stimulation as the basis for conceptual de- of learning so that it can be interactive, the gains
velopment and cognitive understanding. Students are minimized. Effective learning is inhibited
also need to have a range of manipulative skills where the IWB is given a novelty value by the
if they are to take part in lessons without loss of teacher so that it becomes something different,
self-esteem as technologically incompetent. Even where the physical surroundings are not condu-
so good practitioners ensure that all students have cive to IWB use and where the lesson lacks pace.
access to the board, and are given help if there are It is not sufficient to argue that the use of the
signs of unhappiness with the medium. IWB will, of itself, bring the classroom into the
It is only when basic technological fluency and Twenty First Century and the visually stimulated
pedagogical understanding has been achieved that environment. Effective teaching requires that the

789
Interactive Whiteboards in the Web 2.0 Classroom

technology and the pedagogy are directed towards Cuckle, P., & Clarke, S. (2003). Secondary
enhanced and structured understanding. “I love school teachers, mentors and student teachers’
my board because it gives so much to the kids,” as views in the value of ICT in teaching. Technol-
one teacher said, may be the clue that enthusiasm ogy, Pedagogy and Education, 12(3), 377–391.
can be regenerated not just in the students but in doi:10.1080/14759390300200168
the staff also.
Eekelen, I., Boshuizen, H., & Vermunt, J. (2005).
Self-regulation in higher education teacher
learning. Higher Education, 50(3), 447–447.
References
doi:10.1007/s10734-004-6362-0
Abrahamson, D. (2003). Embodied spatial articu- Ernest, P. (1994). The impact of beliefs on the
lation: a gesture perspective on student negotia- teaching of mathematics. In Bloomfield, A.
tion between kinaesthetic schemas and epistemic & Harries, T. (Eds.). Teaching and Learning
forms in learning mathematics. In McDougall, Mathematics. Derby: Association of Teachers of
D. E., & Ross, J. A. (Eds.). Proceedings of the Mathematics.
26th Annual Meeting of the International Group
for the Psychology of Mathematics Education 2, Fersha, A., Resmerita, S., Holzmann, C., &
791-797. Reichor, M. (2005). Orientation sensing for
gesture-based interaction with smart artefacts.
Armstrong, V., Barnes, S., Sutherland, R., Curran, Computer Communications, 28(13), 1552–1563.
S., Mills, S., & Thompson, I. (2005). Collaborative doi:10.1016/j.comcom.2004.12.046
research methodology for investigating teaching
and learning: The use of interactive whiteboard Glazer, E. M., & Hannafin, M. J. (2006). The
technology. Educational Review, 57(4), 457–469. collaborative apprenticeship model: Situated
doi:10.1080/00131910500279551 professional development within school settings.
Teaching and Teacher Education, 22(2), 179–193.
Belli, M. (2005). Technology is not white and doi:10.1016/j.tate.2005.09.004
black. Times Educational Supplement. Retrieved
May 1, 2008, from: http://www.tes.co.uk/search/ Glover, D., Miller, D., Averis, D., & Door, V.
story/?story_id=2063467 (2007). The evolution of an effective pedagogy
for teachers using the interactive whiteboard
Belshaw, D. (2007). Flattening the world: How to in mathematics and modern languages: An
harness Web 2.0 tools to engage learners inside empirical analysis from the secondary sector.
and outside the classroom. In-service develop- Learning, Media and Technology, 32(1), 5–20.
ment day Kings Heath Leading Edge Partnership. doi:10.1080/17439880601141146
Retrieved November 27, 2007, from: http://www.
slideshare.net/edte.ch Glover, D., & Miller, D. J. (2003). Players in the
Management of Change: introducing interactive
Conole, G., & Dyke, M. (2004). What are the whiteboards into schools. Management in Edu-
affordances of information and communica- cation, 17(1), 20–23. doi:10.1177/08920206030
tion technologies? ALT-J, 12(2), 113–124. 170010701
doi:10.1080/0968776042000216183
Goldin-Meadow, S., & Wagner, S. M. (2005).
How our hands help us learn. Trends in Cog-
nitive Sciences, 9, 234–241. doi:10.1016/j.
tics.2005.03.006

790
Interactive Whiteboards in the Web 2.0 Classroom

Gray, C., Pilkington, R., Hagger-Vaughan, L., & Mason, J. (2000). Asking mathematics questions
Tomkins, S. A. (2007). Integrating ICT into class- mathematically. International Journal of Math-
room practice in modern foreign language teaching ematical Education in Science and Technology,
in England: Making room for teachers’ voices. 31(1), 97–111. doi:10.1080/002073900287426
European Journal of Teacher Education, 30(4),
Matthew, K., Callaway, R., Letendre, C., Kimbell-
407–429. doi:10.1080/02619760701664193
Lopez, K., & Stephens, E. (2002). Adoption of
Hall, I., & Higgins, S. (2005). Primary school ICT by teacher educators: one-on-one coaching.
students’ perceptions of interactive white- Technology, Pedagogy and Education, 11(1),
boards. Journal of Computer Assisted Learn- 45–62. doi:10.1080/14759390200200122
ing, 21(2), 102–117. doi:10.1111/j.1365-2729-
McFarlane, A. (2005). The taming of the white-
.2005.00118.x
board. Times Educational Supplement. Retrieved
Hennessy, S., Wishart, J., & Whitelock, D. (2007). May 1, 2008, from: http://www.tes.co.uk/search/
Pedagogical approaches for technology-integrated story/?story_id=2153815
science teaching. Computers & Education, 48(1),
Miller, D., & Glover, D. (2004). Enhancing
137–152. doi:10.1016/j.compedu.2006.02.004
mathematics teaching through new technology:
Jewitt, C. (2004). Knowledge, Literacy and The use of the interactive whiteboard. Retrieved
Learning: Multimodality and New Technologies. May 1, 2008, from: http://www.keele.ac.uk/depts/
London: Routledge Falmer. ed/iaw/nuffield.htm.
Jones, K. (2004). Using Interactive Whiteboards Miller, D. J., Glover, D., & Averis, D. (2005).
in the Teaching and Learning of Mathematics: a Developing Pedagogic Skills for the Use of the
research bibliography. Micro Math, 20(2), 5–6. Interactive Whiteboard in Mathematics. British
Educational Research Association, Glamorgan.
Kennewell, S. (2001). Interactive whiteboards
Retrieved September 25, 2007, from: http://
— Yet another solution looking for a problem.
www.keele.ac.uk/depts/ed/iaw/docs/BERA%20
Information Technology in Teacher Education,
Paper%20Sep%202005.pdf.
39, 3–6.
Miller, D. J., & Glover, D. C. (2006) Enhanced
Kent, M. (2006). Our journey into whiteboard
secondary mathematics teaching: gesture and
hell. Times Educational Supplement. Retrieved
the interactive whiteboard. Paper presented at
May 1, 2008, from: http://www.tes.co.uk/search/
the British Educational Research Association
story/?story_id=2287557
Annual Conference, University of Warwick,
Kirschner, P., & Selinger, M. (2003). The state September 6-9.
of affairs of teacher education with respect to
Miller, D. J., Glover, D. C., Averis, D., &
information and communications technology .
Door, V. (2005). The Interactive White-
Technology, Pedagogy and Education, 12(1),
board — A literature survey. Technology,
5–17. doi:10.1080/14759390300200143
Pedagogy and Education, 14(2), 155–170.
Knezek, G., & Christensen, R. (2002). Impact of doi:10.1080/14759390500200199
new information technologies on teachers and stu-
Mooij, T. (2004). Optimising ICT effectiveness
dents. Education and Information Technologies,
in instruction and learning. Computers & Edu-
7(4), 369–376. doi:10.1023/A:1020921807131
cation, 42(1), 25–44. doi:10.1016/S0360-1315-
(03)00063-0

791
Interactive Whiteboards in the Web 2.0 Classroom

Piaget, J., & Inhelder, B. (1974). The child’s Triggs, P., & John, P. (2004). From transaction to
construction of quantities. London: Routledge transformation: information and communication
& Keegan Paul. technology, professional development and the
formation of communities of practice. Journal
Rasmussen, C., Stephan, M., & Allen, K. (2004).
of Computer Assisted Learning, 20(6), 426–439.
Classroom mathematical practices and gestur-
doi:10.1111/j.1365-2729.2004.00101.x
ing. The Journal of Mathematical Behavior, 23,
301–323. doi:10.1016/j.jmathb.2004.06.003 Vygotsky, L. S. (1978). Mind in society. Cam-
bridge, MA: Harvard University Press.
Robertson, J. W. (2003). Stepping out of the box:
rethinking the failure of ICT to transform schools. Watson, A., & De Geest, E. (2005). Principled
Journal of Educational Change, 4(94), 323–344. teaching for deep progress: improving math-
doi:10.1023/B:JEDU.0000006047.67433.c5 ematical learning beyond methods and materials.
Educational Studies in Mathematics, 58, 209–234.
Schrum, L., Burbank, M. D., Engle, J., Chambers,
doi:10.1007/s10649-005-2756-x
J. A., & Glassett, K. F. (2005). Post-secondary edu-
cators’ professional development: Investigation Watson, D. (2006). Understanding the relation-
of an online approach to enhancing teaching and ship between ICT and education means exploring
learning. The Internet and Higher Education, 8(4), innovation and change. Education and Informa-
279–289. doi:10.1016/j.iheduc.2005.08.001 tion Technologies, 11(3/4), 199. doi:10.1007/
s10639-006-9016-2
Smith, J., Higgins, S., Wall, K., & Miller, J.
(2005). Interactive whiteboards: boon or band- Weiss, D. F. (2005). Keeping it real: The rationale
wagon? A critical review of the literature. Journal for using manipulatives in the middle grades.
of Computer Assisted Learning, 21(2), 91–101. Mathematics Teaching in the Middle School,
doi:10.1111/j.1365-2729.2005.00117.x 11(5), 238–242.
Sturcke, J. (2004). ICT strategy? ‘Just keep tak- Xhakli, K. (2008). Going online: How innova-
ing the tablets’. Times Educational Supplement. tive projects can transform education and ICT
Retrieved May 1, 2008, from: http://www.tes. opportunities. Kosovo: IPKO Institute.
co.uk/search/story/?story_id=2055679
Ziolkowski, R. (2004). Interactive Whiteboards:
Taylor, L. (2004). How student teachers develop Impacting Teaching and Learning. Media &
their understanding of teaching using ICT. Jour- Methods, 40(4), 44.
nal of Education for Teaching, 30(1), 43–56.
doi:10.1080/0260747032000162307
Tearle, P. (2003). Enabling teachers to use informa- KEY TERMS
tion and communications technology for teaching
and learning through professional development: Artifact (BE Artefact): Artifact is an object
influential factors. Teacher Development, 7(3), or item. However it can also be the on screen
457–472. doi:10.1080/13664530300200222 representation of an object or an item.
Gesture: This is a term encompassing human
Tinzmann, M. B., Jones, B. F., Fennimore, T. F., actions here associated with the use of the interac-
Bakker, J., Fine, C., & Pierce, J. (1990). What tive whiteboard e.g. hand and body movements
Is the Collaborative Classroom? NCREL: Oak and facial expressions. There is evidence that users
Brook. develop consistent hand and facial gestures e.g. in

792
Interactive Whiteboards in the Web 2.0 Classroom

seeking responses, rejecting wrong responses and Presentation: Presentation is the use of the
that learners assimilate these as part of the teaching software potential of the interactive whiteboard
package offered by individual teachers. to enhance the way in which words, concepts,
Interactive Whiteboard (IWB): An interac- ideas and relationships are displayed. Design,
tive whiteboard consists of a computer linked to color, movement and more complex virtual ma-
a data projector and to a touch sensitive large nipulatives offer a superior way of showing data
electronic screen usually fixed to a wall. Images on an interactive whiteboard with the intention
from the computer are then displayed onto the of prompting learner participation. The use of a
whiteboard by means of the data projector. These variety of means of display may meet the needs
images can be manipulated at the electronic screen of learners with differing learning styles.
usually by means of a special pen or a finger (this Social Constructivist Approaches: These
depends on the properties of the electronic screen). are based upon the complex interaction between
The term interactive whiteboard often refers only teacher and learner, or between learners, and re-
to the electronic screen. late to the way in which we learn from each other
Interactivity: Interactivity is an approach to with greater facility once the social network of
learning in which teacher and learner interact to the context is known and when the culture of the
ensure understanding, enhance conceptual devel- learning group has been developed.
opment and stimulate debate. Learning is stimu- Virtual Manipulatives: A virtual manipula-
lated through participation rather than through rote tive is a computer program that represents a piece
or passive learning which characterises didactic of equipment on a computer screen. Examples
approaches. include a cannon that can fire cannon balls, a pro-
Motivation: In this context, is an outcome of tractor for measuring angles and a geoboard where
presentation because of the greater interest offered you can place and manoeuvre “elastic bands” on
to learners and the reinforcing of concepts through a grid on “nails.” Virtual manipulatives are most
learner engagement. commonly written in Flash and JavaScript.

This work was previously published in Handbook of Research on Web 2.0 and Second Language Learning, edited by M. Thomas,
pp. 506-525, copyright 2009 by Information Science Reference (an imprint of IGI Global).

793
794

Chapter 3.8
Web 2.0 Technologies:
Social Software Applied to Higher
Education and Adult Learning
Teresa Torres-Coronas
Universitat Rovira i Virgili, Spain

Ricard Monclús-Guitart
Universitat Rovira i Virgili, Spain

Araceli Rodríguez-Merayo
Universitat Rovira i Virgili, Spain

M. Arántzazu Vidal-Blasco
Universitat Rovira i Virgili, Spain

M. José Simón-Olmos
Universitat Rovira i Virgili, Spain

Abstract INTRODUCTION

Web 2.0 technologies are playing an important role Education has traditionally been conducted face-
in building social capital through increasing flows of to-face, with professors performing outstanding
information, and building on knowledge and human magisterial classes in front of the learners. During
capacity of learning. The purpose of this chapter the centuries, students and professors have shared
is to show the role that social software, a compo- the same time and same space frame. Nowadays,
nent of Web 2.0 technologies, can play in higher things are quite different. Information technology
education and adult learning. This chapter focuses (IT) is a reality affecting the whole education sys-
on the role of Web 2.0 technologies in promoting tem from primary school to higher education and
learning. New learning paradigms and pedagogical adult learning. IT is having a considerable impact
applications are also discussed. on the learning providers, on the learning process
itself and, of course, on any agent involved in the
DOI: 10.4018/978-1-60566-739-3.ch059 process.

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Web 2.0 Technologies

History has demonstrated that technology and other forms of e-learning. These efforts are
affects education profoundly. Considering the predicated upon an assumption that “university
definition of technology broadly, one may say that students are inherently inclined towards using
prehistoric people used primitive technologies to the internet as a source of information within
teach skills to their young (Frick, 1991). Whenever their day-to-day lives and, it follows, disposed
a new medium entered the picture, a new wave of towards academic use of the internet” (Selwyn,
educational delivery arrived. Radio, television, 2008, p. 12).
and now computers have all impacted the field But, today, the traditional approach to e-learn-
of distance education. Though some studies (see ing is currently changing from the use of Virtual
Russell, 1999) report no significant differences in Learning Environment (VLE) to e-learning 2.0,
performance between face-to-face instruction and an approach that combines the use of complemen-
technology supported environments. tary tools and Web services -such as blogs, wikis,
Nowadays, campuses are networked, faculty trackback, podcasting, videoblogs, and other
post their notes on Web pages, students access the social networking tools- to support the creation
library from their rooms, and entire classes can of ad-hoc learning communities. In this context,
have discussions via chat software (Rice-Lively, most of the current research tends to be concerned
2000). This development has recently come to with the potential of the worldwide Web and
be labeled under the by now commonly accepted other internet applications to accelerate university
term e-learning (Hudson, 2003). students’ learning and knowledge-building, and
The European e-Learning Action Plan 2001 support interactivity, interaction and collaboration
(European Commission, 2001) defines e-learning (Selwyn, 2008).
as the use of new multimedia technologies and This proposal aims to provide an introduc-
the Internet to improve the quality of learning tory perspective on the learning impacts of new
by facilitating access to resources and services media and Web 2.0 information and communica-
as well as remote exchanges and collaboration. tion technologies on the e-learning environment.
This requires new e-interaction and e-commu- Web 2.0 technologies are playing a crucial role
nication competencies and a reorganization of in building of social capital through increasing
e-learning structures. Components can include flows of information, and building on knowledge
content delivery in multiple formats, manage- and human capacity for learning. Social software
ment of the learning, and a networked com- has emerged as a major component of the Web
munity of learners (Gunasekaran, McNeil, & 2.0 technology movement. But, how can social
Shaul, 2002). Internet/World Wide Web have software play a role in higher education and adult
meant that opportunities have been identified learning? To answer this question, this proposal
for developing distance learning activity into a will focus on the role of Web 2.0 technologies in
more advanced online environment. It is known promoting learning. Pedagogical applications,
as Virtual Learning Environment (VLE), which which stem from their affordance of collabora-
eliminate geographical barriers while providing tive knowledge discovery, will be discussed. At
increased convenience, flexibility, individualized the same time the chapter will also explore the
learning, and feedback over traditional classroom pedagogical methodology involved considering
(Kiser, 1999). Higher education institutions de- that e-learning Web 2.0 leads us from constructiv-
vote substantial resources to providing students ism to navigationism. Finally, some suggestions
with access to internet-based information, VLEs are made for future research in this field.

795
Web 2.0 Technologies

BACKGROUND platform or an attitude rather than technology


(Downes, 2005).
Social Software and Web 2.0 Web 2.0 technologies are already having a
Technologies as a Must for a Digital significant impact on the way in which we com-
Life municate in both our personal and professional
lives. Mejias (2005) wrote down a list of non-
The term social software is generally attributed definitive kinds of social software applications,
to Tim O’Reilly. Social software includes a large arranging technologies according to their social
number of tools used for online communication, function (learning, selling, classifying, defining
e.g. instant messaging, text chat, internet fora, communities, and so on). Mejias (2005) stated
Weblogs (or blogs for short), wikis, social network that most social software products incorporate
services, social guides, social bookmarking, social functions from more than one category and,
citations, social libraries and virtual worlds. also, most of them pose challenges to pedagogi-
O’Reilly (2005) presented Web 2.0 as a second cal approaches. And, these challenges are today
stage in the development of the Web. He describes instructors’ challenges.
Web 2.0 as an “architecture of participation’’ where Organizational structures in the 21st century
collective intelligence generates a ‘‘network ef- are also increasingly networked and with virtual
fect’’ leading to Websites that become more valu- teams becoming the norm. Virtual team working
able as more people participate (O’Reilly, 2003). requires tools that enable the exchange of docu-
For McGee and Begg (2008) “Web 2.0 represents ments and information and collaborative creation.
a group of Web technologies with a user-centric Wikis and blogs have taken relatively little time to
focus that actively change and evolve with user become part of the suite of tools used for collab-
participation” (p. 164). Web 2.0 is referred to as a orative virtual projects. In this new organizational
technology (Franklin & Van Harmelen, 2007) and landscape, enterprise social or “collaborative
at the same time as a community-driven online software is probably the most visible current

Table 1. Different types of social software and its applications

Social software Applications


Multiplayer gaming environments Multi-User Dungeons (MUDs), Massively-Multiplayer Online Games (MMOGs).
Discourse facilitation systems Synchronous: instant messaging (IM) and chat (e.g. Windows® Live Messenger,
AOL Instant Messenger, Yahoo® Instant Messenger, Google™ Chat, Skype™); chat.
Asynchronous: e-mail, bulletin boards, discussion boards, moderated commenting systems (e.g.
Slashdot)
Content management systems Blogs, wikis, document management (e.g. Plone™) and, Web annotation utilities.
Product development systems Especially for Open Source software (e.g. Sourceforge.net®, Libresource)
Peer-to-peer (P2P) file sharing Napster®, Gnutella, BitTorrent™, eMule, iMesh
systems
Selling/purchasing management eBay™
systems
Learning management systems Blackboard, WebCT, Moodle
(LMSs):
Relationship management systems Friendster®, Orkut
Syndication systems list-servs, RSS aggregators
Distributed classification systems Flickr®, del.icio.us.

796
Web 2.0 Technologies

challenge. Interpersonal communication has Learning Paradigm Shifts


become an integral part of the process of content
creation, hence the value placed on communities Since many years ago, different theories have been
and networks” (Abell, Chapman, Phillips, Stewart developed to explain how we learn. Behaviorism,
& Ward, 2006, pp. 244-245). cognitivism, and constructivism are the three
Enterprise Social Software is a term describing broad learning theories most often utilized in the
social software in “enterprise” (business) con- creation of learning environments. Neither of
texts -definition provided by Wikipedia- [http:// these views can be regarded as exclusively right
en.wikipedia.org/wiki/enterprise_social_soft- or wrong. It is, however, necessary to know that
ware/]. It includes social and networked modi- constructivism is presently accepted as the most
fications to company intranets and other classic relevant of the three. In the pedagogical arena
software platforms used by large companies to it is a must to analyze how these models allow
organize their communication’. Enterprise 2.0 is instructors to create the circumstances best suited
a paradigm shift. Organizations are increasingly to facilitate student learning.
focusing on leveraging internal information and The first one, behaviorism, is a worldview
on connecting people to people and people to that assumes a learner is essentially passive,
content. Web 2.0 tools and techniques focus on responding to environmental stimuli. It stems
collaboration and information/knowledge shar- from the work of Pavlov –the father of classical
ing. Business Information Survey explores the conditioning- and Skinner –the father of operant
penetration of Web 2.0 tools. Results show that conditioning. Behavior theorists define learning
there is great strategic interest in social technology as nothing more than the acquisition of new be-
and Web 2.0 tools and techniques, but not much havior. Learning is “any more o less permanent
serious deployment yet (Foster, 2008). But, in change in behaviour which is result of experience”
our digital world, digital natives (Prensky, 2001) (Borger & Seaborne, 1966, p.16). The behaviourist
eagerly embrace social software developing the definition of learning focuses on the behavioural
skills necessary to engage with social and techni- outcomes of learning, rather than on knowledge,
cal change, and to continue learning throughout attitudes and values.
the rest of their lives.. After the behavioural theories came cognitive
As workers live Web 2.0 digital lives, organi- ones. The most influential theorists were Piaget
zations also will need to update their e-learning and Vygotsky. Cognitivism theories seek to explain
corporate practices. In that sense, Trondsen how the brain processes and stores new informa-
(2006) predicts strong uptake of virtual worlds tion. People are rational beings that require active
in corporate learning and notes a number of pilot participation in order to learn, and whose actions
projects underway in company learning contexts. are a consequence of thinking. The learner is
As students live Web 2.0 digital lives, instructors viewed as an information processor.
need to begin to deeply explore and develop new Constructivism as a paradigm posits that learn-
learning paradigms with these technologies and ing is an active, constructive process. According
practices. And, finally, as the students of today to a constructivist view, learning is seen as the
grow into the leaders of tomorrow, they will bring individualized construction of meanings by the
these technologies into their organizations, making learner. The learner is an information construc-
their use an essential part of the future of world tor. Constructivist learning theories posit that
of work and life-long learning. knowledge is built by the learner, not supplied by
the teacher (Piaget, 1967). People, by reflecting
on their experiences, actively construct their own

797
Web 2.0 Technologies

subjective representations of objective reality. 2004) provides a summary of the connectivist


Each of us generates our own mental models, learning skills required within a navigationist
which we use to make sense of our experiences. learning paradigm:
Learning, therefore, is simply the process of ad-
justing our mental models to accommodate new • Learning is a process of connecting spe-
experiences. cialized nodes or information sources.
In the present landscape of technological • Capacity to know more is more critical
change, important transformations are underway than what is currently known.
in terms of how we teach and learn. There is a • Nurturing and maintaining connections is
growing shift on the need to support the acquisition needed to facilitate continual learning.
of knowledge and competencies to continue learn- • Ability to see connections between fields,
ing throughout life. “With respect to ICT, we are ideas, and concepts is a core skill.
witnessing the rapid expansion and proliferation • Currency (accurate, up-to-date knowledge)
of technologies that are less about “narrowcast- is the intent of all connectivist learning
ing”, and more focussed on creating communities activities.
in which people come together to collaborate, • Decision making is itself a learning
learn and build knowledge” (McLoughlin & Lee, process.
2007, p. 664). So, constructivist approaches have
grown to include social constructivism, which Connectivist learning skills are required to
refers “to learning as the result of active partici- learn within a navigationist learning paradigm.
pation in a community” where new meanings are And this is why Brown (2006) states that “con-
co-constructed” (Brown, 2006, p. 111). Different nectivism is part and parcel of navigationism,”
learning strategies have been designed based on (p. 117) a learning paradigm that needs further
a community supported constructionist approach development. The main practical implication of
in which constructionism strategy –a strategy Brown’s work is that teachers and trainers should
connected with experiential learning and based become coaches and mentors within the knowl-
upon constructivist theories of learning- is situated edge and digital era and learners should acquire
in a supportive community context (Bruckman, navigating skills for a navigationist learning para-
1998). This approach emphasizes the importance digm. To enhance e-learning Web 2.0 over time,
of social aspect of learning environment. The it is vital for instructors to ground their designs
construction of new knowledge is the aim of these on established learning theories and report how
learning theories. related learning experiences are integrated with
But beyond constructivism and social con- Web 2.0 tools so instructors can determine what
structivism new paradigms are emerging. Brown Web 2.0 tools have the greatest effect on learner
(2006) focus on navigationism as the last learning motivation and performance. E-learning Web 2.0
paradigm shift. In this new learning paradigm is the key solution to equipping people with the
the emphasis will be on knowledge navigation. evolving knowledge and skills that will be needed
Learning activities will be focused on exploring, to adapt to the continuously changing nature of
connecting, evaluating, manipulating, integrat- the information society. At the same time, the
ing and navigating. Learning will take place major aim in education is to produce autonomous
when learners solve contextual real life problems learners. For Franklin and van Harmelen (2007, p.
through active engagement in problem-solving 21) “the growing Personal Learning Environment
activities, and networking and collaboration. (PLE) movement has a significant Web 2.0 fol-
Siemens’ principles of connectivism (Siemens, lowing which claims that PLEs are social software

798
Web 2.0 Technologies

Table 2. Comparison of characteristics of Web 1.0 versus Web 2.0 educational Websites. Source: McGee
and Begg (2008, p. 167)

Web 1.0 Web 2.0


Course Websites using content management systems. Faculty blogs, student discussion groups. Podcasts.
An expert (course director) produces a syllabus which resides on a Students in a course contribute to syllabus content with questions
curriculum Website. and answers to supplement expert materials.
Single Website, which displays the same content and design for all Personal Websites, with customized data sources and layout for
users. individual users
Posting problem based learning cases to a curriculum Website. Small groups have their own Website to which they add learning
objectives and educational content related to their coursework

tools that help or enable learners to take control key differences between Web 1.0 and Web 2.0
of their own education” and learning processes (see Table 2). The new user-centered paradigm in
throughout their lives. which users are both producers and consumers of
content and services has evolved from previous
Web developments. The Web before the dot.com
POSSIBILITIES AND APPLICATIONS crash is usually referred to as Web 1.0. O’Reilly
OF WEB 2.O TOOLS (2005) cites a number of examples of how Web
2.0 can be distinguished from Web 1.0, such as
As Owen, Grant, Sayers, and Facer (2006) state Web 1.0 was mainly a platform for information,
“Web 2.0 will lead to e-Learning 2.0, to a rethink- but Web 2.0 is also a platform for participation.
ing of the relationship between technology and Web 1.0 tools can be used for the delivery of the
learning, to the development of educational prac- course materials and for communication but Web
tices that place the learner at their heart through 2.0 tools (such as blogs) can be integrated in a e-
the creation of collaborative, community-based learning environment to a shift from a “knowledge
learning experiences. To explore this further we transfer model” to a “knowledge construction
touch now on the key theme of the potential shift model” as presented by Virkus (2008).
in thinking from ‘e-learning’ to ‘c-learning’ (p. If one were to apply Web 2.0 concepts, “the
10). Virtual communities of learning also offer lecture notes could become wikis (Wikipedia), the
the promise of bridging the worlds of work and slides would become an image sharing collection
education. (akin to Flick®), and students would subscribe
Some of the key attributes of social software to audio and video recordings (on a site like
in relation to education are that it (Owen et al., iTunes™), ideally all within an integrated ‘‘virtual
2006): “Delivers communication between groups, learning environment.’’ This online environment
enables communication between many people, would allow students to create their own views of
provides gathering and sharing resources, delivers their learning material and combine, with their own
collaborative collecting and indexing of informa- notes and external information resources. In Web
tion, allows syndication and assists personaliza- 2.0 parlance this is a ‘‘mash up,’’ where content
tion of priorities, has new tools for knowledge from different sources is combined by a user to
aggregation and creation of new knowledge and, create something new” (McGee & Begg, 2008,
delivers to many platforms as is appropriate to the p. 167). Web 2.0 is suitable for educational and
creator, recipient and context”. lifelong learning, because our knowledge society
To help apply Web 2.0 to education McGee is built on digital environments of work and social
and Begg (2008, p. 167) summarize briefly the communication, and educational practices must

799
Web 2.0 Technologies

Table 3. Educational applications of Web 2.0. Source: Franklin & Van Harmelen (2007, pp. 5-7)

Web 2.0 tool Description Educational application


Blogs A system that allows an author to publicly display time-ordered A blogger can build up a corpus of interre-
articles. lated knowledge.
Teachers can use a blog for course announce-
ments, news and feedback.
Blogs can be used with syndication technolo-
gies to enable groups of learners and teachers
to easily keep track of new posts.
Wikis A system that allows one or more people to build up a corpus of Wikis can be used for the creation of anno-
knowledge in a set of interlinked Web pages. tated reading lists by one or more teachers
Wikis are suited to the incremental accretion
of knowledge by a group, or production of
collaboratively edited material.
Social It provides users the ability to record (bookmark) Web pages, and To build up collections of resources.
bookmarking tag those Groups of users with a common interest can
records with significant words (tags) that describe the pages being team together to use the same bookmark-
recorded. ing service to bookmark items of common
interest.
Media-sharing Sstore user-contributed media that allows users to search for and Podcasts can be used to record lectures
services display content. Podcasts can be used to supply audio tutorial
Compelling examples include YouTube™ (movies), iTunes® material
(podcasts and vidcasts), Flickr® (photos), Slideshare (presenta- Instructional videos and seminar records can
tions), DeviantArt (art work) and Scribd (documents). be hosted on video sharing systems.
Social networking Systems that allow people to network together for various pur- LinkedIn® acts, at a professional level, as a
and social presence poses, such as Facebook© and MySpace® (for social networking model of educational use in the way in which
systems / socialising), LinkedIn® (for professional networking), Second it can be used to disseminate questions across
Life™ (virtual world) and Elgg (for knowledge accretion and the community for users seeking particular
learning). information.
There are a wide variety of educational ex-
periments being carried out in Second Life.
Collaborative These allow users in different locations to collaboratively edit For collaborative work over the Web.
editing tools the same document at the same time, such as Google™ Docs &
Spreadsheets.
Syndication and A world of newly added and updated shared content. A feed Feed Readers enable students and teachers to
notification reader (or aggregator) is used to centralize all the recent changes become aware of new blog posts in educa-
technologies in the sources of interest, and a user can easily use the reader/ag- tional blogging scenarios, to track the use of
gregator to view recent tags in social bookmarking systems, to keep
additions and changes. This relies on protocols called RSS (Re- track of new shared media, and to be aware of
ally Simple Syndication) and Atom to list changes (these lists of current news
changes are called feeds, giving rise to the name feed reader).

foster a creative and collaborative engagement • Both instructors and students must value
of learners with this digital environment in the an educational approach where learner
learning process (Guntram, 2007, p. 17). participation and contribution are balanced
Table 3 summarizes some educational applica- with acquisition.
tions of Web 2.0 tools included in Franklin and • A pedagogical approach must be used that
Van Harmelen’s (2007) work. reflects contribution-oriented activities
To embed Web 2.0 tools and processes within where students create some of their own
mainstream higher education practice the follow- learning resources.
ing need to be in place (Collis & Moonen, 2008, • The approach must be scaffolded in prac-
p. 100): tice by interlinked support resources for

800
Web 2.0 Technologies

both instructors and students. Uncertainty beginning to understand the opportuni-


must be reduced as much as possible for ties that mobiles technologies provide
the students in terms of what is expected of for learning. As Wilson (2006) points out
them, and to what standard. “Web platforms that allow moblogging
• The processes as well as the products (blogging from mobile phones), vlog-
produced by the students must be as- ging (video blogging) and other forms of
sessed as part of overall course assessment 3G-enabled participation are increasingly
practices. popular and show clearly the potential for
user-generated 3G content to be integrated
In higher education and adult learning educa- in an architecture of participation” (p. 239).
tional applications of Web 2.0 tools add extra value Mobile technology will play a key role in
to the learning experience and have an unlimited the new e-learning 2.0 paradigm.
potential. So far, we have briefly summarized • e-Learning 2.0 indicators need to be further
the increasingly varied ways in which these new developed in order to monitor progress in
tools can be used to construct the navigationist the use of Web 2.0 in formal and informal
learning paradigm. This new learning paradigm education.
2.0 represents an opportunity to revolutionize • Education methods, learning communities
the way human beings learn, interact, innovate organization are essential aspects in this
and develop. context.
• Research also needs to provide a holistic
view of students’ actual use of the social
FUTURE TRENDS software in higher education and adult
learning.
Different subjects need to be explored in detail to • Another key issue for any future research
step up research —educational, socio-economic is to explore what forms of knowledge
and technological — in the field of e-learning 2.0 students obtain from social software and,
and in the use of Web 2.0 tools in higher education most importantly, how students use such
and adult learning. knowledge. In-depth qualitative research
should be carried out to understand how is
• Special attention need to be devoted to us- built through Web 2.0 technologies.
ing emerging technologies (GRID, Web • The concept of virtual campus and virtual
3.0) for the development of innovative ap- networks for cooperation and collaboration
plications for education and training. In this needs to be revisited.
new technological environment, the ques- • In the virtual world, social networking
tion of how to motivate and socialize the functions can enable learners to aggregate
student as an active learner needs also to into communities of interest and evolve
be raised. As Hvid and Godsk (2006) state into communities of learning or practice.
“e-learning platforms needs an aesthetic We need to understand the formation of
perspective instead of mainly addressing these communities and ways to facilitate
usability and function”. (p. 210) the contribution of cybersocial network-
• In the near future, portable and personal ing to the learning and engagement of stu-
technologies will offer new opportuni- dents and teachers (Computing Research
ties to connect people and to create new Association, 2005).
e-learning 2.0 environments. We are only

801
Web 2.0 Technologies

• E-learning Web 2.0 may be able to reach Borger, R., & Seaborne, A. (1966). The psychology
learners who are disadvantaged by the digital of learning. Harmondsworth: Penguin Books.
divide. If it also important to define a research
Brown, T. H. (2006). Beyond constructivism: Nav-
agenda which takes into account individual
igationism in the knowledge era. Horizon, 14(3),
differences in learning, and special needs
108–120. doi:10.1108/10748120610690681
education to exploit the potential of Web 2.0
technology to provide remedial measures in Bruckman, A. (1998). Community support for con-
the case of disability, exclusion, difficulty in structionist learning. Computer Supported Cooper-
gaining access to learning, or where conven- ative Work: The Journal of Collaborative Comput-
tional education does not work. ing, 7, 47–86. doi:10.1023/A:1008684120893
• Finally, special attention needs to be giv-
Collis, B., & Moonen, J. (2008). Web 2.0 tools and
en to the promotion of gender equality in
processes in higher education: Quality perspec-
building e-learning 2.0 communities and
tives. Educational Media International, 45(2),
social capital.
93–106. doi:10.1080/09523980802107179
Bearing in mind all these agendas, e-learning Computing Research Association. (2005). Cyber
2.0 are likely to be a fertile research field. infrastructure for education and learning for the
future: A vision and research agenda. Retrieved
September 15, 2008, from http://www.cra.org/
Conclusion reports/cyberinfrastructure.pdf
Downes, S. (2005). e-Learning 2.0. eLearn Maga-
Each new wave of technological innovation promises
zine: Education and Technology in Perspective.
to revolutionize education, as we know it. The emer-
Retrieved September 15, 2008, from http://www.
gence of e-learning Web 2.0 is currently affecting
elearnmag.org/subpage.cfm?section=articles&ar
most colleges, universities, and corporations. Now it
ticle=29-1
is time to step back and question the pedagogical prin-
ciples that inform our learning paradigms because European Commission. (2001). The elearning
Web 2.0 technologies have to be implemented taking action plan. Designing tomorrow’s education.
into account pedagogical perspectives. The use of Retrieved September 15, 2008, from http://
Web 2.0 technologies in higher education and adult eurlex.europa.eu/LexUriServ/LexUriServ.
learning is still a new technological phenomenon do?uri=COM:2001:0172:FIN:EN:PDF
which will only “become valuable in education if
Foster, A. (2008). Business information survey.
learners and teachers can do something useful with
Business Information Review, 25(1), 13–31.
it” (OECD, 2001, pp. 24-25).
doi:10.1177/0266382107088221
Franklin, T., & Van Harmelen, M. (2007). Web 2.0
References for content for learning and teaching in higher
education, JISC, Bristol. Retrieved September
Abell, A., Chapman, D., Phillips, P., Stew- 15, 2008, from http://www.jisc.ac.uk/media/
art, H., & Ward, S. (2006). Roles in the e- documents/programmes/digitalrepositories/
landscape: Who is managing information? Web2-content-learning-and-teaching.pdf
Business Information Review, 23(4), 241–251.
doi:10.1177/0266382106072249 Frick, T. (1991). Restructuring education through
technology. Bloomington, IN: Phi Delta Kappa
Educational Foundation.

802
Web 2.0 Technologies

Gunasekaran, A., McNeil, R. D., & Shaul, D. O’Reilly, T. (2003). Architecture of participation.
(2002). E-learning: Research and applications. In- Retrieved September 15, 2008, from http://www.
dustrial and Commercial Training, 34(2), 44–53. oreillynet.com/pub/wlg/3017
doi:10.1108/00197850210417528
O’Reilly, T. (2005). What is Web 2.0? Retrieved
Guntram, G. (Ed.). (2007). Open educational September 15, 2008, from http://www.oreillynet.
practices and resources. OLCOS Roadmap 2012. com/pub/a/oreilly/tim/news/2005/09/30/what-is-
Salzburg Research EduMedia Group, Salzburg. Web-20.html?page=1
Retrieved September 15, 2008, from, available
OECD. (2001). OECD E-learning: The partner-
at: http://www.olcos.org/cms/upload/docs/ol-
ship challenge. Paris: OECD.
cos_roadmap.pdf
Owen, M., Grant, L., Sayers, S., & Facer, K.
Hudson, B. (2003). Promoting collaboration in
(2006). Social software and learning. FutureLabs.
and international online learning community. In-
Retrieved September 15, 2008, from http://www.
dustrial and Commercial Training, 35(3), 88–93.
futurelab.org.uk/research/opening_education.
doi:10.1108/00197850310470294
htm
Hvid, M., & Godsk, M. (2006, June). The plea-
Piaget, J. (1967). Six psychological studies. New
sure of e-learning - Towards aesthetic e-learning
York, NY: Vintage Books.
platforms. In Proceedings of the 12th International
Conference of European University Information Prensky, M. (2001). Digital natives, digital
Systems, University of Tartu & EUNIS, Tartu, immigrants. On the Horizon, 9(5). Retrieved
Estonia, (pp. 210-212). September 15, 2008, from http://www.marcpren-
sky.com/writing/Prensky%20-%20Digital%20
Kiser, K. (1999). 10 things we know so far about
Natives,%20Digital%20Immigrants%20-%20
online training. Training (New York, N.Y.), 36(11),
Part1.pdf
66–74.
Rice-Lively, M. L. (2000). Bordeless education
McGee, J. B., & Begg, M. (2008). What
at UT-Austin GSLIS. Texas Library Journal,
medical educators need to know about “Web
76(2), 58–60.
2.0”. Medical Teacher, 30(2), 164–169.
doi:10.1080/01421590701881673 Russell, T. L. (1999). The no significance differ-
ence phenomenon. Raleigh, NC: North Carolina
McLoughlin, C., & Lee, M. J. W. (2007, Decem-
State University Press.
ber). Social software and participatory learning:
Pedagogical choices with technology affordances Selwyn, N. (2008). An investigation of differences
in the Web 2.0 era. In ICT: Providing choices for in undergraduates’ academic use of Internet. Ac-
learners and learning. Proceedings ASCILITE tive Learning in Higher Education, 9(1), 11–22.
(pp. 664-675), Singapore. Retrieved September doi:10.1177/1469787407086744
15, 2008, from http://www.ascilite.org.au/confer-
Siemens, G. (2004). Connectivism: A learning
ences/singapore07/procs/mcloughlin.pdf
theory for the digital age. Retrieved September 15,
Mejias, U. (2005). A nomad’s guide to learning 2008, from http://www.elearnspace.org/Articles/
and social software. Retrieved September 15, connectivism.htm
2008, from http://knowledgetree.flexiblelearning.
net.au/edition07/download/la_mejias.pdf

803
Web 2.0 Technologies

Trondsen, E. (2006). Virtual worlds for learning or video conferencing. This terms covers a wide
and training. Menlo Park, CA: SRI Consulting set of applications and processes, such as Web-
Business Intelligence. based learning, computer-based learning, virtual
classrooms, and digital collaboration.
Virkus (2008). Use of Web 2.0 technologies in
Personal Learning Environments (PLE): A
LIS education: experiences at Tallinn University,
learning environment in which learners manage
Estonia. Program: Electronic Library and Infor-
their own learning by selecting, integrating and
mation Systems, 42(3), 262-274.
using various software tools and services. It takes
Wilson, J. (2006). 3G to Web 2.0? Can mobile advantages of Web 2.0 affordances such as col-
telephony become an architecture of participa- laborative information and knowledge sharing.
tion? Convergence: The International Journal of Social Capital: A cross-disciplinary concept
Research into New Media Technologies, 12(2), referring to the benefits of social networks and
229–242. doi:10.1177/1354856506066122 connections. Social capital is constructed and
maintained in the interaction between individuals
or groups. Social networks promote different types
of social capital: bonding –referring to horizontal
Key Terms
ties between individuals-, bridging – referring
Collaborative Learning: An educational to ties that cut across different communities- or
approach based the idea that learning is a natu- linking –referring to vertical ties.
rally social act. The learner actively constructs Social Software: Software that allows the
knowledge by formulating ideas into words, and creation of communities and resources in which
these ideas are built upon through reactions and individuals come together to learn, collaborate
responses of others. In other words, collaborative and build knowledge. It is also known as Web 2.0
learning is not only active but also interactive. It and it supports social interaction and collabora-
is a student-centered approach in which social tive learning. Current typical examples include
software tools are currently used for building and Flickr® and YouTube™ –as audiovisual social
sharing knowledge. software.
Connectivism: A learning theory for the digital Virtual Learning Environments (VLE):
era. It is based upon the idea that knowledge is A set of teaching and learning tools designed
networked and so the act of learning takes place to enhance a student’s learning experience by
inside virtual networks and communities through including computers and the Internet in the learn-
social interaction. It is a networked model of ing process. The principal components of a VLE
learning. package include curriculum mapping, student
E-Learning (electronic learning): Technolo- tracking, online support for both teacher and
gy-supported learning and delivery of content via student, electronic communication, and Internet
all electronic media. These may include Internet, links to outside curriculum resources. There are
intranets, computer-based technology, or interac- a number of commercial VLE software packages
tive television. They may also include the use of available, including Blackboard, WebCT, Lotus®
e-technology to support traditional methods of LearningSpace, and COSE.
learning, for example using electronic whiteboards

This work was previously published in Handbook of Research on E-Learning Applications for Career and Technical Education:
Technologies for Vocational Training, edited by V. C.X. Wang, pp. 779-790, copyright 2009 by Information Science Reference
(an imprint of IGI Global).

804
805

Chapter 3.9
SWELS:
A Semantic Web System
Supporting E-Learning
Gianluca Elia
University of Salento, Italy

Giustina Secundo
University of Salento, Italy

Cesare Taurino
University of Salento, Italy

Abstract the SWELS system is proposed by describing the


methodology adopted for organizing and modeling
This chapter presents a prototypal e-learning sys- its knowledge base, by illustrating its main func-
tem based on the Semantic Web paradigm, called tionalities, and by providing the design of the tool
SWELS (Semantic Web E-Learning System). The followed by the implementation choices. Finally,
chapter starts by introducing e-learning as an ef- future developments of SWELS will be presented,
ficient and just-in-time tool supporting the learning together with some remarks regarding the benefits
processes. Then a brief description of the evolution for the final user in using such system.
of distance learning technologies will be provided,
starting from first generation e-learning systems
through the current Virtual Learning Environ- Introduction
ments and Managed Learning Environments, by
underling the main differences between them and In a context of rapid environmental and technologi-
the need to introduce standards for e-learning with cal change, characterized by an increasing obsoles-
which to manage and overcome problems related to cence of knowledge, organizations need to accelerate
learning content personalization and updating. Fur- the renewal and to increase the effectiveness of their
thermore, some limits of the traditional approaches managerial competences. Such continuous change
and technologies for e-learning will be provided, is a determinant of continuous learning processes
by proposing the Semantic Web as an efficient and that calls for the capacity to organize at all levels of
effective tool for implementing new generation e- the organization new working processes that have
Learning systems. In the last section of the chapter, to be more knowledge intensive, multidisciplinary,
and collaborative.
DOI: 10.4018/978-1-60566-034-9.ch006

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
SWELS

This requires a profound rethinking of the dimensional vision of the knowledge available,
processes supporting the design, development, therefore users are obliged to follow the learning
and delivery of learning (McCrea et al., 2000) modules according to a linear path designed for a
in a way that the learning process becomes more generic learner). In addition, there is the need to
effective, just-in-time, and customized. optimize the processes related to learning resource
As a consequence, learning should not be a organization and aggregation, and the subsequent
passive activity which is only done when people access and reuse of such resources with respect
are in the educational institutions without knowing to a not scheduled learner profile.
how the knowledge is used in the real world. It Our focus here is on the creation of a Web-
should be a continuous and active process per- based learning environment that enables fast,
formed under a specified goal and situation where just-in-time and relevant learning. Indeed, cur-
the knowledge is really needed. Moreover, as huge rent Web-based solutions do not meet the above
amount of knowledge becomes available through mentioned requirements, and some pitfalls are for
Internet in the information society, it becomes example information overload, lack of accurate
possible for people to access the knowledge they information, and content that is not machine-
need when necessary. In such a circumstance, understandable.
the most important thing about having a lot of These limits suggest the application of Se-
knowledge is to know how to find the knowledge, mantic Web technologies (Barnes-Lee, 2000)
to be ready to understand and master the new to e-learning as means for implementing new
knowledge, and to create knowledge for future generation e-learning systems. The Semantic Web
use to close the loop of knowledge production technologies support the innovation process in
and consumption. For these reasons, the goal of a learning environment, exploiting the opportu-
education and learning should be augmented to nity to create and manage data that are machine
include training of learning capability and creativ- understandable and not only machine readable
ity of the learners (Mizoguchi, 2000). (Secundo et al., 2004).
Such considerations are some prominent An effective way to apply the Semantic Web
drivers of the e-learning. Since e-learning ap- approach to e-learning could be the use of the
plications are accessible from anywhere at any ontology backbone, which allows the ontology-
time, ICT-based learning environment has been based description of the learning materials
gaining increasing attention from the research (knowledge base), adding small semantic an-
community. notations to each learning object created (Nejdl,
In this context, the recently emerged (VLE) 2001). By using an ontology-based approach,
Virtual Learning Environments revealed them- learning resources can be easily organized into
selves very effective from the pedagogical point customized learning patterns and delivered on
of view, especially if they are compared with the demand to the learner, according to her/his profile
previous (CBT) Computer Based Training and and knowledge needs.
(WBT) Web Based Training systems. Moreover, such an approach allows to virtu-
However, VLE did not completely solve the ously combine the content description process
problems related to the organization and naviga- with the content navigation one: content descrip-
tion of the learning materials. Indeed, most of tion to easily identify the learning resources
the current Web-based learning solutions show required to achieve the desired learning goals;
some limits in accessing the right knowledge, as content navigation to minimize the required time
well as in the learning pattern navigation process for accessing the learning resources by adopt-
(given that they do not allow a complete and multi-

806
SWELS

ing the right approach of exploring the learning will be provided, starting from first generation
space. of e-learning systems (CBT) through the current
Therefore, according to this approach and to Virtual Learning Environments and MLEs (Man-
the alignment between e-learning and Knowledge aged Learning Environments), by underling the
Management we present an application of the main differences between them and the need to
KIWI approach called SWELS (Semantic Web introduce standards for e-learning with which
E-Learning System). This tool is a prototypal to manage and overcome problems related to
e-learning system based on the Semantic Web learning content personalization and updating.
paradigm which main functionalities are: Furthermore, some limits of the traditional ap-
proaches and technologies for e-learning will be
• The creation of an ontology-based view; provided, especially referring to the knowledge
• The semantic representation and visual- organization and access as well as to the learning
ization of learning modules (knowledge content navigation. At this point, the Semantic
base); Web will be proposed as an efficient and effective
• Learning modules (knowledge base) tool for implementing new generation e-learning
accessing; systems, since that the application of such tech-
• The visualization of the structure of the nology to e-learning provides an ontology-based
ontology. description and organization of learning materials
around small pieces of semantically annotated
Moreover, SWELS provides an innovative learning objects. In the last section of the chapter
functionality to learners--the opportunity to navi- the SWELS e-learning system is proposed. The
gate a domain ontology explicitly. By explicitly description of such an innovative solution starts
navigating the domain ontology, learners not with some insights on the methodology adopted
only have the direct access to the knowledge they for organizing and modeling the Knowledge
need inside the knowledge base, but also they Base of SWELS, then the main functionalities
are empowered in reaching the following goals of the e-learning system will be illustrated; fol-
(Secundo et al., 2004): lowing, the design of the tool together with the
implementation choices will be provided; finally,
1. The complete exploration of the knowledge future developments and some remarks will be
base, keeping the awareness and the visibility presented regarding the benefits for the final user
of the learning path performed to reach the (the learner) in using such system.
extracted knowledge;
2. The gradual, but deep, understanding of the
semantic structure of the knowledge domain E-Learning: A Technology
they are exploring, through the comprehen- Facilitating the
sion of the meanings of concepts and rela- Learning Processes
tions of the ontology.
Learning is a critical support mechanism for or-
The chapter starts by introducing e-learning ganizations to compete, not only from the point
as an efficient and just-in-time tool supporting of view of education, but also from the point of
the learning processes, arisen from the learning view of the New Economy (Drucker, 2000). The
requirements of the new, dynamically changing incredible velocity and volatility of today’s mar-
knowledge society. Then a brief description of kets require just-in-time methods for supporting
the evolution of distance learning technologies the need-to-know of employees, partners, and

807
SWELS

Table 1. Summary of problems and needs in education (Adapted from: Koper, 2004)

Dimension Problems/Needs
I. Changes in Societal 1. Current higher education infrastructure cannot accommodate the growing college-aged population and
Demands life-long learning enrolments, making more distance education programs necessary.
2. Knowledge and information are growing exponentially and Lifelong learning is becoming a competitive
necessity.
3. Education is becoming more seamless between high school, college, and further studies.
II. Changes in Learning 4. Instruction is becoming more learner-centred, non-linear, and self-directed.
Teaching process 5. There is an increasing need for new learning and teaching strategies that a) is grounded in new instruc-
tional design research and b) exploit the capabilities of technology.
6. Learning is most effective when learners are engaged in solving real-world problems; learning environ-
ments need to be designed to support this problem-centred approach.
7. Students demand more flexibility; they are shopping for courses that meet their schedules and circumstances.
8. Higher-education learner profiles, including online, information-age, and adult learners, are changing.
9. Academic emphasis is shifting from course-completion to competency.
10. The need for faculty development, support, and training is growing.
11. Instructors of distance courses can feel isolated.
III. Changes in Organization 12. There is a shift in organizational structure toward decentralization.
of Educational Institutions 13. Higher education outsourcing and partnerships are increasing.
14. Retention rates and length of time taken to completion concern administrators, faculty members, stu-
dents and tax payers.
15. The distinction between distance and local education is disappearing.
16. Faculty members demand reduced workload and increased compensation for distance courses.
17. Traditional faculty roles are shifting or unbundling.

distribution paths. overcome with e-learning; such problems and


Time, or the lack of it, is the reason given by needs are summarized and grouped on several
most businesses for failing to invest in learning. e-learning domain dimensions (Koper, 2004):
Therefore, learning processes need to be fast and But, what does e-learning mean? E-learning
just-in-time. Speed requires not only a suitable is the use of Internet technologies to create and
content of the learning material (highly specified, deliver a rich learning environment that includes
not too general), but also a powerful mechanism a broad array of instruction and information
for organizing such material. Also, learning must resources and solutions, the goal of which is to
be a customized on-line service, initiated by user enhance individual and organizational perfor-
profiles and business demands. In addition, it must mance (Rosenberg, 2006). E-learning is not just
be integrated into day-to-day work patterns and concerned with providing easy access to learning
needs to represent a clear competitive edge for resources, anytime, anywhere, via a repository
the business. In a few words, learning needs to of learning resources, but is also concerned with
be relevant to the (semantic) context of the busi- supporting such features as the personal defini-
ness of people and organizations (Adelsberger tion of learning goals, and the synchronous and
et al., 2001). asynchronous communication, and collaboration,
In this scenario, Web-based learning environ- between learners and between learners and instruc-
ments have been gaining increasing attention from tors (Kolovski et al., 2003). It aims at replacing
the research community, since e-learning appli- old-fashioned time/place/content predetermined
cations can represent real facilitator of the learn- learning with a just-in-time/at work-place/custom-
ing processes both in business and in academic ized/on-demand process of learning (Stojanovic
contexts. The following table (Table 1) underlines et al., 2001).
some problems and needs that can be effectively

808
SWELS

Traditional learning process could be char- E-Learning Systems:


acterised by centralised authority (content is From Virtual Learning
selected by the educator), strong push delivery Environments to Managed
(instructors push knowledge to students), lack of Learning Environments
a personalisation (content must satisfy the needs
of many), and the linear/static learning process In the 90s the primary impact of the Internet tech-
(unchanged content). The consequences of such nologies on distance learning was mainly on the
organisation on the learning are expensive, slow possibility of having different ways for aggregat-
and too unfocused (problem-independent) learn- ing and delivering learning content. Indeed, the
ing process. application of such technologies to the learning
But dynamically changed business environ- processes has introduced a set of opportunities
ment puts completely different challenges on and advantages: the possibility to generate and
learning process--fast, just-in-time (cheap) and rel- transport on the Web multimedia audio/video
evant (problem-dependent) learning, as mentioned flows at low costs (therefore promoting the dif-
above. This can be solved with the distributed, fusion of synchronous learning environments on
student-oriented, personalised, nonlinear/dynamic asynchronous ones); the use of standard tech-
learning process--e-learning. The principle behind nologies for information exchange that allow to
e-learning is that the tools and knowledge needed dynamically and effectively structure and navigate
to perform work are moved to the knowledge learning content; the possibility for learners to
workers--wherever and whoever they are. acquire knowledge and to continuously revise
In the recent years, new breeds of IS (In- and update it by adapting the learning environ-
formation System) known as LMS (Learning ment to their needs. In other words, Web-based
Management Systems) and LCMS (Learning learning environments shifted from stand-alone
Content Management Systems) are evolving to technologies towards highly integrated e-learning
enable learning in organisations (Brennan et al., and knowledge management infrastructures and
2001). In essence, LMS replace isolated and frag- tools enabling the creation of learning commu-
mented learning programmes with a systematic nities and supporting the collaboration between
means of assessing and raising competency and members and organizations. However, in the last
performance levels throughout the organisation, years Internet technologies failed in the process
by offering a strategic IS solution for planning, of creating and managing learning contents in a
delivering, and managing all learning events, way that they could be easily and dynamically
including both online and classroom-based learn- reused and updated. This because of the inability
ing (Greenberg, 2002). LMS are often coupled of trainers and learning managers to create learn-
with LCMS which facilitate the management ing materials that could be fast and easily adapted
and administration of the learning content for the to the learning needs of learners as well as to the
online learning prgrammes in the form of learning new ways of content delivery. Such considerations
objects (Brennan et al., 2001). have been driven the shift from first generation
e-learning systems, based on the delivery of Web-
based learning content and on the basic Internet

809
SWELS

standards, towards second generation ones, based • Low level formats for learning content;
on “ad-hoc” e-learning standards (Damiani et • Metadata for content description
al., 2002). • Personalization of content according to the
Second generation e-learning systems are learner profile and to the linguistic/social/
based on the creation of VLEs that have risen from cultural environment;
the integration between e-learning and knowledge • XML-based models and languages for
management solutions. The primary aim of VLEs structuring and describing dynamic learn-
is to allow people to share knowledge, interests ing patterns (i.e., Educational Modeling
and experiences, thereby encouraging the cre- Language);
ation of Virtual Learning Communities based on • Technologies and methodologies for in-
blended learning solutions in which face-to-face teroperability with Internet/Intranet deliv-
and virtual classroom meetings are combined ery infrastructures.
with Web-based learning patterns to provide to
learners a complete, interactive and effective Nowadays there are several key international
learning experience. Nevertheless, if from one players (including IEEE, IMS, ARIADNE, ADL
side VLEs revealed themselves very effective and AICC) that are focusing their efforts on the
from the pedagogical point of view (especially if issues of interoperability and reuse, resulting in a
considered in relation to the e-learning platform of multitude of standards that can be used for building
first generation), from the other side they showed interoperable learning systems. These attempts at
some limits as regards to the problem of learning building learning platforms for interoperability
content aggregation and organization, and the are mainly targeted to ease the need of LMSs for
subsequent access and reuse by learners with a adaptation to standards, but as a consequence,
non scheduled user profile. The classification and learners can be expected to gain more freedom. For
management of contents are the strength points of example, the goal of SCORM (Sharable Content
the so-called MLE. MLEs privilege the content Object Reference Model) is to provide a reference
design, creation, and management in respect to model for content that is durable (survives system
the content delivery infrastructure, considered as changes), interoperable (between systems), acces-
an element with which the content has to inter- sible (indexed and searchable) and reusable (able to
operate. The main goal of MLEs is to manage in be modified in different contexts and by different
an integrated way a complete system for analyz- tools). This will hopefully allow students to move
ing, developing, and evaluating competences, more freely between LMSs and even to combine
for scheduling and organizing learning patterns, several services from different LMSs.
for managing roles and virtual classrooms, for
tracking the learners and for final evaluation of
the competences reached (Secundo et al., 2004). Some Limits of Traditional
The complete separation proposed by MLEs Approaches and
between the management infrastructure and the Technologies For E-Learning
final output of the learning material is enabling
the development and the diffusion of standards Current approaches, models and technologies for
for e-learning in several applicative contexts e-learning introduce, on the other hand, several
(Lockwood et al., 2001): problems. First, most content providers have large
monolithic systems where adaptation to standards
• DRM (Digital Right Management) and pri- will not significantly change the underlying
vacy management; teacher-learner model. Students will be presented

810
SWELS

with material in a context often leading up to some a new learning framework is required, based on
form of (standardized) test. New and more interest- the key points mentioned above and opened to a
ing methods for learning--such as techniques for multitude of new services. In order to be effec-
collaboration, annotation, conceptual modeling, tive, it needs a powerful language for expressing
and so forth.--will not profit from such adapta- facts about resources and schemas that will allow
tion. Second, even though monolithic, closed machines as well as humans to understand how
or proprietary systems will be able to exchange these facts are related without relying on heu-
learning resources, course-like structures and keep ristics. Moreover, there is a need for expressing
track of students with the help of those standards, facts about remote (identifiable) objects without
they will need to go through yet another process accessing remote data stores.
of adaptation to the next big batch of agreements
on learning technologies, such as e.g. profiling
and tracking of student performance. Third, the Emerging Perspectives of
current perspective on metadata is too limited. E-Learning in the Semantic Age
Anyone who has something to say about a learning
resource should be able to do so. This includes In an e-learning environment, the learning content
learners, teachers and content contributors such should be oriented around small modules (the so
as authors and providers. Communicating this called learning objects) coupled with associated
metadata is equally important as it can help, direct semantics (the metadata) so that learners are able
or encourage others to actively participate and to find what they need in a specific moment and
learn. Proposed solutions, such as the adoption context. Furthermore, these modules should be
of SCORM, will result in learning resources (and related by a “dependency network” or “concep-
their metadata) that will reappear in different ver- tual Web” to allow individualised learning. Such
sions and formats rather than dynamically evolve a dependency network permits, for example, the
and improve (Naeve et al., 2001). learning objects to be presented to the learner in
In a few words, today many of the e-learning an orderly manner, with prerequisite material be-
systems available on market lack in specific func- ing presented first. Additionally, in an e-learning
tionalities for the creation and delivery of dynamic, environment, the learner should be able to add
modular learning paths that match the knowledge extra material and links (i.e., annotate) to the
needs in a contextualized (according to learner’s learning objects for their own benefit or for that
current activities) and individualized (according of later students. This framework lends itself to
to learner’s experiences, competences profiles, an implementation based on the Semantic Web,
learning history and personal preferences) way. incorporating cooperating software agents, which
This suggests a strong integration among additionally make use of appropriate Web services
e-learning and knowledge management func- to provide the functionality. The facilities the ap-
tionalities to define a rich learning environment plications based on these technologies can provide,
with wealth and variety of resources available include allowing e-learning content to be created,
just-in-time to learners, both through structured annotated, shared, and discussed, together with
and unstructured knowledge objects through in- supplying resources such lecture notes, student
teraction with other people (Elia et al., 2006). The portfolios, group projects, information pages,
key to success is therefore the ability to reduce discussion forums, and question-and-answer bul-
the cycle time for learning and to adapt “content, letin boards. Moreover, such applications allow
size and style” of learning to the learner and to the students to benefit from more interaction with
business. Therefore, to overcome such problems, their peers (for example, sharing resources found

811
SWELS

on the Web), as well as with the instructors and • Annotated. Every resource identifiable
tutors, by also providing an easy way for sharing by an URI can be annotated, with personal
and archiving information, whether of general notes and links by anyone.
interest or specific to a group project they are • Extended. In terms of content (structured,
involved in (Kolovski et al., 2003). by means of XML descriptors), permitting
T first generation WWW was a powerful tool multiple versions to exist. Indeed, successive
for research and education, but its utility is ham- editing of the content can be done via special
pered by the inability of the users to navigate easily RDF-schemas allowing private, group con-
the huge amount of sources for the information sensus or author-specific versions of a com-
they require. The Semantic Web is a vision to solve mon base document. The versioning his-
this problem. It is proposed that a new WWW tory will be a tree with known and unknown
architecture will support not only Web content, branches, which can be traversed with the
but also associated formal semantics (Barnes-Lee, help of appropriate versioning tools.
1998). The idea is that the Web content and the • Shared by, and communicated to, any-
related semantics (or metadata) will be accessed one who has expressed an interest in such
by Web agents, allowing these agents to reason content. RDF is application independent.
about the content and produce intelligent answers Since the metadata is expressed in a stan-
to user queries. The Semantic Web, in practice, dard format, which is independent of the
comprises a layered framework: an XML layer for underlying schemas, even simplistic ap-
expressing the Web content (the structure of data); plications can understand parts of complex
a RDF (Resource Description Framework) layer RDF graphs. If the learner’s favourite tool
for representing the semantics of the content (the does not support the corresponding sche-
meaning of data); an ontology layer for describing mas, it can at least present them in a rough
the vocabulary of the domain; and a logic layer to graph, table or whatever standard form it
enable intelligent reasoning with meaningful data has for describing resources and their prop-
(Stojanovic et al., 2001). erties. If more advanced processing soft-
Within an e-learning framework, the Semantic ware is available (such as logic engines),
Web provides the technology that allows a learning more advanced treatment of the RDF de-
object to be (Naeve et al., 2001): scriptions is possible.
• Certified. There is no reason why only big
• Described with metadata. Since a resource organizations should certify learning re-
can have uses outside the domains fore- sources. Individuals, such as teachers, may
seen by the provider, any given description want to certify certain content as a qual-
(metadata instance) is bound to be incom- ity learning resource that is well suited for
plete. Because of the distributed structure of specific learning tasks.
RDF, a description can be expanded or new
descriptions following new formats (sche- Apart from these uses, it is possible to invent
mas) can be added. This allows for creative new schemas describing structures, personaliza-
uses of content in new and unforeseen ways. tion, results from monitoring and tracking, pro-
Hence, one of the most important features cesses and interactions that can enrich the learning
of the current Web - the fact that anyone can environment in various ways.
link anything to anything--has been carried The key property of the Semantic Web ar-
over into RDF. chitecture (common-shared-meaning, machine-
processable metadata), enabled by a set of suitable

812
SWELS

agents, establishes a powerful approach to satisfy learning content provides efficient solutions to
the e-learning requirements: efficient, just-in-time, the following problems (Koper, 2004):
and task relevant learning. Learning material is
semantically annotated and for a new learning • The development of Web-based courses
demand it may be easily combined in a new that are flexible, problem-based, non-
learning course. According to his/her preferences, linear, incorporate multimedia and are
a learner can find useful learning material very adaptive to learner characteristics, is ex-
easily. The process is based on semantic querying pensive and extremely time-consuming. A
and navigation through learning materials, enabled semantic framework can help the course
by the ontological background. So, the Semantic developers in the structuring and integra-
Web can be exploited as a very suitable platform tion of the development work. In addition,
for implementing an e-learning system, because authoring and design support agents and
it provides all means for the ontology develop- tools could be created to help the develop-
ment, the ontology-based annotation of learning ers to do their jobs more effectively and
materials, as well as their composition in learning efficiently.
modules and proactive delivery of the learning • An explicit notation of learning content
materials through e-Learning portals. can preserve and share knowledge about
In the following table (Table 2), the most effective learning designs. It gives the
important characteristics (or pitfalls) of the possibility to build and share catalogues
traditional learning and improvements achieved of effective learning patterns that can be
using the e-learning environment are showed; communicated very precisely and can be
furthermore, a summary view of the possibil- adapted to other contexts, problems, and
ity to use the Semantic Web for realizing the content.
e-learning requirements is presented (Drucker, • Instantiation of an e-learning course in
2000; Stojanovic et al., 2001). current LMSs (Learning Management
An important aspect related to the use of Systems) can be a time-consuming job
Semantic Web in educational contexts is how that has to be repeated for every new run
to represent a course in a formal, semantic way of the course. One has to assign users,
so that it can be interpreted and manipulated by create groups, but also has to set-up the
computers as well as humans (i.e., the creation communication and collaboration servic-
and management of data that are machine under- es (e.g., discussion forums, workspaces,
standable and not only machine readable). This etc.) mostly by hand. A representation of a
process is known in the literature as “educational course that includes a specification of the
modeling.” A semantic model is developed using set-up of the services enables the automa-
a variety of methods: literature research, expert tion of this instantiation process.
group discussions, validation sessions, and so • When the representation of the learning
forth, and the result is described with a formal material includes a semantic, higher level
modeling language, like UML. The UML class description of the interactive processes
diagrams can be translated to RDF-Schema and/ that occur during the learning process,
or OWL Web Ontology Language, depending software agents can interpret these to sup-
on the richness of the model. XML-Schema’s port learners and staff in managing the
(XSD) and other semantic bindings like Topic workflow of activities in learning. These
Maps can also be generated from the UML mod- agents can also support the filtering of the
els (Koper, 2004). A semantic representation of

813
SWELS

Table 2. Differences between training and e-learning and main benefits of applying Semantic Web tech-
nologies to e-learning (Adapted from Drucker, 2000; Stojanovic et al., 2001)

Dimension Training e-Learning Semantic Web


Delivery Push--Instructor determines Pull--Learner de- Knowledge items (learning materials) are distributed on the
agenda termines agenda Web, but they are linked to commonly agreed ontologies.
This enables the creation of user-specific learning patterns,
by semantic querying for topics of interest.
Responsiveness Anticipatory--Assumes to know Reactionary --Re- Software agents on the Semantic Web may use com-
the problem sponds to problem monly agreed service language, which enables co-
at hand ordination between agents and proactive delivery of
learning materials in the context of actual problems.
The vision is that each user has his/her own personalised
agent that communicates with other agents.
Access Linear-- Pre-defined progression Non-linear --Al- User can describe situation at hand (goal of learning, pre-
of knowledge lows direct ac- vious knowledge) and perform semantic querying for the
cess to knowl- suitable learning material. The user profile is also accounted
edge in whatever for. Access to knowledge can be expanded by semantically
sequence makes defined navigation.
sense to the situ-
ation at hand
Symmetry Asymmetric --Training occurs Symmetric The Semantic Web offers the potential to become an integra-
as a separate activity --Learning occurs tion platform for all business processes in an organisation,
as an integrated including learning activities.
activity
Modality Discrete-- Training takes place Continuous Active delivery of information (based on personalised agents)
in dedicated chunks with defined --Learning runs in creates a dynamic virtual learning environment.
starts and stop the parallel loops
and never stops
Authority Centralized --Content is selected Distributed --Con- The Semantic Web will be as decentralised as possible. This
from a library of materials devel- tent comes from enables an effective co-operative content management.
oped by the educator the interaction of
the participants
and the educators
Personalization Mass produced-- Content must Personalized- A user (using personalised agent) searches for learning
satisfy the needs of many -Content is de- material customised for her/his needs. The ontology is the
termined by the link between user profile and needs, and characteristics of
individual user’s the learning material.
needs and aims to
satisfy the needs
of every user
Adaptiveness Static-- Content and organi- Dynamic-- Con- The Semantic Web enables the use of knowledge provided in
zation/taxonomy remains in tent changes con- various forms, by semantic annotation of content. Distributed
their original authored form stantly through nature of the Semantic Web enables continuous improvement
without regard to environmental user input, ex- of learning materials.
changes periences, new
practices, busi-
ness rules and
heuristics

appropriate resources to be used during highly desirable, since learners have not
the performance of an activity. the same learning pre-requisites, skills,
• Adaptation to individual learner charac- aptitudes or motivations. However, such
teristics (i.e., his/her learner profile) is adaptation can only be done realistically

814
SWELS

when the adaptation is wholly or at least This paper represents an extended version of a
partially automated (therefore, including previous publication that the authors G. Secundo,
descriptions of the conditions for adapta- A. Corallo, G. Elia G. Passiante (2004) published
tion). Otherwise, it becomes a very de- in the proceedings of the International Confer-
manding work for the learner and/or his/ ence on Information Technology Based Higher
her learning manager. Education and Training, May 29th--June 2th, 2004
• A semantic annotation of learning content Istanbul, Turkey.
enables and facilitates sharing and re-use The SWELS system is intended to be an in-
of learning objects (that is one of the major novative tool for knowledge acquisition and com-
objectives in the field of e-learning). This petence development of learners and knowledge
sharing and re-use is needed to make the workers that exploits Semantic Web technologies
content development process more effi- in order to provide an effective and useful support
cient. On the contrary, if learning objects to online learning processes. The system, indeed,
are not semantically represented, it might is conceived as a tool with which to potentially
be hard to find them on local or remote re- overcome the limits of the current e-learning ap-
positories, hard to integrate them into new plications in terms of learning content creation
contexts and--relating to the problem of in- and delivery, that is, the inability of existing tools
teroperability and learning object exchange to create dynamics and adaptive learning paths
among different LMSs--hard to interpret that match the learning profile of learners as well
and structure them in the correct way. as their knowledge needs. SWELS points out a
• An explicit semantic representation can proactive behaviour based on a matching process
serve as a means to create more advanced among the profile of the user, his/her interests
and complex, but consistent learning de- as well as his/her just-in-time choices during
signs than is possible without such a rep- the learning activities, and the learning content
resentation. This is a characteristic of any available in the knowledge base; as a consequence
language with semantic that enables one learning resources can be easily organized into
to write, read, rewrite and share meaning customized learning patterns and delivered on
(e.g., natural language). demand to the user.
Learning materials which SWELS refers to are
focused on “Change Management and Leadership”
SWELS: An e-learning knowledge domain, that has been modeled through
System Based on Semantic a domain ontology. Such an ontology contains the
Web Technologies list of concepts and semantic relations with which
to provide a semantic description of the learning
According to the alignment between e-learning and objects (text files, images, graphs, as well as mul-
knowledge management approaches, a prototypal timedia audio-video files) of the domain.
e-learning system based on the Semantic Web
paradigm has been implemented called SWELS.
Such system has been designed and developed at Knowledge Base
the eBMS (e-Business Management Section) () of Organization and Modeling
the Scuola Superiore ISUFI, University of Lecce
(Italy) and it is the result of a research activity Learning materials (i.e., the knowledge base)
under the KIWI project. are described by means of a domain ontology
that provides a semantic representation of con-

815
SWELS

Figure 1. A representation of knowledge base flexibility

tent, adding small semantic annotations to each edge base, therefore providing to learners the
learning resource. In particular, the knowledge possibility to have an explicit navigation of the
base modeling process can be organized in two domain ontology.
main steps: The two main advantages for the final users
are, from one hand a complete exploration of the
1. Definition of the knowledge base ontology. modeled knowledge base, which allows them to
The ontology definition consists in identify- have a total awareness of the available content,
ing the learning module structure and defines as well as the visibility of the performed learning
the abstract notions and vocabulary that will path to reach the required knowledge. So, learners
be available for the learner to conceptualize are conscious both of the total amount of knowl-
the learning modules. edge present in the knowledge base, and of the
2. Description of the knowledge resources. knowledge extracted till then and of knowledge
Knowledge items are tagged with one con- heritage to explore in the future. From the other
cept belonging to the ontology. In this way, hand, learners can understand, step by step, the
learner can identify each resource and, using semantic structure of the knowledge domain
the ontological relationships, he/she can they’re exploring by surfing the ontology, by
explore new resources tagged by different gradually being aware of the meanings of ontol-
domain concepts. ogy’s concepts and relations.
This approach to the knowledge base orga-
Such a description of learning content allows nization and modeling provides more flexibility
an effective organization of them in the knowl- for learners as regard to the learning content

816
SWELS

Figure 2. Use case diagram

access, since that they can explicitly browse the Creation of the
knowledge base and dynamically configure their Ontology-Based View
learning patterns (Figure 1).
When a learner accesses the SWELS, he/she has
to select the domain ontology and the relation by
Functional Description which creating the ontology view. When predicate
of SWELS is chosen, the tool generates the taxonomic repre-
sentation of the ontology, through a tree-structure
The interaction between the learner and the system (Figure 4).
can be represented through a use case diagram
that shows the main functionalities of the tool
(Figure 2). Semantic Representation
As the use cases show, in order to have access to and Visualization of
the dynamically created learning patterns, learners Learning Modules
have to perform a set of different steps.
The following state chart diagram describes the After the ontology view creation, the learner can
overall behaviour of the system by underling the generate his/her own personalized learning pat-
logic sequence of the states and the list of the state tern by browsing the concepts of the ontology.
transitions related to the user events according to By clicking on each concept, a list of elements
interaction between learner and system described will be shown:
before (see Figure 3).

817
SWELS

Figure 3. State chart diagram

• Concept definition (top of the page); the selected concept and the neighbour concepts.
• List of knowledge resources indexed on Such a graph is organized in the tab “Grafico”
the selected concept (body of the page-- (Graph) of the application, and represents the
with blank relation); semantic boundary of the concept (specifying
• List of knowledge resources indexed on the neighbour concepts, the semantic connec-
concepts linked to the selected concept tions and the direction of these connections). The
through one of the ontology relation- semantic boundary is illustrated through a radial
ships (body of the page – with specified layout (neighbourhood view) --as TGViz one (a
relation). Plug-in of Protégé) and Visualizer one (a Plug-in
of OntoEdit)--to give to the learner an explicit and
Such information is organized in the tab immediate representation of the ontology structure
“Documenti” (Knowledge Resources) as follows (Figure 6). It is important to note that, referring to
(Figure 5). each triple (subject, predicate, object), the direc-
tion of the arrows connecting two concepts goes
Ontology Structure Visualization from the subject to the object of the triple; this
allows the learner to have a unique interpretation
When a learner selects the specific concept which of the semantic map.
he/she is interested to, together with the semantic In this way, two different and complementary
visualization of the learning modules, SWELS representations of the domain ontology are avail-
generates also a knowledge map containing both able: the tree-structure (on the left of the page)

818
SWELS

Figure 4. Ontology view creation

Figure 5. Semantic representation and visualization of learning modules

819
SWELS

Figure 6. Ontology structure visualization

and the graph-structure (on the right). This choice Design of The Tool
allows a better understanding of the knowledge
domain (since that it provides two different ways Concerning the design of SWELS, we decided to
for representing the knowledge available in the adopt the MVC (Model-View-Controller) design
domain) and gives the learner the opportunity pattern (Figure 9) since that it allows enterprise
to select and extract the right learning resources applications to support multiple types of users
according to his/her “learning profile.” with multiple types of interfaces. By representing
the logic architecture of the system with such a
Learning Module Accessing “Three Tier” model, it is possible to keep sepa-
rated core business model functionalities from the
By clicking on the button “Risorsa” (Resource), presentation and the control logic that uses those
in the list of the knowledge resources indexed in functionalities. Such separation permits multiple
the tab “Documenti” (Figure 5), the learner has views to share the same enterprise data model,
the direct access to the chosen learning module; which makes supporting multiple clients easier to
in this way, the selected learning module will be implement, test, and maintain (Sun Microsystems,
launched in a new browser window and he/she Inc., 2002).
can attend it autonomously (Figure 7). According to the Three Tier model adopted for
Moreover, the learner can also access metadata the SWELS design, the first diagram proposed is
describing knowledge resources, by clicking on the package diagram that shows developed class
resource name. In this way, Dublin Core metadata packages and the dependencies among them
(Dublin Core Metadata Initiative, 2006) will be (Fowler et al., 1999) (Figure 10).
shown (Figure 8).

820
SWELS

Figure 7. Learning module accessing

Going on in the description of the SWELS • Jakarta Struts, an open source framework
design, following are shown the class diagrams for creating Java Web applications that uti-
describing the types of the objects in the system lize a MVC architecture. The framework
and the various kinds of static relationships that gives three key components: a “request”
exist among them. In particular, we propose the handler provided by the application devel-
class diagram related to the ontology view cre- oper that is mapped to a standard URI, a
ation (Figure 11), the class diagram related to the “response” handler that transfers control to
semantic visualization of the learning modules another resource which completes the re-
(Figure 12), and the class diagram related to the sponse, a tag library that helps developers
ontology structure visualization (Figure 13). create interactive form-based applications
with server pages (The Apache Software
Technological Issues Foundation, 2006).
• Oracle9iAS Toplink, an ORM (Object
With regard to implementation choices, SWELS Relational Mapping) framework for imple-
is a J2EE Web-based application, developed ac- menting the ‘Model’ layer that is free only
cording to the MVC (Model-View-Control) pat- for non-commercial applications.
tern (which implies together the use of Servlets as
well as JSPs technologies), by using two suitable Furthermore, the ontology is codified in RDFS
frameworks: and is stored in a relational database. The DBMS

821
SWELS

Figure 8. E-learning metadata

is Oracle 9i; the relational database schema for the Empirical Evidence
application is the following (Figure 15):
Finally, the standard for e-learning metadata During the exciting experience of conceptualization,
is Dublin Core; the implementation of SCORM design and implementation phases of SWELS, some
1.2 is a work-in-progress. attempts to validate the effectiveness of the whole
These implementation choices give the tool a approach were made. Specifically, we refer to a pro-
high level of flexibility and scalability; indeed, cess realized on empirical evidence basis to acquire
it can be used on several knowledge bases by some insights for improving the overall system. This
developing a specified domain ontology and by process is articulated in two test phases--the ‘alpha
exploiting the potentialities of ORM framework. test’ phase and the ‘beta test’ phase.
The “Alpha test” phase was performed by the
team involved in the implementation of the system
itself. From one side, software developers tested

Figure 9. The MVC (Model-View-Controller) design pattern

822
SWELS

Figure 10. Package diagram

Figure 11. Class diagram for the ontology view creation

823
SWELS

Figure 12. Class diagram for the semantic visualization of the learning modules

many times and under different conditions each (platform and contents) as an additional learn-
functionality of the system. They also executed ing tool during the attendance of the module on
a general test for the overall system to evaluate “Change Management and Leadership.” After
its robustness and the coherence of data manage- one week, at the end of this module, a face-to-
ment and tracking systems. From the other side, face discussion meeting was organized with the
the subject matter experts, after the design phase participation of an outstanding professor in this
and the coordination of teams involved into the field. Final impressions of master participants
content creation process, executed a double-layer about SWELS were extremely positive, because
control: one for the exactness of how each topic they represented a sort of personal assistant to
was expressed, and one for the semantic link deepen and clarify some difficult concepts of the
about different topics. Both tests revealed a set module and, above all, to have a systemic vision
of enhancements that have been implemented in of the general topic.
the new version of SWELS.
The “Beta test phase” was performed by Future Developments of Swels
involving a group of 20 students attending an
International master program at the eBMS of The next steps that we aim to develop in the
Scuola Superiore ISUFI,. They used SWELS future are:

824
SWELS

Figure 13. Class diagram for the ontology structure visualization

• The implementation of an interactive radi- (i.e., assessment, difficulty level, etc.) and
al layout layer (i.e., an interactive graphi- the learner profile expressed in terms of
cal interface to activate the “conceptual interests and knowledge gaps (by tracking
semantic boundary”). In our opinion, this the learning pattern dynamically created
improvement could make SWELS more by learners).
effective, since that learners can immedi- • The integration of SWELS into a LMS,
ately access learning modules, by directly that means the development of a personal
clicking on the concepts shown in the ra- learning agent integrated into a LMS that
dial graph. proactively configures and recommends
• The ontological representation of two fur- personalized learning paths to the learners
ther learning dimensions: the typology and according to their learning profile.
characteristics of the learning resource

825
SWELS

Figure 14. Integrating struts framework in MVC architectures

Finally, a large scale experimentation of the through simple and complex semantic
system should be organized in order to evaluate assertions;
the effectiveness of SWELS and, more in general, • A more direct access to the knowledge do-
of the learning approach embedded in SWELS. main by explicitly navigating and browsing
the ontology map;
• A more flexible structure of the learning
Conclusion materials that can be easily recombined and
described for other purposes and learning
SWELS platform is the result of applying Semantic goals in other knowledge domains.
Web technologies to e-learning. Such a strategic
choice allows learners and knowledge workers to
increase the effectiveness of their learning process In our beliefs, this approach could provide a new
since it enables a personalized access to learning way in which students learn, since it is based on a
materials as well as a complete and deep under- learner-centric strategy characterized by:
standing of the knowledge domain.Indeed, from the
point of view of the final users, the main benefits • The role of personal tacit knowledge and
of using SWELS are: learning experiences as the starting point
and the knowledge background of future
• The explicitation of tacit knowledge con- learning patterns;
tained in the knowledge base conceptualiza- • A solution-oriented approach for creating
tion process and held in the minds of subject just-in-time new learning patterns;
matter experts as well as domain designers; • The possibility to fulfill the personal skill
• The systematization of knowledge through gap by actively participating and self-ex-
an explicit indexing of knowledge resources ploring the knowledge base;

826
SWELS

Figure 15. Relational schema of the database implemented

• A stimulus to the “knowledge curiosity” of • A dynamic creation of learning paths, start-


learners in deepening specific knowledge ing from different knowledge resources
domains; semantically annotated, according to the
• The development of knowledge, skills and learner interests and knowledge needs, ex-
attitudes conceived as capacity for effective pressed by them in real time.
actions and problem solving;
• A set of customized training curricula con-
sistent to learners’ needs, their own time
and place, without compromising its effec-
tiveness (Keegan, 2000);

827
SWELS

References Fowler, M., & Scott, K. (Eds.). (1999). UML Dis-


tilled Second Edition--A Brief Guide to the Stan-
Adelsberger, H., Bick, M., Körner, F., & Paw- dard Object Modelling Language. A. Wesley.
lowski, J. M. (2001). Virtual education in business
information systems (VAWI)--Facilitating col- Greenberg, L. (2002). LMS and LCMS: What’s
laborative development processes using the Essen the Difference? Learning Circuits, ASTD.
learning model. In H. Hoyer (Ed.), 20th ICDE Keegan, M. (Ed.). (2000). e-Learning, The engine
World Conference on Open Learning and Distance of the knowledge economy. Morgan Keegan &
Education. The Future of Learning – Learning for Co.
the Future: Shaping the Transition.
Kolovski, V., & Galletly, J. (2003). Towards e-
Barnes-Lee, T. (1998), Semantic Web Roadmap. learning via the Semantic Web. In B. Rachev, &
W3C Design Issues. A. Smrikarov (Eds.), In Proceedings of the 4th
Barnes-Lee, T. (2000). What the Semantic Web International Conference on Computer Systems
can represent. and Technologies--CompSysTech’2003 (pp. 591
- 596). ACM New York, NY, USA.
Brennan, M., Funke, S., & Andersen, C. (2001).
The learning content management system--a new Koper, R. (2004). Use of the Semantic Web to
e-Learning market segment emerges. IDC White solve some basic problems in education. Journal
Paper., of Interactive Media in Education, (6): 1–23.

Damiani, E., Corallo, A., Elia, G., & Ceravolo, Lockwood, F., & Gooley, A. (Eds.). (2001). Inno-
P. (2002, November). Standard per i learning vation in open and distance learning - successful
objects: Interoperabilità ed integrazione nella development of online and Web-based learning.
didattica a distanza. Paper presented at the In- London, Kogan Page.
ternational Workshop eLearning: una sfida per McCrea, F., Gay, R. K., & Bacon, R. (2000).
l’Università - Strategie Metodi Prospettive, Riding the big waves: a white paper on the B2B
Milan, Italy. e-Learning Industry. San Francisco/Boston/New
Drucker, P. (2000). Need to know: Integrating e- York/London: Thomas Weisel Partners LLC.
learning with high velocity value chains. Delphi Mizoguchi, R. (2000). IT revolution in learning
Group White Paper. technology, In Proceedings of SchoolNet 2000,
Dublin Core Metadata Initiative. (2006). Dublin Pusan, Korea (pp. 46-55).
Core Metadata Terms., Retrieved on May 26, Naeve, A., Nilsson, M., & Palmer, M. (2001),
2003, from http://dublincore.org E-learning in the semantic age. CID, Centre For
Elia, G., Secundo, G., & Taurino, C. (2006), To- User Oriented It Design. Stockhom, Sweden.
wards unstructured and just-in-time learning: The Retrieved on September 30, 2006
“Virtual eBMS” e-Learning system. In A Méndez- Nejdl, W. (2001). Learning repositories--Tech-
Vilas, A. Solano-Martin, J. Mesa González, & nology and context. In A. Risk (Ed.), ED-Media
J.A. Mesa González (Eds.), m-ICTE2006: Vol. 2001 World Conference on education multimedia,
2. Current Developments in Technology-Assisted Hypermedia and Telecommunications: Vol. 2001,
Education (pp. 1067-1072). FORMATEX, Bada- N. 1.
joz, Spain.

828
SWELS

Rosenberg, M. J. (Ed.). (2006). Beyond E-Learn- Stojanovic, L., Staab, S., & Studer, R. (2001,Oc-
ing: Approaches and technologies to enhance tober 23-27). eLearning based on the Semantic
knowledge, learning and performance. Pfeiffer. Web. In W. A. Lawrence-Fowler & J. Hasebrook
(Eds.), In Proceedings of WebNet 2001--World
Secundo, G., Corallo, A., Elia, G., & Passiante, G.
Conference on the WWW and Internet, Orlando,
(2004). An e-Learning system based on semantic
Florida. (pp.1174-1183). AACE.
Web supporting a Learning in Doing Environment.
In Proceedings of International Conference on Sun Microsystems, Inc. (2002). Java Blueprints-
Information Technology Based Higher Education -Model-view-controller.
and Training--ITHET 2004. IEEE XPlore,
The Apache Software Foundation. (2006). Apache
Struts.

This work was previously published in Semantic Knowledge Management: An Ontology-Based Framework, edited by A. Zilli;
E. Damiani; P. Ceravolo; A. Corallo and G. Elia, pp. 120-145, copyright 2009 by Information Science Reference (an imprint
of IGI Global).

829
830

Chapter 3.10
Web Services Discovery with
Rough Sets
Maozhen Li
Brunel University, UK

Bin Yu
Level E Limited, UK

Vijay Sahota
Brunel University, UK

Man Qi
Canterbury Christ Church University, UK

ABSTRACT INTRODUCTION

Web services are emerging as a major technology Web services are emerging as a major technology
for building service-oriented distributed systems. for developing service-oriented distributed
Potentially, various resources on the Internet can systems. Potentially, many resources on the
be virtualized as Web services for a wider use by Internet or the World Wide Web can be virtualized
their communities. Service discovery becomes an as services for a wider use by their communities.
issue of vital importance for Web services applica- Service discovery becomes an issue of vital
tions. This article presents ROSSE, a Rough Sets importance for Web service applications. As
based Search Engine for Web service discovery. shown in Figure 1, discovered services can either
One salient feature of ROSSE lies in its capabil- be used by Web service applications or they
ity to deal with uncertainty of service properties can be composed into composite services using
when matching services. A use case is presented workflow languages such as BPEL4WS (Andrews
to demonstrate the use of ROSSE for discovery of Curbera, Dholakia, Goland, Klein, Leymann et al.,
car services. ROSSE is evaluated in terms of its 2003). UDDI (Universal Description, Discovery
accuracy and efficiency in service discovery. and Integration, http://www.uddi.org) has been

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Web Services Discovery with Rough Sets

proposed and used for Web service publication name, service category, and aspects related to the
and discovery. However, the search mechanism quality of a service. In addition to OWL-S, another
supported by UDDI is limited to keyword matches. prominent effort on Semantic Web services is
With the development of the Semantic Web WSMO (Roman, Keller, Lausen, Bruijn, Lara,
(Berners-Lee, Hendlet, & Lassila, 2001), services Stollberg et al., 2005), which is built on four key
can be annotated with metadata for enhancement concepts—ontologies, standard WSDL based
of service discovery. The complexity of this Web services, goals, and mediators. WSMO
metadata can range from simple annotations, to stresses the role of a mediator in order to support
the representation of more complex relationships interoperation between Web services.
between services based on first order logic. However, one challenging work in service
One key technology to facilitate this semantic discovery is that service matchmaking should be
annotation of services is OWL-S (Martin, able to deal with uncertainty in service properties
Paolucci, McIlraith, Burstein, McDermott, when matching service advertisements with
McGuinness et al., 2004), an OWL (Web Ontology service requests. This is because in a large-scale
Language, http://www.w3.org/TR/owl-features/ heterogeneous system, service publishers and
Reference) based ontology for encoding properties requestors may use their pre-defined properties
of Web services. OWL-S ontology defines a to describe services, for example, in the form
service profile for encoding a service description, of OWL-S or WSMO. For a property explicitly
a service model for specifying the behavior of a used in one service advertisement, it may not be
service, and a service grounding for invoking the explicitly used by another service advertisement
service. Typically, a service discovery process within the same service category. As can be seen
involves a matching between the profile of a from Table 1, the property P1 used by the service
service advertisement and the profile of a service advertisement S1 does not appear in the service
request using domain ontologies described in advertisement S2. When services S1 and S2 are
OWL. The service profile not only describes the matched with a query using properties P1, P2 and
functional properties of a service such as its inputs, P3, the property P1 becomes an uncertain property
outputs, pre-conditions, and effects (IOPEs), but when matching S2. Similarly, the property P3
also non-functional features including service becomes an uncertain property when matching

Figure 1. A layered structure for service-oriented systems

831
Web Services Discovery with Rough Sets

S1. Consequently, both S1 and S2 may not be ROSSE DESIGN


discovered because of the existence of uncertainty
of properties even though the two services are ROSSE considers input and output properties
relevant to the query. individually when matching services. For the sim-
It is worth noting that properties used in plicity of expression, input and output properties
service advertisements may have dependencies, used in a service request are generally referred
for example, both P1 and P3 may be dependent to as service request properties. The same goes
properties of P2 when describing services S1 and to service advertisements.
S2 respectively. Both S1 and S2 can be discovered Figure 2 shows ROSSE components. The
if P1 and P3 (which are uncertain properties in Irrelevant Property Reduction component takes
terms of the user query) can be dynamically a service request as an input (step 1), and then
identified and reduced in the matching process. To it accesses a set of advertised domain services
increase the accuracy of service discovery, a search (step 2) to remove irrelevant service properties
engine should be able to deal with uncertainty of using the domain ontology (step 3). Reduced
properties when matching services. properties will be marked in the set of advertised
In this article, we present ROSSE, a Rough domain services (step 4). Once invoked (step 5),
Sets (Pawlak, 1982) based Search Engine for Web the Dependent Property Reduction component
service discovery. One salient feature of ROSSE accesses the advertised domain services (step 6)
lies in its capability to deal with uncertainty in to discover and reduce indecisive properties which
service properties (attributes) when matching will be marked in advertised domain services (step
service advertisements with service requests. 7). Invoked by the Dependent Property Reduction
Experiment results show that ROSSE is more component (step 8), the Service Matching and
effective in service discovery than existing Ranking component accesses the advertised
mechanisms such as UDDI keyword matching domain services for service matching and ranking
and OWL-S matchmaking. (step 9), and finally it produces a list of matched
The remainder of this article is organized as services (step 10).
follows. The ROSSE Design section presents the In the following sections, we describe in depth
design details of ROSSE. The ROSSE Case Study the design of ROSSE components for service
section gives a case study to demonstrate the matchmaking and discovery. Firstly, we introduce
use of ROSSE for discovery of car services. The Rough sets for service discovery.
ROSSE Implementation and Evaluation section
evaluates ROSSE from the aspects of accuracy Rough Sets for Service Discovery
and efficiency in service discovery. The Related
Word section discusses some related work, and Rough sets method is a mathematic tool that can
the Conclusion and Future Work section concludes deal with uncertainty in knowledge discovery. It
the article. is based on the concept of an upper and a lower

Table 1. Two service advertisements with uncertain service properties

832
Web Services Discovery with Rough Sets

approximation of a set as shown in Figure 3. For X = {x ∈ U : [ x]P  X ≠ ∅} (2)


a given set X, the yellow grids (lighter shading)
A

represent its upper approximation, and the green For a property used by a service request p ∈
grids (darker shading) represent its lower ap- PA, we have
proximation. We introduce Rough sets for service
discovery in the following way. • ∀x ∈ X , x definitely has property p.
• ∀x ∈ X , x possibly has property p.
Let • ∀x ∈ U - X , x absolutely does not have
• Ω be a domain ontology. property p.
• U be a set of N service advertisements, U =
{s1, s2, ..., sN}, N ≥ 1. The use of “definitely,” “possibly” and
• P be a set of K properties used in the N “absolutely” are used to encode properties that
service advertisements, P = {p1, p2, ..., pK}, cannot be specified in a more exact way. This
K ≥ 2. is a significant addition to existing work, where
• PA be a set of M properties used in service discovery of services needs to be encoded in a
advertisements which are relevant to a precise way, making it difficult to find services
service request R within the domain ontology which have an approximate match to a query.
Ω, Advertised domain service properties may be
• PA = {pA1, pA2, ..., pAM}, PA ⊆ P, M ≥ 1. irrelevant (having no effect on service matching)
• X be a set of service advertisements relevant or relevant (having an impact on service matching).
to the service request R, X ⊆ U. Certain properties used by advertised services
• X be the lower approximation of the set X. may be redundant which can be reduced without
• X be the upper approximation of the set X. losing essential classificatory information. The
concept of the reduct is fundamental for Rough
According to the Rough sets theory, we sets theory (Winiarski, 2001). Service property
have reduction can be considered as a process of finding
a smaller (than the original one) set of properties
X = {x ∈ U : [ x]P ⊆ X } (1) with the same or close classificatory power as
A

Figure 2. ROSSE components

833
Web Services Discovery with Rough Sets

Figure 3. Approximation in Rough sets

the original set. For a service query, the most For each property used in a service request,
relevant properties of advertised services can be the Irrelevant Property Reduction component
determined after property reduction. uses Algorithm 1 to remove irrelevant properties
from advertised services. For those properties
Reducing Irrelevant Properties used in service advertisements that have a
nomatch result, they will be treated as irrelevant
When searching for a service, a service request properties. Service advertisements are organised
may employ some properties which are irrelevant as service records in a database. Properties are
to the properties used in a service advertisement organised in such a way that each property uses
within one domain ontology. These irrelevant one column to ensure the correctness in the
properties used in service advertisements should following reduction of dependent properties. As
be removed before the service matchmaking a property used in one service advertisement
process is performed. might not be used in another one, some properties
may have empty values. For a service request, a
Let property with an empty value in a service record
• pR be a property used in a service request. becomes an uncertain property. If a property in an
• pA be a property used in a service advertise- advertised service record is marked as nomatch,
ment. the column associated with the property will be
marked as nomatch. As a result, all properties
Following the work proposed in (Paolucci, within the column including uncertain properties
Kawamura, Payne, & Sycara, 2002), we define (i.e., properties with empty values) will not be
the following relationships between pR andpA: considered in service matchmaking.

• exact match, pR and pA are equivalent or pR Reducing Dependent Properties


is a subclass of pA.
• plug-in match, pA subsumes pR. Properties used by service advertisements may
• subsume match, pR subsumes pA. have dependencies. Dependent properties are
• nomatch, no subsumption between pR and indecisive properties which have no effect on
pA.

834
Web Services Discovery with Rough Sets

Algorithm 1. Reducing irrelevant properties from service advertisements


1: for each property pA used in service advertisements
2: for all properties used in a service request
3: if pA is nomatch with any pR
4: then pA is marked with nomatch;
5: end if
6: end for
7: end for

service matching. Building on the work proposed Specifically, service advertisements with the
in (Jensen, Shen, & Tuson, 2005), we designed maximum number of nonempty property values
Algorithm 2 to reduce dependent properties from are used in the algorithm as targets to find
advertised services. indecisive properties. The targeted services can
Let still be uniquely identified without using these
indecisive properties. All possible combinations of
• Ω, U, P, PA be defined as in the Rough Sets individual indecisive properties are checked with
for Service Discovery section. an aim to maximally remove indecisive properties
• PAD be a set of LD decisive properties for which may include uncertain properties whose
identifying service advertisements relevant values are empty. In the mean time, the following
to the service request R in terms of Ω, service discovery process is speeded up due to
• PAD = { p AD1, p AD2,..., p AL
D
D
}, PAD ⊆ PA, LD ≥ 1. the reduction of dependent properties.
• PAIND be a set of LIND indecisive properties for
identifying service advertisements relevant Computing Match Degrees
to the service request R in terms of Ω,
• PAIND = { p AIND , IND,..., p AL
1 pA2
IND
IND
}, PAIND ⊆ PA, LIND ≥ The Service Matching and Ranking component
1. uses the decisive properties to compute the match
• IND() be an indiscernibility relation. degrees of advertised services related to a service
• f be a mapping function from a property to request.
a service advertisement. Let

Then • Ω, U, P, PA be defined as in the Rough Sets


for Service Discovery section.
IND ( PAIND ) =
• PR be a set of M properties used in a service
{( x, y ) ∈ U : ∀p AiIND ∈ PAIND , f ( x, p AiIND ) =
request R. PR = {PR1, PR2,..., PR3}, M ≥ 1.
f ( y, p AiIND )} • PAD be a set of LD decisive properties for
(3) identifying service advertisements relevant
to the service request R in terms of Ω,
PAD =PAIND - PA (4) • PAD = { p AD1, p AD2,..., p AL
D
D
}, LD ≥ 1.
• m(pRi, pAj) be a match degree between a
For a service request, the Dependent Property property PRi and a property PAj in terms of
Reduction component uses Algorithm 2 to find Ω, PRi ∈ PR,1 ≤ i ≤ M, PAj ∈PAD , 1 ≤ j ≤ LD.
the decisive properties in service advertisements. • v(PAj) be a value of the property PAj, PAj∈PAD,
1 ≤ j ≤ LD.

835
Web Services Discovery with Rough Sets

Algorithm 2. Reducing dependent properties from advertised services


S is a set of service advertisements with the maximum number of nonempty
property values relevant to a service request;
PA is a set of properties used by the S set of service advertisements;
PAD is a set of decisive properties, PAD ⊆ PA;
PAIND is a set of individual indecisive properties, PAIND ⊆ PA ;
PAIND_Core is a set of combined indecisive properties,
PAIND_Core ⊆ PAIND;
PAD = Ø; PAIND = Ø; PAIND_Core = Ø;
1: for each property p∈ PA
2: if p is an indecisive property for identifying the S set of services
3: then
4: add p into PAIND;
5: PAIND_Core = Ø;
6: add p into PAIND_Core;
7: end if
8: end for
9: for i=2 to sizeof(PAIND)-1
10: calculate all possible i combinations of the properties in PAIND;
11: if any combined i properties are indecisive properties for identifying
the S set of services
12: then
13: PAIND_Core = Ø;
14: add the i properties into PAIND_Core;
15: continue;
16: else if any combined i properties are decisive properties
17: then break;
18: end if
19: end for
20: PAD = PA-PAIND_Core;
21: return PAD;

• S(R, s) be a similarity degree between a be given a match degree of 100% if it has an


service advertisement s and the service exact match relationship with a property used in
request R, s ∈ U. a service request. A match degree of 50% will
be given if it has a plug-in relationship with a
Algorithm 3 shows the rules for calculating service request property and the relationship is
a match degree between a property used in a out of five generations. Similarly, a property used
service request and a property used in a service in a service advertisement will be given a match
advertisement. A decisive property with an degree of 50% if it has a subsume relationship with
empty value has a match degree of 50% when a service request property and the relationship is
matching each property used in a service request. out of three generations.
A property used in a service advertisement will

836
Web Services Discovery with Rough Sets

Algorithm 3. The rules for calculating match degrees between properties used in service requests and
service advertisements respectively
1: for each property pAj ∈ PAD, v(pAj) ≠ NULL
2: for each property pRi ∈ PR
3: if pAj is an exact match with pRi
4: then m(pRi, pAj) = 1;
5: else if pAj is a plug-in match with pRi
6: then if pRi is the kth subclass of pAj and 2≤k≤5
7: then m(pRi, pAj) = 1-(k-1)×10%;
8: else if pRi is the kth subclass of pAj and k>5
9: then m(pRi, pAj) = 0.5;
10: end if
11: else if pAj is a subsume match with pRi
12: then if pAj is the kth subclass of pRi and 1≤k≤3
13: then m(pRi, pAj) = 0.8-(k-1)×10%;
14: else if pAj is the kth subclass of pRi and k>3
15: then m(pRi, pAj) = 0.5;
16: end if
17: end if
18: end for
19: end for
20: for each property pAj ∈ PAD, v(pAj) = NULL
21: for each property pRi ∈ PR
22: m(pRi, pAj) = 0.5;
23: end for
24: end for

Each decisive property used for identifying Using the formula (5), ROSSE calculates a
service advertisements has a maximum match matching degree for each service advertisement
degree when matching all the properties used in related to a service request. The similarity
a service request. S(R, s) can be calculated using degrees are used to produce a lower and an upper
formula (5). approximation set of discovered services.
LD M
S ( R, s ) = ∑ ∑ max(m( p Ri , p Aj )) LD
j =1 i =1 ROSSE CASE STUDY
(5)
In this section, we present a use case of ROSSE
to discover vehicle services. Figure 4 shows the

837
Web Services Discovery with Rough Sets

Figure 4. Ontolgogies used in the search scenario


Query: Car, mini cooper, ABS, UK, BMW, Exhaust 1.0, Price,

ontologies used in this scenario defining the Building a Decision Table


classifications of vehicles, objects, exhausts, loca-
tions, configurations, brands respectively. Two A service decision table is used to compute
ontologies are used to classify configurations of dependent properties among services. As the
vehicles represented respectively by e1-e5 and g1- number of services registered with ROSSE can
g4. Relevant vehicle services are registered with be tremendous, the decision table is constructed
ROSSE. In the following sections, we describe by sampling registered services. For a specific
how services are matched in terms of the follow- query, ROSSE randomly selects a certain number
ing query to search for car services that sell red of services records. A service record is selected as
BMW mini coopers that have an exhaust of 1.0, long as one of its properties has a valid relation-
and are configured with ABS, manufactured in ship with a property used in a service query. The
the UK. Price information is also provided by relationship can be exact, plug-in or subsume as
the car services. defined in algorithm 1 which is described in the
Reducing Irrelevant Properties section.

838
Web Services Discovery with Rough Sets

Table 2. A segment of the decision table used for Table 3. Computed dependent properties
discovery of car services

Table 2 shows a segment of the decision table relevant to the service. Such a property might
with 13 service records for discovery of car be an inherent property of the service. ROSSE
services. As can be seen from Table 2, properties deals with properties marked with 0 as uncertain
of advertised services that are relevant to the car properties when matching services.
service query are f6, g2, d3, f3, c2, b2, b3, d2, c1,
e1/g1, d1, b6. If a property in a service record is Computing Dependent Properties
marked with 1, this means that the property is used
by the service in its advertisement. For example, Once a service decision table is constructed, the
the service S1 has properties of f6, g2, d3, f3, c2, next step is to compute dependent properties.
d1, and b6 in its advertisement. A property marked Using the algorithm 2 presented in the Reduc-
with 0 in a service record means that the service ing Dependent Properties section, properties g2,
does not have the corresponding property in its d3, f3, and c2 are indecisive properties which
advertisement, for example, properties such as b2, are reduced from the decision table in matching
b2, d2, c1, and e1/g1 are not used by the service services as shown in Table 3. Table 4 shows the
S1 for advertisement. However, it should be noted segment of the decision table without dependent
that a property marked with 0 in a service record properties.
does not necessarily mean this property is not

Table 4. The segment of the decision table without


Table 5. Computation of matching degrees
dependent properties

839
Web Services Discovery with Rough Sets

Computing Match Degrees ROSSE IMPLEMENTATION AND


EVALUATION
Decisive properties are used for computing the
similarities between an advertised service and a ROSSE is implemented with Java on a Pentium IIII
service request. For each decisive property used 2.6G with 512M RAM running Red Hat Fedora
in a service advertisement and a property used in Linux 3. Figure 5 shows the homepage of ROSSE.
the service query, a maximum matching degree It has two registries for service registration, a
can be computed using ontologies defined in UDDI registry and an OWL registry. The UDDI
Figure 4. Table 5 shows the matching degrees of registry is used to register services with WSDL
the decisive properties used in the exemplified 13 interfaces, and the OWL-S registry is used to reg-
service records. It should be noted that both e1 and ister services with OWL-S interfaces. The UUID
g1 refers to the same property Configuration, but of a WSDL service registered with the UDDI
they use different ontology definitions as shown in registry is used to uniquely identify semantic
Figure 4. The matching degree of Configuration annotation records of the registered service. In
to the ABS property used in the query is computed this way, WSDL services registered with ROSSE
in such way that a mean of two matching degrees can be matched with semantic inferences instead
using the two ontology definitions (i.e., 100% and of using keywords only. jUDDI (http://ws.apache.
90%) is computed which is 95%. org/juddi) and mySQL (http://www.mysql.com)
It is worth noting that for an uncertain property are used to build the UDDI registry and UDDI4J
which is marked with the number of 0 in a box of (http://uddi4j.sourceforge.net/) is used to query
Table, a matching degree of 50% is given. Based the registry. OWL-S API (http://www.mindswap.
on the formula (5) presented in the Comupting org/2004/owl-s/api) is used to parse OWL-S docu-
Match Degrees section, the similarity degree ments to register services with OWL-S interfaces
between an advertised service and a service query with the OWL-S registry in ROSSE.
can be computed. In the car service query case, ROSSE provides graphical user interfaces to
for example, service S1 has a similarity degree of register services. Figure 6 shows a page to register
66.25% and service S13 has a similarity degree a vehicle service that has a WSDL Interface, and
of 74.375%. Figure 7 shows the four steps used to semantically

Figure 5. ROSSE user interface

840
Web Services Discovery with Rough Sets

Figure 6. Registering a service that has a WSDL interface

Figure 7. Annotating a vehicle service with semantic information

annotate the vehicle service. Figure 8 shows the terms of its functional input and output properties
registration of a zip code finding service with an using formula (5). As shown in Figure 5, ROSSE
OWL-S interface in ROSSE. can discover services with WSDL interfaces or
For a service request, ROSSE computes a OWL-S interfaces. It can also discover the best
matching degree for each service advertisement in service from service advertisements which has

841
Web Services Discovery with Rough Sets

Figure 8. Registering OWL-S services with ROSSE

Figure 9. Pizza ontology structure


Size
Delivery
has attributes of Price
Pizza
Location
Pizzashop
Nuttopping Hot Pizza
Vegetabletopping has attributes of Vegetarian
Fruittopping Pizza Meat has attributes of Meattopping
Tomatotopping Pizza
Fishtopping

Mushroom Veneziana

Rosa Caprina

Soho Fiorentina

the highest matching degree related to a service between properties used in service queries and
request. service advertisements. We implemented a light
In this section, we evaluate the accuracy weighted reasoning component in ROSSE to
and efficiency of ROSSE in service discovery. overcome a high overhead incurred by RACER.
We compare ROSSE with UDDI and OWL-S The component uses the Protégé OWL API (http://
respectively. RACER (Haarslev & Möller, 2001) protege.stanford.edu/plugins/owl/api/) to parse
was used by OWL-S to infer the relationships OWL documents.

842
Web Services Discovery with Rough Sets

We designed Pizza services for the tests using In the tests conducted for group 1, both OWL-S
the Pizza ontology defined by http://www.co- and ROSSE have a precision of 100%. This is
ode.org/ontologies/pizza/pizza_20041007.owl. because all service advertisements in this group
Figure 9 shows the Pizza ontology structure. do not have uncertain properties (i.e., properties
The approach adopted here can be applied to with empty values). UDDI discovered 4 services,
other domains—where a specific ontology can but only 2 services were relevant to the service
be specified. The use of service properties needs query with a precision of 50%, and a recall of
to be related to a particular application-specific 66.7%. In the tests of the last 3 groups where
ontology. advertised services have uncertain properties,
OWL-S cannot discover any services producing
a precision of 0 and a recall of 0. Although UDDI
ROSSE Accuracy in Service can still discover some services in these tests,
Discovery the precision of each group is low. For example,
in the tests of group 3 and group 4 where the
Precision and recall are standard measures that service property certainty rates are 50% and 30%
have been used in information retrieval for respectively, UDDI cannot discover any relevant
measuring the accuracy of a search method or a services. ROSSE is more effective than both UDDI
search engine (Rijsbergen, 1979). We performed and OWL-S in dealing with uncertain properties
4 groups of tests to evaluate the precision and when matching services. For example, ROSSE is
recall of ROSSE in service discovery using 10 still able to produce a precision of 100% in the
service records in each group. Each service had 5 tests of the last 3 groups albeit with a low recall
properties of which 2 properties were dependent which is 33.3%.
properties. For a service query, each group had
3 relevant services. The 10 services in group 1 ROSSE Efficiency in Service
did not have uncertain properties, but group 2 Discovery
had 3 services with uncertain properties, group
3 had 5 services with uncertain properties and We have registered 10,000 Pizza service records
group 4 had 7 services with uncertain proper- with ROSSE for testing its efficiency in service
ties. Properties such as Size, Price, Nuttoping, discovery. Service discovery involves two pro-
Vegetariantopping, and Fruittopping were used cesses, one is service matchmaking and the other
by the advertised services. Table 6 shows the is service accessing (i.e., accessing matched ser-
evaluation results. vices). We compared the efficiency of ROSSE in
matching services with that of UDDI and OWL-S

Table 6. ROSSE accuracy in service discovery


Service UDDI OWL-S ROSSE
Property
Certainty Precision Recall Precision Recall Precision Recall
Rate

100% 50% 66.7% 100% 100% 100% 100%

70% 33.3% 33.3% 0 0 100% 33.3%

50% 0 0 0 0 100% 33.3%

30% 0 0 0 0 100% 33.3%

843
Web Services Discovery with Rough Sets

Figure 10. ROSSE efficiency in service matchmaking

25000
20000
Performance

15000
(ms)

10000
5000
0
0 2000 4000 6000 8000 10000 12000
Number of Services

UDDI Matchmaking OWL-S Matchmaking


ROSSE Matchmaking

respectively, and the evaluation results are plotted RELATED WORK


in Figure 10. We also compared their efficiency
in accessing matched services, and the results are Service matchmaking is becoming an issue of vital
plotted in Figure 11. importance in service-oriented systems. UDDI
From Figure 10 we can see that UDDI has has been proposed to support service publication
the least overhead in matching services. This is and discovery. However, the search mechanism
because UDDI only supports keyword based exact supported by UDDI is limited to keyword matches
matching. UDDI does not support the inference and does not support any inference based on the
of the relationships between requested service taxonomies referred to by the tModels. Various
properties and advertised service properties extensions (Miles, Papay, Dialani, Luck, Decker,
which is a time consuming process. We also Payne et al., 2003; Powles & Krishnaswamy, 2005;
observe that ROSSE has a better performance ShaikhAli, Rana, Al-Ali, & Walker, 2003) have
in service matchmaking than OWL-S when the been proposed to complement UDDI with rich
number of advertised services is less than 5500. descriptions and powerful match mechanisms in
This is because ROSSE used a simpler reasoning support of service discovery.
component than RACER which was used by Among the extensions, the UDDI-M approach
OWL-S for matching services. However, the (Miles et al., 2003) is flexible in attaching metadata
overhead of ROSSE in service matchmaking to various entities associated with a service, but
increases when the number of services gets larger. this approach assumes the properties used in
This is due to the overhead caused by a reduction service advertisements and in service requests
of dependent properties. The major overhead are consistent. Semantic Web service technologies
of OWL-S in matching services is caused by such as OWL-S and WSMO have been proposed
RACER which is sensitive to the number of service to enhance service discovery with semantic
properties instead of the number of services. annotations. However, the classical OWL-S
From Figure 11 we can see that the ROSSE matching algorithm (Paolucci et al., 2002) cannot
matchmaking algorithm is most efficient in deal with uncertainty in service properties when
accessing matched services due to its reduction of matching service advertisements with service
dependent properties. The OWL-S has a similar requests. This work has been extended in various
performance to UDDI in this process. ways in applying Semantic Web services for

844
Web Services Discovery with Rough Sets

Figure 11. ROSSE efficiency in accessing matched services

250
200
Performance
150
(ms)
100
50
0
0 2000 4000 6000 8000 10000 12000
Number of Services
UDDI Records OWL-S Records
ROSSE Records

service discovery. For example, Jaeger, Rojec- CONCLUSION AND


Goldmann, Mühl, Liebetruth, and Geihs (2005) FUTURE WORK
introduce “contravariance” in matching inputs
and outputs between service advertisements and In this article we have presented ROSSE for service
service requests using OWL-S. Li & Horrocks discovery. ROSSE is novel in its capability to deal
(2004) introduce a “intersection” relationship with uncertainty of service properties for high
between a service advertisement and a service accuracy in service discovery. The preliminary
request. Majithia, Ali, Rana, and Walker (2004) experimental results achieved so far are encour-
introduce reputation metrics in matching services. aging. However, the following issues need to be
However, these OWL-S based methods still cannot considered for ROSSE enhancement:
deal with missing (uncertain) properties.
WSMO introduces mediators trying to • It has been shown that finding a minimal
support distinct ontologies employed by service reduct in Rough set is a problem of NP-
requests and service advertisements. However, hard when the number of attributes gets
the discovery mechanism (Keller, Lara, Polleres, large (Skowron & Rauszer, 1992). Heuristic
Toma, Kifer, & Fensel, 2004) proposed in WSMO methods need to be investigated to speed up
requires that properties used by both the goals the process in service property reduction.
and services should be consistent. • Services registered with ROSSE could be
Compared with the work mentioned above, tremendous. Scalability is one the issues
ROSSE matchmaking can deal with uncertain that need to be tackled. UDDI Version 3
properties in matching services. It takes all service (http://uddi.org/pubs/uddi_v3.htm) pro-
advertisements belonging to one service category vides larger support for multiple registries,
into one search space to dynamically identify but the specification does not specify how
and reduce irrelevant and dependent properties these registries should be structured for
which may be uncertain properties related to a enhanced scalability in service registration.
service request. Distributed Hash Table (DHT) based Peer-
to-Peer (P2P) systems such as Chord (Stoica,
Morris, Liben-Nowell, Karger, Kaashoek,
Dabek et al., 2003) and Pastry (Rowstron &

845
Web Services Discovery with Rough Sets

Druschel, 2001) have shown their efficiency Li, L. & Horrocks, I. (2004). A software frame-
and scalability in content lookup. Scalabil- work for matchmaking based on semantic web
ity in ROSSE can be improved with DHT technology. Int. J. of Electronic Commerce, 8(4),
structured P2P systems. pp. 39-60.
• Advertised services may be further de-
Majithia, S., Ali, A., Rana, O., & Walker, D. (2004).
scribed in terms of their non-functional
Reputation-Based Semantic Service Discovery,
properties related to QoS such as reliability
Proceedings of WETICE 2004, Italy.
and cost. One challenge is how to model
such QoS data so that functionally matched Martin, D., Paolucci, M., McIlraith, S., Burstein,
services can be evaluated in terms of their M., McDermott, D., McGuinness, D. et al. (2004).
QoS properties. Bringing Semantics to Web Services: The OWL-S
• Currently ROSSE only supports keyword- Approach, Proceedings of the First International
based queries. It is expected that complex Workshop on Semantic Web Services and Web
queries to be supported in ROSSE, for Process Composition (SWSWPC 2004), San Diego,
example, queries with a range or fuzzy California, USA.
queries.
Miles, S., Papay, J., Dialani, D., Luck, M., Decker,
K., Payne, T. et al., (2003). Personalised Grid
Service Discovery, IEE Proceedings Software:
REFERENCES
Special Issue on Performance Engineering, 150(4),
pp. 252-256.
Andrews, T., Curbera, F., Dholakia, H., Goland,
Y., Klein, J., Leymann, F. et al. (2003). Business Paolucci, M., Kawamura, T., Payne, T. & Sycara,
Process Execution Language for Web Services K. (2002). Semantic Matching of Web Service
version 1.1. Capabilities, Proceedings of the 1st International
Semantic Web Conference (ISWC), Berlin.
Berners-Lee, T., Hendler, J. & Lassila, O. (2001).
The Semantic Web, Scientific American, Vol. 284 Pawlak, Z. (1982). Rough sets. International Jour-
(4), pp. 34-43. nal of Computer and Information Science, 11(5),
pp. 341-356.
Haarslev, V. & Möller, R. (2001). Description of
the RACER System and its Applications, Proc. Powles, A. & Krishnaswamy, S. (2005). Extending
of 2001 International Workshop on Description UDDI with Recommendations: An Association
Logics (DL-2001), Stanford, USA. Analysis Approach, Proceedings of WSMDEIS
2005, Miami, USA.
Jaeger, M., Rojec-Goldmann, G., Mühl, G., Lie-
betruth, C. & Geihs, K. (2005). Ranked Matching Roman, D., Keller, U., Lausen, H., Bruijn, J.,
for Service Descriptions using OWL-S, Proc. of Lara, R., Stollberg, M. et al., (2005), Web Service
Communication in Distributed Systems (KiVS) Modeling Ontology, Applied Ontology, 1(1), pp.
2005, Kaiserslautern, Germany. 77 - 106.
Keller, U., Lara, R., Polleres, A., Toma, I., Kifer, Rowstron A. & Druschel, P. (2001). Pastry: Scal-
M. and Fensel, D. (2004). WSMO Web Service able, distributed object location and routing for
Discovery. Retrieved on XXX from, http:// large-scale peer-to-peer systems, Proceedings of
www.wsmo.org/2004/d5/d5.1/v0.1/20041112/ Middleware 2001, pp. 329-350, Lecture Notes in
d5.1v0.1_20041112.pdf Computer Science, Springer.

846
Web Services Discovery with Rough Sets

ShaikhAli, A., Rana, O., Al-Ali, R., & Walker, D. Winiarski, R. (2001). Rough sets methods in Feature
(2003). UDDIe: An Extended Registry for Web Reduction and Classification, Int. J. Appl. Math.
Service, Proceedings of SAINT Workshops, Or- Comput. Sci., 11(3), pp. 565-582.
lando, Florida, USA, 2003.
Jensen, R., Shen, Q., & Tuson, A. (2005). Find-
Skowron, A. & Rauszer, C. (1992). The discernibil- ing Rough Set Reducts with SAT, Proceedings of
ity matrices and functions in information systems, the 10th International Conference on Rough Sets,
Decision Support by Experience - Application of Fuzzy Sets, Data Mining, and Granular Comput-
the Rough Sets Theory, R. Slowinski (ed.), Kluwer ing (RSFDGrC), pp. 194-203, Lecture Notes in
Academic Publishers, pp. 331-362. Computer Science, Springer, Regina, Canada.
Stoica, I., Morris, R., Liben-Nowell, D., Karger, Rijsbergen, C. (1979). Information Retrieval, 1979,
D., Kaashoek, M., Dabek et al. (2003). Chord: A Butterworths: London.
Scalable Peer-to-Peer Lookup Protocol for Internet
Applications, IEEE/ACM Transactions on. Net-
works, 11(1), pp. 17-32.

This work was previously published in International Journal of Web Services Research, Vol. 6, Issue 1, edited by L.-J. Zhang,
pp. 69-86, copyright 2009 by IGI Publishing (an imprint of IGI Global).

847
848

Chapter 3.11
Generating Join Queries for
Large Databases and Web
Services
Sikha Bagui
The University of West Florida, USA

Adam Loggins
Zilliant Inc., USA

Abstract Introduction and Related


Works
In this data-centric world, as Web services and
service oriented architectures gain momentum and As we are working with more and more data, the
become a standard for data usage, there will be a sizes of databases are getting larger and larger.
need for tools to automate data retrieval. In this As businesses are going global, Web services are
article the authors propose a tool that automates becoming a standard for sharing data (Srivastava
the generation of joins in a transparent and inte- et al., 2006; Resende and Feng, 2007). Enterprises
grated fashion in heterogeneous large databases are moving towards service oriented architectures
as well as Web services. This tool reads metadata where several large databases may be layered be-
information and automatically displays a join path hind Web services, hence databases are having to
and a SQL join query. This tool will be extremely become adaptable with loosely-coupled, heterog-
useful for performing joins to help in the retrieval enous systems (Srivastava et al., 2006) too. In such
of information in large databases as well as Web scenarios of Web services and service oriented
services. [Article copies are available for purchase architectures, which may be dealing with several
from InfoSci-on-Demand.com loosely coupled heterogeneous large databases, it
is no longer humanly possible to have handy all the
information on all the tables and primary keys in all
the large databases. Although considerable work is

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Generating Join Queries for Large Databases and Web Services

being done on the challenges associated with Web Since joins are one of the most time-consuming
services addressing the problem of multiple Web and data-intensive operations in relational query
services to carry out particular tasks (Florescu processing, joins have been studied discussed ex-
et. al., 2003; Ouzzani and Bouguettaya, 2004), tensively in the literature. Mishra and Eich (1992)
most of this work is targeted towards work-flow present a very comprehensive study of works that
of applications, rather than coordinating how data have been done on joins. Query optimization is-
can be retrieved from multiple large databases in sues in joins, and devising strategies for distrib-
Web services via SQL (Srivastava et al., 2006). uted join processing have also been discussed by
In this article we try to address one aspect of this many, for example, Kim et al. (1985), Perrizo et
problem of retrieving data from multiple hetero- al. (1989), Segev (1986), Swami and Gupta (1988),
geneous large databases using SQL. Specifically, Yoo and Lafortune(1989), and Yu et al (1985,
we present a tool that automatically formulates 1987). These works have to be extended in the
joins by reading the metadata of databases in the context of databases for Web services and service
context of very large databases or in the context oriented architectures. Srivastava, et. al (2006)
of Web services which may employ the use of addresses the problem of query optimization over
several large heteregenous databases. Web services on a much broader scale.
Let us look at an example of a query presented In this article we present a tool that we have
to a Web service: Suppose a health insurance developed that will: (i) read the meta data of
company needs to verify the salary, health, and databases, that is, search the database model or
travel patterns of a person before determining schema and discover the relationships between the
the amount of health insurance he/she needs to tables using table indexes defined in the database
pay. In a Web service, this will require joining of catalogs; (ii) find efficient join paths between the
several tables. And, of course, no one person will tables to be joined; and, (iii) generate a SQL join
have knowledge of all the primary key/foreign query (in ANSI SQL standard).
key relationships between the tables to join in This rest of the article is organized as follows:
the Web services. Section two briefly describes relational databases
When databases were smaller, it was possible to with respect to the join operation; section three
have knowledge of most of the tables and primary presents an architectural overview of our tool;
key/foreign key relationships in databases, and section four presents the configuration details
SQL join queries could easily be built by joining of our tool; section five describes how we tested
tables in databases. But, in large databases layered our tool and presents some results; and section
behind Web services, it will not be possible to have six presents the conclusion. Some relevant code
knowledge of all the database schemas. portions are presented in the appendices.
The join operation, originally defined in the
relational data model (Codd 1970, 1972), is a funda-
mental relational database operation, facilitating Relational Databases and
the retrieval of information from two relations the Join Operation
(tables). Writing efficient joins is simple for small
databases since few relations are involved and one In relational databases, data is stored in the form
has knowledge of the complete database schema. of tables or relations. Each table has information
But, writing efficient joins is a challenge in large on a particular subject or concept and is composed
database scenarios and Web services where it may of a group of “related” attributes. The attributes
not be possible to have a complete picture of the in a table are all “related” in the sense that they
database schema and it’s relations.

849
Generating Join Queries for Large Databases and Web Services

describe the subject or concept of the table. For where r(a) Ө s(b) defines the join condition; a
example, there could be a table called Employee, and b are the attributes, usually the key fields of
with attributes emp_lastName, emp_midName, the respective tables R and S (usually indexed);
emp_firstName, emp_ssn, emp_birthdate, city, and Ө defines the join condition that must hold
state, homePhone, cellPhone, deptnum, etc. All true between the attributes a and b of R and S,
these attributes describe an Employee. Likewise, respectively. The Ө operation can be any one of
there could be another table called Department, the following: = , ≠ , >, < , ≥ or ≤ . The join condi-
with attributes, dept_Name, dept_Number, tion also includes multiple simple conditions of
dept_manager, dept_location, etc. Here again, the form shown above connected with the logical
all these attributes describe a Department. Now, connective AND (Earp and Bagui (2000), Elmasri
if we want information that is partly in the Em- and Navathe (2007)):
ployee table and partly in the Department table,
for example, if we want to know which employee condition AND condition AND condition
is working for a department located in LA, we
have to perform a “join” of the Employee table and
the Department table on some common attribute An Architectural
(usually the primary key field of one table and the Overview of our Tool
foreign key field of the other table). In this case
we would perform this join with a simple SQL We developed a tool that extracts configuration
query where Employee.deptnum = Department. information from system tables in databases in
dept_Number. Web services. The primary keys of the tables in
Usually, when a join query is composed, one the databases are extracted. Primary key/foreign
has to determine which tables contain the infor- key relationships are determined. A list of table
mation needed to answer the query, and has to indexes is also constructed – this information is
join those tables by the key fields. This is possible also obtained from the metadata of the databases.
if there are few tables and one has a conceptual This information is then stored in an XML con-
idea of the databases. But how do we compose figuration document. A Java interface is then used
joins when there are hundred of tables, with an to activate a GUI that takes, as input parameters,
unknown (large number) of attributes per table the names of tables that need to be joined. Then, a
in a database – the scenario for Web services. search routine is called that generates the table’s
Moreover, the conceptual schema of the databases neighbor nodes (tables) – that is, information on
could be constantly evolving or changing. which table is linked to which table. From here,
So, the join operation is used to combine re- join paths are generated, from where a valid join
lated tuples from two relations into single tuples path is then selected, and the final product is a SQL
that are stored in a resulting new relation. The join query generated from the valid join path.
desired relationship between the tuples or some Below we present the algorithm of our tool.
attributes in the tuples is specified in terms of the
join condition. In its simplest form, the join of two Algorithm of our Tool
tables or relations, R and S is written as:
Input: Activate GUI, input parameters, that is, the
R |X| r(a) Ө s(b) S names of the tables that have the final information
that is required (tables to be joined).

850
Generating Join Queries for Large Databases and Web Services

Figure 1. Architectural overview of our tool


Databases in Web Services

Extract information from the


Systems Tables

Determine Primary key/Foreign Store Configuration information


key relationships in XML doc

Java interface used to activate


GUI – read in set of target tables Search routine activates
to join

Generate table neighbor nodes

Generate join paths

Validate/select join paths

Generate SQL join query

Output: SQL Query. a. Read the primary keys of the tables


b. Check for primary key/foreign key re-
Method: lationships and generate primary key/
foreign key relationship table
1. For all the metadata in the required database 2. Call Search Routine
catalogs in Web services 3. Generate join paths
4. Generate SQL query

851
Generating Join Queries for Large Databases and Web Services

Figure 2. UML for tool

Figure 1 presents the architectural overview SELECT DISTINCT scr.PKTABLE_NAME AS lo-


of our tool. calTable,
scr.PKCOLUMN_NAME AS localCol,
scr.FKTABLE_NAME AS foriegnTable,
Tool configuration details scr.FKCOLUMN_NAME AS foriegnCol,
scr.FK_NAME AS key
The UML class diagram of the tool is presented FROM System_CROSSREFERENCE scr
in Figure 2. WHERE scr.PKTABLE_NAME = ‘-tablename-’
The first step is to extract the list of tables in
the database from the information available in And,
the Systems Table of the databases. This is done
using the following SQL query: SELECT DISTINCT scr.PKCOLUMN_NAME AS local-
Col,
SELECT table_name scr.FKCOLUMN_NAME AS foriegnCol
FROM System_tables FROM System_CROSSREFERENCE scr
WHERE table_type = ‘TABLE’ WHERE scr.PKTABLE_NAME = ‘-tablename-’
AND scr.FKTABLE_NAME = ‘-tablename2-’
The next step is to determine the primary key/
foreign key relationships between the tables and This configuration information is stored using
indexes used for the relationships. This is done an XML document, presented in appendix 1. The
using the following SQL queries: XML document creates a table of the form:

852
Generating Join Queries for Large Databases and Web Services

Table(localTable, localColumn, foreignTable, Now, we need information that is partly in


foreignColumn) table A, partly in table D and partly in table I.
That is, we want to see if tables A and D can be
Then, a search routine, as can be seen from joined and if tables A and I can be joined. We need
UML diagram (Figure 2), reads, as input param- to determine if the following links, as shown in
eters, a list of table names to find relationships Figure 3, exist.
between, and searches for the relationships be- The algorithm takes the first table: A, and finds
tween the tables by looking up the primary key/ all its neighbors. A’s neighbors are all the tables
foreign key relationships between the tables. The that A links to. Now suppose for example, it was
code of the Searcher class is presented in appendix found that A’s neighbors are B, C, and D, that is,
2. The sampleQueryBuilder class develops the A links to B, C, and D, as in Figure 4.
GUI. This class takes in the database type (in That is, A’s primary key is in tables B, C and
this case, hypersonic) as the input parameter, the D as the foreign key, as shown in Figure 5.
XML configuration file (presented in appendix So, one of the targeted links have been found,
1), the table names, and calls the Searcher class, A-D. But, all the targeted links have not been
which then creates the nodes. found, so the algorithm keeps running until all
the targeted links have been found. So now, the
The Search Routine algorithm stores the links A-B, A-C, A-D, as
shown in Figure 4.
The search routine reads, as input parameters, a Next the routine finds the neighbors of tables B,
list of table names to find relationship between, C, and D. Now suppose for example, B’s neighbor
and searches for relationships between the tables. is E, C’s neighbor is F and D’s neighbor is G and
Below we present the algorithm of our search H. So now we have found the following links
routine. A-B, A-C, A-D, A-B-E, A-C-F, A-D-G, A-D-H,
as shown in Figure 6.
Algorithm of Search Routine Next search routine finds the neighbors of E,
F, G and H. Now suppose that E links to I and J,
Until all targets are found and F links to K and L, as shown in Figure 7.
a. Read in the tables to be joined The algorithm now stops since the targeted
b. Determine the relationships between the tables have been found: A links to D directly:
tables A-D. And, A links to B which links to E which
c. For all tables links to I: A-B-E-I. The algorithm now keeps
i. Find all neighbors only these two paths as the valid join paths. Using
ii. If targets are found stop else find neigh- these join paths, the next step was to generate a
bors of new tables SQL join query.

For example, let us assume that we have the Generating the SQL Query
following tables in a database schema:
Our algorithm to create the SQL join query goes
{A, B, C, D, E, F, G, H, I, J, K, L} down the shortest join path first. The shortest
path is A-B-E-I. So, the joins will be in the form
And assume that you do not know the pri- (see Box 1).
mary key/foreign key relationships between the As shown in Figure 8.
tables.

853
Generating Join Queries for Large Databases and Web Services

Testing the Tool (JBPM) in the context of Web services. The


hypersonic database (HSQLDB), freely avail-
We tested our tool using the hypersonic database able on the Web at http://www.hsqldb.org/, is a
and the Java Business Process Management leading SQL relational database engine written in

Figure 3. Are these tables linked?

D A I

Linked? Linked?

Figure 4. Neighbors of Table A

B C D

Figure 5. Primary key/Foreign key relationships

TableA(PrimaryKeyOfA, AttributeA1, AttributeA2)


TableB(PrimaryKeyOfB, AttributeB1, PrimaryKeyOfA)
TableC(PrimaryKeyOfC, AttributeC1, PrimaryKeyOfA)
TableD(PrimaryKeyOfD, AttributeD1, PrimaryKeyOfA)

Figure 6. Neighbors of Tables B, C, and D

B C D

E F G H

854
Generating Join Queries for Large Databases and Web Services

Figure 7. Neighbors of Tables E, F, G, and H

B C D

E F G H

I J K L

Box 1.
Table A INNER JOIN Table B ON A.X = B.X
Table B INNER JOIN Table E ON B.X = E.X
Table E INNER JOIN Table I ON E.X = I.X
Table A INNER JOIN Table D ON A.X = D.X

Figure 8. Join paths

A
1st join 4 th join
B nd D
2nd join

3rd join
I

Java. The HSQLDB database engine offers both any Java application. JBPM (or JBOSS JBPM)
in-memory and disk-based tables and supports presently has a fully-formed table schema of 38
embedded and server nodes. tables with multi-table relationships. Figure 9
The JBPM data model, available at http://www. shows a snapshot of a portion of the complicated
jboss.com/products/jbpm, is a business friendly multi-table schema of the JBPM data model.
open source piece of software with an architecture These tables house data that record the state of a
that will run standalone or can be embedded with process as it progresses through its life cycle. The

855
Generating Join Queries for Large Databases and Web Services

Figure 9. Schema of the JBPM data model

JBPM_ProcessInstance table, for, example, stores schema of the JBOSS data source is given in
the process instance id along with start and end Figure 9). We selected the JBPM_ACTION and
dates; the JBPM_ProcessDefinition table stores JBPM_BYTEBLOCK tables. This was an arbitrary
the various definitions that are present for use by selection, and the user can select any table or any
the application along with their versions. number of tables by selecting the select another
tab (shown in Figure 11). We will illustrate this
Running the Application software by using two tables. After the user selects
the tables, as shown in Figure 11, the user clicks
Step 1: From the GUI interface (shown in Figure the find relations tab.
10), the user selects a data source (of the database).
We selected JBOSS, as shown in Figure 10. Step 3: Once you click find relations, you will
get join path displayed in Figure 12.
Step 2: The next step will be to select the tables
that you want to join from that data source (the

Figure 11. Selecting tables using the GUI


Figure 10. The GUI interface

856
Generating Join Queries for Large Databases and Web Services

Figure 12. The join path

From Figure 12’s output screen we can see that results of some of the test runs. The first column
there is join path from the JBPM_ACTION table to shows the tables that needed to be joined. The
the JBPM_NODE table to the JBPM_PROCESS- second column shows the number of tables that
DEFINITION table to the JBPM_MODULEDEF- the algorithm needed for the output. The third
INITION table to the JBPM_BYTEARRAY table column shows the resulting number of 2-table
to the JBPM_BYTEBLOCK table. joins that the algorithm required for the output.
This join path generates the ANSI SQL join And the last column shows the number of mil-
query shown in Figure 13. liseconds it took to produce the output for this
particular join.
Results On the average, six tables needed to be joined
to obtain the results of 2 table joins, and the time
Using the large database, JBPM, we ran several averaged 53.2 mil seconds; 4 tables needed to be
tests to test our software. Below we present the joined to obtain the results of 3 table joins, and

Figure 13. ANSI SQL join query generated

857
Generating Join Queries for Large Databases and Web Services

Table 1.
Total Number of Number of mil
Nu mber of Number of
different tables seconds it took
Tables to be joined tables to be 2-t able joi ns
joined to get re- t o b u i ld t h e
joined required
sult query
JBPM_ACTION, JBPM_BYTEARRAY 2 4 8 94
JBPM_ACTION, JBPM_RUNTIMEACTION 2 2 4 32
J BPM_NODE, J BPM_ ACTION, J BPM_
3 2 4 47
EVENT
JBPM_TASKINSTANCE, JBPM_SWIMLA-
3 4 8 47
NEINSTANCE, JBPM_SWIMLANE
JBPM_LOG, JBPM_COMMENT, JBPM_PRO-
3 4 8 31
CESSINSTANCE
J BPM _VA R I A BLEI NSTA NCE , J BPM _
LOG, JBPM_TOKEN, JBPM_TOKENVARI- 4 3 6 31
ABLEMAP
JBPM_ACTION, JBPM_PROCESSDEFINI-
3 2 4 47
TION, JBPM_TRANSITION
JBPM_PROCESSDEFINITION, JBPM_VARI-
A BLEI NSTA NCE, J BPM _ PROCESSI N- 3 2 4 94
STANCE
JBPM_PROCESSINSTANCE, JBPM_NODE 2 2 4 78
JBPM_LOG, JBPM_VARIABLEINSTANCE 2 1 2 31
JBPM_VARIABLEINSTANCE, JBPM_PRO-
2 2 4 47
CESSDEFINITION
J BPM _ I D_ PE R M ISSIONS , J BPM _ I D_
2 - - 16
GROUP
JBPM_POOLEDACTOR, JBPM_PROCESSIN-
2 3 6 62
STANCE
JBPM_TIMER, JBPM_TASK 2 2 4 31
J BPM _ AC T ION , J BPM _ PRO C E S SI N -
2 3 6 125
STANCE
JBPM_PROCESSDEFINITION, JBPM_RUN-
2 2 4 47
TIMEACTION
JBPM_PROCESSINSTANCE, JBPM_COM-
2 2 4 47
MENT
JBPM_SWIMLANE, JBPM_DELEGATION 2 1 2 140
JBPM_ACTION, JBPM_EVENT, JBPM_LOG,
JBPM_SWIMLANE, JBPM_SWIMLANEIN- 5 5 6 204
STANCE
JBPM_PROCESSDEFINITION, JBPM_PRO-
CESSINSTANCE, JBPM_NODE, JBPM_TO- 5 5 6 266
KEN, JBPM_TASKINSTANCE

858
Generating Join Queries for Large Databases and Web Services

Figure 14. Number of mil seconds to build a SQL join query for 2 tables

Figure 15. Number of mil seconds to build a SQL join query for 3 tables

the time averaged 62.5 seconds; six tables needed Conclusion


to be joined to obtain the results of 4 table joins,
and the time averaged 46.5 mil seconds; six tables This tool will generate join paths and a SQL
needed to be joined to obtain the results of 5 table join query from large databases as well as large
joins, and the time averaged 235mil seconds. databases in Web services. The user only needs
Figure 14 presents a graphical representation to know which tables he/she wants to join, but
of the number of milliseconds it took to build a does not have to know the join path needed to join
query for two table joins, and Figure 15 presents them (the primary key/foreign key relationships
a graphical representation of the number of mil- between the tables). The software finds the join
liseconds it took to build a query for three table path between the tables selected, and displays a
joins. SQL join query. This software is a very important
step forward in the process of retrieving informa-

859
Generating Join Queries for Large Databases and Web Services

tion efficiently from large databases as well as Ouzzani, M., & Bouguettaya (2004). A. Efficient
large databases used in Web Services and Service access to Web services. IEEE Internet Comput-
Oriented Architectures. ing, 8(2), 34-44.
Perrizo, W., Lin, J.Y.Y., & Hoffman, W. (1989,
June). Algorithms for distributed query process-
Acknowledgment
ing in broadcast local area networks. IEEE Trans.
Know. Data Eng., 1(2), 215-225.
The authors would like to thank the Editor, Dr.
Ghazi Alkhatib, and the referees of this article for Resende, L., & Feng, R. (2007, June 12-14).
their constructive suggestions, which led to the Handling Heterogeneous Data Sources in a SOA
present improved version of this article. Environment with Service Data Objects (SDO).
SIGMOD’07, Beijing, China.
Segev, A. (1986, March). Optimization of join
References
operations in horizontally partitioned database
systems. ACM Trans. Database Systems, 11(1),
Codd, E. (1970). A Relational Model for Large
48-80.
Shared Data Banks, CACM, 13(6).
Srivastava, U., Munagala, K., Widom, J., &
Codd, E. (1972). Further Normalization of the
Motwani, R. (2006, September 12-15). Query
Data Base Relational Model. Data Base Systems.
Optimization over Web Services. VLDB ’06,
Prentice Hall.
Seoul Korea.
Earp, R., & Bagui, S. (2000). Oracle’s Joins,
Swami, A., & Gupta, A. (1988). Optimizing large
Oracle Internals, 2(3), 6-14.
join query. Proceedings of SIGMOD, 8-17.
Elmasri, R., & Navathe, S.B. (2007). Fundamen-
Yoo, H., & Lafortune, S. (1989, June). An intel-
tals of Database Systems, (5th ed.). Boston, MA:
ligent search method for query optimization by
Pearson Education.
semi-joins. IEEE Trans. Knowl. Data Eng., 1(2),
Florescu, D., Grunhagen, A., & Kossmann, D. 226-237.
(2003). XL: A platform for Web services. Proceed-
Yu, C. T., Chang, C. C., Templeton, M., Brill, D., &
ings of First Biennial Conference on Innovative
Lund, E. (1985). Query Processing in a fragmented
Data Systems Research (CIDR).
relational database system: Mermaid. IEEE Trans.
Kim, W., Reiner, D. S., & Batory, D. S. (1985). Software Eng. SE-11, 8, 795-810.
Query Processing in Database System. New York:
Yu, C. T., Guh, K. C., Zhang, W. Templeton, M.,
Springer-Verlag.
Brill, D., & Chen, A. L. P. (1987). Algorithms to
Mishra, P., & Eich, M.H. (1992). Join Processing process distributed queries in fast local networks.
in Relational Databases. ACM Computing Surveys, IEEE Trans. Comput. C-36, 10, 1153-1164.
24(1), 63-113.

860
Generating Join Queries for Large Databases and Web Services

Appendix A

Storing the configuration information using an XML document

<?xml version=”1.0” encoding=”UTF-8”?>


<!--
Document : configDoc.xml
Author : Owner
-->
<config>
<datasource id=”TEST1”>
<driver>org.postgresql.Driver</driver>
<url>jdbc:postgresql://localhost:5432/TEST1</url>
<username>postgres</username>
<password>****</password>
<type>postgres</type>
</datasource>
<datasource id=”school”>
<driver>oracle.jdbc.driver.OracleDriver</driver>
<url>jdbc:oracle:thin:@unix.cslab.uwf.edu:1521:STUDENT _ COURSE</url>
<username></username>
<password></password>
</datasource>
<datasource id=”JBOSS”>
<driver>org.hsqldb.jdbcDriver</driver>
<url>jdbc:hsqldb:E:\Graduate _ Project\guibuildTest\hypersonic\localDB</url>
<username>sa</username>
<password></password>
<type>hypersonic</type>
</datasource>
<DBType id=”oracle”>
<query id=”tables”>
select table _ name
from ALL _ IND _ COLUMNS
</query>
<query id=”table _ relation”>
select con.table _ name as localtable,
concol.column _ name as localColumn,
icol.TABLE _ NAME as foreignTable,
icol.Column _ name as foreignColumn,
con.Constraint _ name as key
from all _ constraints con,
all _ ind _ columns icol,
all _ cons _ columns concol

861
Generating Join Queries for Large Databases and Web Services

where con.constraint _ type = ‘R’


and con.R _ constrant _ name = icol.Index _ Name
and con.constraint _ name = concol.Constraint _ Name
</query>
</DBType>
<DBType id=”hypersonic”>
<query id=”tables”>
select table _ name
from System _ tables
where table _ type = ‘TABLE’
</query>
<query id=”table _ relation”>
select distinct scr.PKTABLE _ NAME as localTable,
scr.PKCOLUMN _ NAME as localCol,
scr.FKTABLE _ NAME as foriegnTable,
scr.FKCOLUMN _ NAME as foriegnCol,
scr.FK _ NAME as key
from System _ CROSSREFERENCE scr
where scr.PKTABLE _ NAME = ‘-tablename-’
</query>
</DBType>
<DBType id=”postgres”>
<query id=”tables”>
select distinct tabs.tablename as table _ name
from pg _ catalog.pg _ tables tabs
where tabs.schemaname = ‘public’
</query>
<query id=”table _ relation”>
select localTab.tablename as localTable,
locCol.column _ name as localColumn,
refTab.tablename as foriegnTable,
refCol.column _ name as foriegnColumn,
keys.FK as key
from pg _ catalog.pg _ indexes localTab,
constraint _ column locCol,cvs
constraint _ column refCol,
pg _ catalog.pg _ indexes refTab,
(select locTab.conname as localTabIndex,
refTab.conname as foriegnTabIndex,
con.conname as FK
from pg _ catalog.pg _ constraint con,
pg _ catalog.pg _ constraint locTab,
pg _ catalog.pg _ constraint refTab
where con.contype = ‘f’

862
Generating Join Queries for Large Databases and Web Services

and con.conrelid = locTab.conrelid


and con.confrelid = refTab.conrelid ) keys
where localTab.indexname = keys.localTabIndex
and refTab.indexname = keys.foriegnTabIndex
and locCol.constraint _ name = keys.localTabIndex
and refCol.constraint _ name = keys.foriegnTabIndex
and localTab.tablename = locCol.table _ name
and refTab.tablename = refCol.table _ name
</query>
</config>

This work was previously published in International Journal of Information Technology and Web Engineering, Vol. 4, Issue 2,
edited by G. I. Alkhatib; D. C. Rine, pp. 45-60, copyright 2004 by IGI Publishing (an imprint of IGI Global).

863
Section IV
Utilization and Application

This section introduces and discusses the utilization and application of Web technologies. These particu-
lar selections highlight, among other topics, the application of semantic Web technologies to e-tourism,
e-banking, and in car repairs as well as the adoption of Web services in digital libraries. Contributions
included in this section provide excellent coverage of today’s online environment and insight into how
Web technologies impact the fabric of our present-day global village.
865

Chapter 4.1
The Role of Web Services:
A Balance Scorecard Perspective
Pauline Ratnasingam
University of Central Missouri, USA

Abstract as they are used to facilitate the seamless flow of


business transactions and are known to offer many
This chapter aims to examine the extent of Web benefits.
services usage and quality, applying the balanced However, studies also show that the lack of
scorecard methodology in a small business firm as effective use, quality, and security in Web service
an exploratory case study. This chapter contributes applications is one of the main reasons why firms fail
to guidelines and lessons learned that will inform, to realize the full potential of their IT investments
educate, and promote small businesses on the impor- (Benko & McFarlan, 2003). It is imperative that
tance of maintaining the quality of Web services. small businesses focus on the quality of Web services
and their operations given the extent to which Web
service applications are used in business processes
Introduction in this fast changing market conditions. Enforcing
and maintaining the quality of Web services does
The Internet, a rapidly expanding global computer not only involve a set of security analyses and audit
and communication infrastructure, has facilitated procedures that most firms conduct periodically, but
the emergence of digitization and globalization rather it is a continual process that needs to align
that in turn has permitted businesses to extensively a rigorous methodology. Such methodology is the
engage in foreign investments. Forrester Research balanced scorecard, which is a set of quantifiable
suggests that e-commerce in the U.S. will grow measures that aim to monitor and manage a firm’s
19%, reaching $230 billion by 2008. Today, firms strategic performance. This chapter aims to examine
are attempting to attain their value chain goals by the extent of Web services usage and quality by
offering to sell products and services. Web services applying the balance scorecard methodology in a
have become a significant part of small business, small business firm.
The balanced scorecard is needed to align, moni-
DOI: 10.4018/978-1-60566-042-4.ch016 tor, and adjust the factors that impact the quality

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
The Role of Web Services

of Web services. Previous studies applying the their outcomes?) measures (as in how to achieve
balanced scorecard in the context of Web services their outcomes?) targets, that is, accountability
and quality is limited. Only 10% of the organi- (as in how do we know that we have achieved
zations executed their implementation strategy it?), and initiatives (as in what actions to take?).
to apply the balanced scorecard methodology Further, the balanced scorecard is based on three
because they experienced barriers in formulat- time dimensional timelines, namely, yesterday,
ing a vision, allocating resources (i.e., human today, and tomorrow. The next section presents
resources), and managing change (Niven, 2003). a discussion of the four perspectives.
This chapter aims to examine the extent of Web
services usage and quality, applying the balanced 1. The learning and growth perspective aims
scorecard methodology in a small business firm to measure the human, information, and or-
as an exploratory case study. The next section ganizational capital. Human capital includes
discusses the theory of balanced scorecard and the skills, knowledge, expertise, the extent of
Web services followed by the development of a training given to employees, and the business
framework which integrates the critical success cultural attitudes. Do small business employ-
factors. Then we discuss the research method and ees have the skills/competencies to operate
provide a description of the background informa- the Web service application and align it with
tion of the firm. We then test the framework via their internal business processes effectively
an exploratory case study and report the findings. in order to meet their customers’ objectives
The findings contribute to guidelines and lessons of using Web services? Information capital
learned that will inform, educate, and promote aims to measure effective communication
small businesses on the importance of maintaining and information sharing. Do small business
the quality of Web services. Finally, we conclude employees possess the information required
the chapter with contributions and directions for to achieve objectives? Organizational
future research. capital aims to monitor the soft areas of the
employees, such as, learning and growth,
culture, leadership, knowledge sharing, and
Background Information: teamwork. Do small businesses have the
The Balanced Scoreboard ability to sustain growth and change that in
turn enhances the quality of Web services?
The balanced scorecard deployed to measure 2. The internal business process perspective
the effective use and quality of Web services aims to measure performance that permits
among small businesses focuses on a system that small businesses to be aware of the quality
enforces measurement and feedback, thereby of their products and services. Web services,
imposing quality, continuous improvement, em- considered as system quality, are defined
ployee empowerment, and strategic performance as “the conformance to explicitly stated
that aim to sustain the competitive and strategic functional and performance requirements,
objectives. The balanced scorecard measures the explicitly stated development standards,
performance of Web services in a small business and implicit characteristics that are expected
firm from four perspectives, namely, learning and of all professionally developed software”
growth, internal business processes, customer, (Solano, De Ovalles, Rojas, Padua, &
and financial perspectives, which are discussed Morales, 2003, p. 67). Similarly, Ortega,
below. Each of these four perspectives is further Perez, and Rojas (2000) suggest that product
categorized by their objectives (as in what are effectiveness should include characteristics

866
The Role of Web Services

such as timeliness, functionality, reliability, line of small businesses is achieved. What


usability, efficiency, maintainability, and accountability do small businesses that use
probability. Small businesses need to be Web services have to financial stakeholders?
aware of the following questions when In many respects, the financial perspec-
assessing the quality of their Web services tive represents “end in mind” of the small
performance. Does our internal business business strategic vision. Small business
processes applying Web services conform managers are able to examine the outcomes
to the mission of small businesses? Does of their business performance that provide
the internal business processes meet our strategic financial feedback and show the
customer requirements? There are two types trends of their business performance using
of processes under strategic management. Web services overtime. Figure 1 presents the
First, the mission oriented-process focuses perspectives of the balanced scorecard.
on the strategic goals of small businesses,
and second, the support processes are more
repetitive and are used in the daily opera- Web Services
tions that in turn enforce benchmarking. The
balanced scorecard provides a diagnostic Web services are creating a service-oriented ar-
feedback into the various internal processes, chitecture that provides technical solutions and e-
thereby guiding and improving the busi- collaborations among value chain partners (Chan,
ness processes involved in the use of Web Kellen, & Nair, 2004; Yen, 2004). Web services
services on a continuous basis. What must refer to a new breed of Web applications as self-
small businesses do well internally in order contained, self-describing modular applications
to achieve the objectives they set forth to that can be published, located, and invoked across
achieve quality in Web services? Where the Web. Businesses use existing software com-
does the Web services “process” start, and ponents specified as services to perform business
where does it end? operations in a “service-oriented architecture.”
3. The customer perspective focuses on meet- Similarly, Web services refer to a set of software
ing the needs of the customers, retaining applications or components developed using a
existing customers, and gaining customer specific set of application programming interface
satisfaction. What do customers expect (API) standards and Internet-based communica-
or demand from the use of Web services? tion protocols. The objective is to enable these
Dimensions of customers experience include applications or components to invoke function
time, quality, price or cost, accessibility, calls and exchange data among themselves over
reputation, and relationship. Who do we the standard Internet infrastructure. We define
define as our customers? How do our cus- Web services as “modular Internet-based business
tomers see us? How do Web services create functions that perform specific business tasks to
value for our customers? facilitate business interactions within and beyond
4. The financial perspective aims to provide the organization.” Further, Web services generate
timely and accurate financial information. sustainable competitive advantage for firms sup-
By implementing a centralized database, it porting their core competencies and adding value
is hoped that processing can be standardized to the execution of the corporate strategy. The
and automated. Further, both risk assess- technology’s fullest potential will not be realized
ment and cost benefit analysis can be easily if it is used only for improving the operational ef-
conducted in order to ensure that the bottom ficiency of existing business processes. Therefore,

867
The Role of Web Services

Figure 1. The balanced scorecard perspectives

we focus on the quality of Web services usage roles. First, it helps the Web service requestors
by applying a rigorous methodology called the and providers to find and connect with one an-
balanced scorecard methodology as it measures other; secondly, it creates a trusted environment
the effective use and quality of Web services essential for carrying out mission-critical business
from four perspectives, namely, learning and activities, thereby contributing to technology trust.
growth, internal business processes, customer, Finally, the top layer consists of a diverse array of
and financial perspectives, thereby providing a application services. It is in this top layer where
holistic view. the day-to-day operations will be most visible to
employees, customers, and trading partners. The
The Web Services Architecture top layer performs the service binding and invoca-
tion. Similarly, three are layers of Web services,
The Web services architecture is made up of three namely, basic services, composite services, and
layers of technology. At the foundation (the bottom managed services. While basic services manage
layer) is the software standards and communica- the publication, discovery, selection, and binding,
tion protocols that provide a common language composite services facilitate the coordination,
for Web services and enable applications to be conformance, monitoring, and quality of service.
connected. Standards and protocols are often Finally, managed services provide the market
cited as the strategic value that Web services bring certification rating and operations support.
(Hagel, 2002; Hagel & Brown, 2001). The service Table 1 presents the three layers of the Web
grid (the middle layer) provides a set of shared services architecture adapted from Hagel and
utilities from security to third-party auditing, to Brown (2001).
billing and payment so that critical business func-
tions and transactions over the Internet can be
conducted. This layer publishes the services that
serve as entry points for queries to find service
descriptions. In short, the service grid plays two

868
The Role of Web Services

Table 1. The Web services architecture

Top Layer – Application Service


Web services runs on any application platform as long as it has a Web server connected to the Internet.
Middle Layer – Service Grid
The service grid layer provides four types of utilities:
     (1) Service management utilities (i.e., provisioning, monitoring, ensuring quality of service, synchronization, conflict resolution)
     (2) Resource knowledge management utilities (i.e., directories, brokers, registries, repositories, data transformation)
     (3) Transport management utilities (i.e., message queuing, filtering, metering, monitoring, routing, resource orchestration)
     (4) Shared utilities (i.e., security, auditing, and assessment of third party performance, billing and payment)
Bottom Layer – Software Standards and Communication Protocols
Software standards include:
     (1) Web service description language (WSDL) to describe the Web service
     (2) Universal, description, discovery and integration (UDDI) to advertise, syndicate as a central organization for registering, finding,
and using Web services)
     (3) Web services flow language (WSFL) to define work flows
     (4) XML (format for data exchange and description)
     (5) Communication protocols including simple object access protocol (SOAP) to communicate that are for calling Web services,
HTTP, and TCP/IP)

Framework of the Balanced services by deploying the balanced scorecard


Scorecard for Web Services framework for small businesses based on the
objectives (what are the outcomes?) measures
The framework of balanced scorecard for Web (how to achieve the outcomes?), targets, that is,
services was developed by integrating the theory of accountability and initiatives (how do we know
balanced scorecard and Web services. The frame- that we have achieved it?), and initiatives (what
work consists of critical success factors or indicators actions to take?) for all the four perspectives (i.e.,
that make up the objectives, measures, targets, and learning and growth, internal business processes,
initiatives. The goal of these critical success factors customer, and financial perspectives).
is to evaluate the effective use and quality of Web In this study we interviewed the managers
service applications. Table 2 below illustrates the of the firm in the agricultural industry. Initial
framework of the balanced scorecard in Web ser- entry into the case site was obtained by mak-
vices, which serves as a measurement tool, thereby ing telephone calls to key representatives in the
ensuring the quality of Web services. organization. A brief description and purpose of
the study was discussed before requesting them
to participate. The telephone conversation was
Research Method followed by an e-mail to the respondents with an
attachment of the file describing the purpose of
Case studies were chosen as an appropriate method the study. Once confirmation of their willingness
to evaluate the effective use and quality of Web to participate in the study was received, appoint-
services among small businesses as it elicited ment dates for interview sessions were arranged
subtle and rich data needed, thereby increasing with the managers in the firm. Evidence for the
our understanding in the use and quality of Web exploratory case study data came from the hand
services applying the balanced scorecard methodol- written notes taken during the two (90 minutes)
ogy (Yin, 1994). face-to-face interviews and the tape recorded
We attempted to identify the critical success data. In addition, analysis of existing documents
factors in the effective use and quality of Web relating to Web service applications, day-to-day

869
The Role of Web Services

Table 2. Framework of the balanced scorecard in Web services

Balanced Scorecard Perspectives Relationship to Effective use and Quality of Web Services via the
Critical Success Factors (or Indicators)
(1) (1) Learning and growth perspective Objectives: employees must be well trained and are expected to perform
     How can we continue to improve and create value in their day-to-day business operations applying Web services
the use of Web services? Measures: provide online training, user manuals, standard operating proce-
dures, help desk, and reward employees with high productivity
Targets: fewer customers and stakeholder complaints
Initiatives: ongoing monitoring of employees performance, focus on the
culture, climate, and commitment of the organization.
(2) Internal business process perspective Objectives: to achieve high quality and productivity of the services provided
What processes do we need to excel further when using via Web services
Web services? Measures: apply best business practices, reliable, accurate, and timely infor-
mation. Focus on the usability, interoperability of the system
Targets: increased profit and fewer customers and stakeholders complaints
Initiatives: ongoing audit and applying the quality assurance plan on a
regular basis
(3) Customer perspective Objectives: satisfaction of customers and stakeholders, reputation of the firm
How can we enhance our business reputation with our and the quality of their products and services
customers via Web services? Measures: open, frequent communications, providing excellent quality
services, provide value for money and reliable operations
Targets: increase in profit and sales
Initiatives: training employees, ongoing weekly meetings with employees;
regular review and reflection of the goals and mission of the company
(4) Financial perspective Objectives: increase in profits, rate of return
How are we perceived by our shareholders and other Measures: increased productivity, increase quality of services provided, ap-
stakeholders invested in our firm? ply the return on capital employed, economic value added and free cash flow
Targets: profit figures, increased shareholders value
Initiatives: advertising their company products, attending trade shows,
business conferences and seminars, word of mouth, control operating costs,
maximize potential use of Web services

interactions, operating procedures, Web sites, and government for beautifying the land, such as for
internal security policies were analyzed. The tape the city of Blue Springs and the Tiffany Springs
recorded data was transcribed to identify concepts highway. Their main form of Web service appli-
via pattern matching to the balanced scorecard cation is their business-to-consumer (B2C) Web
framework. site implemented in 2003, which has embedded
IT solutions, e-mail, and fax. Table 3 presents the
background information of Firm A.
Findings: Background In this section we report the findings of the
Information of Firm A exploratory case study from the hand written notes
taken during the face-to-face interview with the
Firm A is a small business seed manufacturer and manager of Firm A.
seller of wild flower and bird seeds located in
Kingsville. It is a family owned business with 10 Web Services Quality in the
employees and the owner has been in this business Learning and Growth Perspective
for 23 years. They sell a variety of wild flowers
seeds including the annual mix, shade, and suns. Firm A found that the learning capability in ap-
They also supply bird seeds to regular stores, resi- plying best business practices was important for
dential customers, and fulfill large bids from the their business performance. Although only six

870
The Role of Web Services

Table 3. Background information of firm A

Background Characteristics Firm A


Type of industry Agricultural industry
Size of Firm Small business
Type of ownership Family owned
Number of employees 10
Type of Web service application B2C Web site & embedded IT solutions
Type and number of customers/business partners Regular store customers, government bid contracts, Web customers
Annual revenue in US$ $1.5 million
How long have they being in business? 23 years
Mode of attracting customers Advertise in the local newspaper, Web page

employees were assigned to operate their computer factors: timeliness of obtaining information
systems, they had to abide to best business prac- or processing the transaction, accuracy as in
tices that included changing their passwords every achieving integrity in the content of the message,
10 days and not disclosing pricing information to confidentiality, access rights, non-repudiation,
any other employees other than the manager and reusability, portability, reliability, security and
two other employees in the accounts department. efficiency of the Web service applications were
Each employee was given limited access rights considered in this perspective.”
and was unable to see the detail information of The manager indicated that, “Further, with
the transaction. industry-accepted standards and protocols, Web
The manager noted, “We continually improve services provided a standard interface allowing
our employee skills by providing training to our integration among heterogeneous platforms,
employees in order to increase the potential use thus facilitating efficient and effective collabo-
of Web services.” ration between departments that use different IT
Web services were deployed to facilitate systems. Finally, Web services’ service-oriented
information sharing and collaboration among architecture allows firms to build a flexible IT
employees and business units. infrastructure that enables faster decision-making
and response to market changes.”
Web Services Quality in Improving Hence, through the orchestration of modu-
Internal Business Processes lar, loosely coupled software components, Web
services enable an “assemble line approach” to
The internal business process perspective focused software development, resulting in a responsive
on the quality and use of Web services in activi- IT infrastructure for designing and building
ties such as supply chain management, customer faster application development and enterprise
relationship management, and research and de- applications.
velopment. The balanced scorecard provided a We argue that Web services technology, with
systemic quality approach to assess Web services its industry-accepted standards and protocols, can
as it allowed the software processes to be ef- enhance internal business operations by enabling
ficiently and effectively balanced. The manager process automation, increasing interoperability
noted, “We applied the quality assurance plan and reducing integration complexity, and improv-
which included the following critical success ing process design.

871
The Role of Web Services

Web Services Quality in the shareholders and there is a balance between


Improving Customer Retention growth and productivity. Further, return on capi-
and Relationships tal employed and economic value indicators are
added.”
The customer perspective is the core of the business
strategy. The manger noted, “Our Web services
and IT solutions offered us with unique business Lessons Learned
processes and customer value propositions that
determined the correct business processes thereby The findings suggests a cyclic process that was
creating better customer relationships.” created with the use of the balanced scorecard ap-
The manager also noted, “We have key attri- proach to evaluate the quality of Web services ap-
butes of Web services that create customer value plications and in order to integrate quality, provide
propositions such as; enhanced customer intimacy a strategic map, and indicate how information will
via open communications, improved customer be disseminated so that the potential use of Web
retention, and better customer value. Beyond services can be attained. The processes adapted
the quality and specifications of its products and from Kaplan and Norton (2000) included:
services we try to satisfy our customers by meet-
ing their needs and offering quality goods and • Analysis of the sector, its development
services.” and role of company: Refers to the identi-
These attributes that serve as critical success fications of the key goals in the use of Web
factors were derived from the balanced scorecard services and establishing the characteris-
methodology which comprises of objectives, tics and requirements for the industry.
measures, targets, and initiatives. • Establishing or confirming the com-
The manager indicated, “Web services made pany’s strategic plan: Refers to the es-
our firm’s IT infrastructure more flexible and tablishment or confirmation of a strategic
adaptable, affording the organizational agility to plan, intensifying the internal and external
meet the ongoing customers’ changing require- analysis of the earlier processes, and ensur-
ments.” ing that agreements are arrived towards the
quality of Web services.
Web Services Quality in • Translating the strategy into operational
Improving Financial Position terms: Refers to the actual actions taken
to ensure that best business practices, stan-
The use of the balanced scorecard methodology dards, and quality procedures that were
allowed for improved capability of learning and followed in the use of Web services. For
innovation, better internal business processes, and example, in the financial perspective they
the enhanced customer value that in turn served aim to create value for the shareholders
as performance drivers for increased financial and there is a balance between growth and
returns. productivity. Return on capital employed
The manager noted, “Web services has directly and economic value indicators are added.
influenced our shareholder value as it influenced Likewise, in the customer’s perspective,
our firm’s financial strategy, productivity and the growth in terms of volume generated
revenue growth.” from customers was examined. Further,
Further, he added, “For example, in the fi- segments that value quality and innovation
nancial perspective we aimed to create value for were emphasized. In the internal business

872
The Role of Web Services

process perspective they try to differen- growth, internal business processes, customer,
tiate between basic operating processes and financial perspectives) and tested it via an
and operating excellence in the support exploratory case study of a small business firm
services via the use of Web services. The within the agricultural industry. Then we reported
product and service quality were measured the findings based on the impact of Web services
through the product quality index indica- quality on the four perspectives. The findings
tor using the market share in order to gain suggest that the lessons learned evolve over a set
profit from the investment in the financial of processes that are aimed at integrating quality
perspective. Finally, in the learning and into the potential use of Web services.
growth perspective, three strategic objec- The contribution of this study was attributed
tives were identified, namely, basic com- to the framework, which provided guidelines for
petencies and skills (referring to the skills measuring Web services usage and quality, apply-
expected of the employees), technology ing the key Web services features that gave rise to
(referring to the Web services applications business values that were matched with strategic
used in the value chain), and climate for initiatives that small businesses can have in each
action (referring to organization commit- of the business perspectives. Web services further
ments that must be implemented by the hu- drive the quality of IT strategy as it is fully aligned
man resources department). with overall business strategy, thereby focusing on
• Aligning the organization with the strat- the key organizational capabilities and sustainable
egy: Refers to the alignment of business competitive advantage.
unit goals with the organization’s goal in This study contributes to the theory as it extends
the use of Web services. the current literature of Web services to include the
• Making the strategy everyone’s daily balanced scorecard framework of measure to its
job: Refers to the linking of the employees usage and quality. Further, the study contributes
with the business unit and the organiza- to practice as small businesses using Web services
tion’s strategy. will benefit from the balanced scorecard as they
• Making the strategy an ongoing process: will have a system which provides them with
Refers to proposing a process for manag- timely, cost-effective, scalable, manageable and
ing the strategy by integrating the tactical reliable feedback on their strategic performance.
management and the strategic management The business issues including effective organi-
in the use of Web services. zational performance and successful strategy
• Promoting change through management implementation will be greatly enhanced. Further,
leadership: Refers to the involvement the balanced scorecard gives a holistic view of
of the management team in the change the small business by simultaneously examining
process. its performance from four perspectives. It is able
to translate an organization’s vision and strategy
into specific objectives, monitored through a
Conclusion coherent set of performance indicators. Future
research should aim to apply the balanced score-
In this chapter we discussed the role of Web card framework for Web services via a survey or
services usage and quality, applying the bal- multiple case studies with firms across different
anced scorecard methodology. We developed a sections in the industry so that the findings can
framework that presented the balanced scorecard’s contribute to a generic framework.
four business perspectives (i.e., learning and

873
The Role of Web Services

References Niven, P. R. (2003). Balanced scorecard. Step-


by-step for government and non-profit agencies.
Benko, C., & McFarlan, F. W. (2003). Connect- John Wiley & Sons.
ing the dots: Aligning projects with objectives in
unpredictable times. Boston: Harvard Business Ortega, M., Perez, M., & Rojas, T. (2000). A model
School Press. for software product quality with a system focus.
Paper presented at the 4th World Multi Conference
Chan, S. S., Kellen, V., & Nair, C. (2004). Adopting on Systematic Cybernatics.
Web services for B2B e-collaboration: Research
directions. Paper presented at the International Solano, J., De Ovalles, M. P., Rojas, T., Padua, A.
Resources Management Association Conference, G., & Morales, L. M. (2003, Winter). Integration
Innovations through Information Technology (pp. of systemic quality and the balanced scorecard,
1191-1193). privacy and security in e-business. Information
Systems Management, 66–81. doi:10.1201/1078
Hagel, J. (2002). Edging into Web services. The /43203.20.1.20031201/40086.9
McKinsey Quarterly, 4.
Yen, V. C. (2004). Applications development with
Hagel, J. III, & Brown, J. S. (2001). Your next Web services. Paper presented at the International
IT strategy. Harvard Business Review, 79(9), Resources Management Association Conference,
105–115. Innovations through Information Technology (pp.
Kaplan, R., & Norton, D. (2000). The strategy 875-876).
focused organization. Harvard Business School Yin, R. K. (1994). Case study research: Design
Press. and methods (2nd ed.). Thousand Oaks, CA: Sage
Publications.

874
The Role of Web Services

Appendix: Case Study Questionnaire

A) Demographic Section

1. Name of your organization?


2. Your job title?
3. How long have you worked in your present organization?
A. Less than a year
B. Between 1-5 years
C. Between 6-10 years
D. Between 11-20 years
E. Over 20 years
4. How long have you had your present position?
A. Less than a year
B. Between 1-2 years
C. Between 3-5 years
D. Between over 5 years
5. What is your organization’s reach?
A. Local
B. Regional
C. National
D. Global
6. What is your organization’s size?
A. Large > 500 employees
B. Small-Medium-Enterprise 1-499 employees
7. Please name the type of industry your organization is or provide the NAICS code?
8. Please indicate which of the following terms best describes your organization’s main business
activity
A. Retail/wholesale trade
B. Manufacturing/distribution
C. Computers/communications
D. Financial services
E. Education
F. Health
G. Government services
H. Other services
I. Other ---------------------------------------------------------------------
9. What is the main role of your organization?
A. Buyer
B. Seller
C. Manufacturer
D. Supplier
10. Please indicate which of the following types of business transactions are actively supported by the
IT systems in your organization?

875
The Role of Web Services

A. Purchase orders
B. Invoices
C. Advance shipping notices
D. Product information
E. Payment transactions
11. Please indicate the type of business applications and tools your organization implemented, or will
be implementing?
A. EDI – Value-Added-Network
B. Internet-based EDI
C. Extranets
D. Intranets
E. E-mail
F. B2C or B2B shopping cart
G. Other types of Web service applications
H. Other ------------------------------------------------------------------------------
12. Please indicate the number of business partners your organization has? (business partners refer to
those who are contracted legally to do business with the firm)
A. 1-20
B. 21-50
C. 51-100
D. 101-499
E. Over 500
13. Approximately what is your annual revenue, in millions? (please estimate)
A. 0-1m
B. 1-10m
C. 10-100m
D. 100-500,
E. 500-1,000m
F. Over 1,000m
14. Approximately how much does your organization spend annually on information technology?
(please estimate)
A. 0-100,000
B. 100,000-500,000
C. 500,000-1m
D. 1-5m
E. Over 5m
15. Please indicate how your organization chose your business partners?
A. Advertising on your Web page
B. Screening of business partners
C. Based on past reputations
D. Other ---------------------------------------------------------------------------------
16. When did your organization start implementing a balanced scorecard system? Please indicate the
year.

876
The Role of Web Services

17. Please indicate the importance of implementing a balanced scorecard system in your organization’s
current business strategy
A. No importance/not considered
B. Small consideration
C. Nominally part of strategy
D. Substantially part of strategy
E. Crucial to strategy
18. On average how often does your organization meet face-to-face with your business partners?
A. At least once per week
B. At least once per month
C. More than once a month
D. Other --------------------------------------------------------------------------------
19. Please indicate how your organization maintains your business partners
A. Verbal agreements
B. Legal business partner agreements (written)
C. Screening of business partners (based on a performance assessment)
D. Other ---------------------------------------------------------------------------------------

Questions on your Firm’s Application of the


Balanced Scorecard on Web Services

How did your firm implement a balanced scorecard methodology?What is the vision of your organization
towards the use of Web services?Do your employees have a solid understanding of your firm’s use of
Web services?Why was the balanced scorecard important in measuring your business performance?What
benefits did your firm experienced from the quality of Web service applications?What risks/challenges
did your firm experienced from using Web services?

Learning and Growth Perspective

Do small business employees have the skills/competencies to operate the Web service application and
align it with their internal business processes effectively in order to meet their customers’ objectives of
using Web services?
Do small business employees possess the information required to achieve objectives?
Do small businesses have the ability to sustain growth and change that in turn enhances the quality
of Web services?
How does the balanced scorecard impact the quality of Web services in the learning and growth
perspective?
How was the data gathered, analyzed, evaluated for feedback so that appropriate actions can be taken
to improve the learning and growth perspective?
What skills do your employees possess regarding performance management and more specifically,
developing performance measures in the balanced scorecard framework?
(This question relates to what qualification requirements do the senior employees who undertake the
role of a balanced scorecard auditor or quality control manager has? The purpose is to ensure that they
have the adequate skills to completely measure the firm’s performance).

877
The Role of Web Services

Do your employees have the right skills/competencies to operate the Web service applications?
Do your employees have the tools and information they require to make effective decisions that impact
customer outcomes?
Do your employees possess the information required to achieve objectives?
Does your organization have the ability to sustain growth and change?
Do your employees face difficulties when accessing critical information needed to serve customers?

Internal Business Process Perspective

Does the internal business processes applying Web services conform to the mission of small busi-
nesses?
Does the internal business processes meet their customer requirements?
What must small businesses do well internally in order to achieve the objectives they set forth to
achieve quality in Web services?
Where does the Web services “process” start, and where does it end?
How does the balanced scorecard impact the quality of Web services in the internal business processes?
How were the data gathered, analyzed, evaluated for feedback so that appropriate actions can be taken to
improve the internal business process perspective?
Do your employees know how their day to day actions contribute to the organization’s success? (how
and why)
What must your firm do well internally in order to achieve the objectives set forth in the customer
perspective?

Customer Perspective

What do customers expect or demand from the use of Web services? Dimensions of customers experience
include; time, quality, price or cost, accessibility, reputation, and relationship.
Who do we define as our customers?
How do our customers see us? How does Web services create value for our customers?
How does the balanced scorecard impact the quality of Web services and its relationship with custom-
ers’ perspectives? What factors are evaluated in the customers’ perspective?
How the data was gathered, analyzed, evaluated for feedback so that appropriate actions can be taken
to improve the customer perspective?
Who does your firm define as your customers?
What do your customers expect or demand from your firm? (factors pertaining to time, quality, price
or cost, accessibility, reputation, relationship and image).
Do you require anything from your customers? (in order to meet your customers demands – is there
anything you need from them?)
You have a unique value proposition for customers (for example cost, technical superiority, customer
intimacy).
What factors does your firm excel in as in evidence of past successes thereby providing accountability
to your stakeholders that you are satisfying customer expectations?
(This question relates to what factors do the firm excel in as in evidence of past successes thereby
providing accountability)

878
The Role of Web Services

Financial Perspective

What accountability do small businesses that use Web services have to financial stakeholders?
How does the balanced scorecard impact the quality of Web services performance from a financial
perspective? What factors are evaluated in the financial perspective?
How are the data gathered, analyzed, evaluated for feedback so that appropriate actions can be taken
to improve the financial perspective?
What reporting mechanisms are deployed in your firm in order to enforce accountability?
Does your firm create significant value from intangible assets such as employee knowledge and in-
novation, customer relationships, and a strong culture. (how and why)
Does your senior management team spend time together discussing variances from plan and other
financial issues? (How often?)
Does your organization clearly define the performance targets for both financial and non-financial
indicators? (how and why)

This work was previously published in Managing Web Service Quality: Measuring Outcomes and Effectiveness, edited by K.
Khan, pp. 342-357, copyright 2009 by Information Science Reference (an imprint of IGI Global).

879
880

Chapter 4.2
Semantic Web Take-Off in a
European Industry Perspective
Alain Léger
France Telecom R&D, France

Johannes Heinecke
France Telecom R&D, France

Lyndon J.B. Nixon


Freie Universität Berlin, Germany

Pavel Shvaiko
University of Trento, Italy

Jean Charlet
STIM, DPA/AP-Hopitaux Paris & Université Paris 6, France

Paola Hobson
Motorola Labs, UK

François Goasdoué
LRI, CNRS et Université Paris Sud XI, France

Abstract complementary tasks. The spectrum of applications


is extremely large and to name a few: corporate
Semantic Web technology is being increasingly ap- portals and knowledge management, e-commerce,
plied in a large spectrum of applications in which e-work, e-business, healthcare, e-government,
domain knowledge is conceptualized and formalized natural language understanding and automated
(e.g., by means of an ontology) in order to support translation, information search, data and services
diversified and automated knowledge processing integration, social networks and collaborative filter-
(e.g., reasoning) performed by a machine. Moreover, ing, knowledge mining, business intelligence and
through an optimal combination of (cognitive) hu- so on. From a social and economic perspective, this
man reasoning and (automated) machine process- emerging technology should contribute to growth
ing (mimicking reasoning); it becomes possible in economic wealth, but it must also show clear cut
for humans and machines to share more and more value for everyday activities through technological
transparency and efficiency. The penetration of Se-
DOI: 10.4018/978-1-60566-066-0.ch001 mantic Web technology in industry and in services

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Semantic Web Take-Off in a European Industry Perspective

is progressing slowly but accelerating as new to leverage existing company information in new
success stories are reported. In this chapter we ways. Present research has turned to focusing on
present ongoing work in the cross-fertilization the seamless integration of heterogeneous and dis-
between industry and academia. In particular, we tributed applications and services (both intra- and
present a collection of application fields and use inter-enterprise) through Semantic Web Services,
cases from enterprises which are interested in the and respectful of the legacy systems already in
promises of Semantic Web technology. The use place, with the expectation of a fast return on
cases are focused on the key knowledge process- investment (ROI) and improved efficiency in
ing components that will unlock the deployment e-work and e-business.
of the technology in industry. The chapter ends This new technology takes its roots in the cogni-
with the presentation of the current state of the tive sciences, machine learning, natural language
technology and future trends as seen by prominent processing, multi-agents systems, knowledge
actors in the field. acquisition, automated reasoning, logics and deci-
sion theory. It can be separated into two distinct
– but cooperating fields - one adopting a formal
Current Situation and algorithmic approach for common sense
automated reasoning (automated Web), and the
As a result of the pervasive and user-friendly other one “keeping the human being in the loop”
digital technologies emerging within our in- for a socio-cognitive Semantic Web (automated
formation society, Web content availability is social Web) which is gaining momentum today
increasing at an incredible rate but at the cost of with the Web 2.0 paradigm1.
being extremely multiform, inconsistent and very On a large scale, industry awareness of Se-
dynamic. Such content is totally unsuitable for mantic Web technology has started at the EC level
machine processing, and so necessitates too much with the IST-FP5 thematic network Ontoweb2
human interpretation and its respective costs in [2001-2004] which brought together around 50
time and effort for both individuals and companies. motivated companies worldwide. Based on this
To remedy this, approaches aim at abstracting experience, within IST-FP6, the Network of Ex-
from this complexity (i.e., by using ontologies) cellence Knowledge Web3 [2004-2008] made an
and offering new and enriched services able to in-depth analysis of the concrete industry needs
process those abstractions (i.e., by mechanized in key economic sectors, and in a complementary
reasoning) in a fully – and trusted - automated way the IST-FP6 Network of Excellence REW-
way. This abstraction layer is the subject of a ERSE4 was tasked with providing Europe with
very dynamic activity in research, industry and leadership in reasoning languages, also in view
standardization which is usually called “Semantic of a successful technology transfer and aware-
Web” (see for example, DARPA, European IST ness activities targeted at the European industry
Research Framework Program, W3C initiative, for advanced Web systems and applications. This
OASIS). The initial application of Semantic Web impetus will continue and grow up in the EU IST-
technology has focused on Information Retrieval FP7 [2007-2013]5.
(IR) where access through semantically annotated The rest of the chapter is organized as follows.
content, instead of classical (even sophisticated) Four prototypical application fields are presented
statistical analysis, aimed to give far better results in Section 2, namely (1) healthcare and biotech-
(in terms of precision and recall indicators). The nologies, (2) knowledge management (KM), (3)
next natural extension was to apply IR in the in- e-commerce and e-business, and finally, (4) mul-
tegration of enterprise legacy databases in order timedia and audiovisual services. Finally, Section

881
Semantic Web Take-Off in a European Industry Perspective

3 reports on a current vision of the achievements involved in many initiatives of the Knowledge
and some perspectives are given. Web Industry Area, including the collection of
business use cases and their evaluation. In order to
Overall Business Needs and more directly achieve close co-operation between
Key Knowledge Processing researchers and industry, each research activity in
Requirements the network was invited to select a use case whose
requirements closely correlated to what would be
Use Case Collection achieved in their research work. Results have been
collected and reported in July 20076.
In order to support a large spectrum of applica- Currently in 2007, this Industry Board con-
tion fields, two EU FP6 Networks of Excellence sisted of about 50 members (e.g., France Telecom,
NoE-Knowledge Web and NoE-REWERSE are British Telecom, Institut Français du Pétrole, Illy
tasked with promoting transfer of best-of-the-art Caffè, Trenitalia, Daimler AG, Thalès, EADS,
knowledge-based technology from academia to …) from across 14 nations and 13 economic
industry. The networks are made up of leading sectors (e.g., telecoms, energy, food, logistics,
European Semantic Web research institutions that automotive,...).
co-ordinate their research efforts while parallel The companies were requested to provide
efforts are made in Semantic Web education to illustrative examples of actual or hypothetical
increase the availability of skilled young research- deployment of Semantic Web technology in
ers and practitioners and last but not the least, in their business settings. This was followed up
pushing the take-up in Business and Industry. with face-to-face meetings between researchers
In order to accelerate the transfer from research and industry experts from the companies to gain
to industry, the objective of an Industry-Research additional information about the provided use
co-operation is to establish a working relationship cases. Thus, in 2004, a total of 16 use cases were
between Semantic Web researchers and an industry collected from 12 companies. In 2007, through
partner, in which research results being produced many workshops and Industry forum sessions at
in an area of Semantic Web research will be proto- major Semantic Web conferences, more than a
typically applied to the industry partner’s selected hundred use cases were available or illustrative
business case. The co-operation not only seeks to of the current trend to introduce Semantic Web
achieve an individual success story in terms of technology in the main stream.
some specific research and a given business case, As shown in Figure 1, where the use cases
but also to establish the value of Semantic Web are broken down according to the industry sec-
technologies to industrial application in a more tor, collected cases are from 9 industry sectors,
general sense. It achieves this by demonstrating with the highest number of the use cases coming
the use of Semantic Web technology in a business from the service industry (19%) and media &
setting, exploring their usefulness in solving busi- communications (18%) respectively. This initial
ness problems and ensuring future applicability by collection of use cases can be found in (Nixon L.
directing researchers towards meeting industrial et al., 2004), and an updated selection is available
requirements in their work. on the Knowledge Web Industry portal7.
In NoE-Knowledge Web, an Industry Board The co-operations have been a very challeng-
was formed at the beginning of the network to bring ing activity, given the early state of much cutting
together potential early adopters of Semantic Web edge Semantic Web research and the differences
technologies from across a wide spread of indus- in perspective between academia and business.
try sectors. Industry Board members have been However, successes have been reported, not only

882
Semantic Web Take-Off in a European Industry Perspective

Figure 1. Breakdown of use cases by industry sector

in the production of some prototypical solutions Use Case Analysis


and demos which can be shown to industry and
business top managers, but also in making re- A preliminary analysis of the use cases has been
searchers more aware of the importance of their carried out in order to obtain a first vision (end
work to solving business problems and the earlier of 2004) of the current industrial needs and
recognition by academics of industry needs and to estimate the expectations from knowledge-
expectations and so integrating them to their based technology with respect to those needs.
research agenda. The industry experts were asked to indicate the
Hence, the Industry-Research co-operations existing legacy solutions in their use cases, the
in NoE-Knowledge Web and NoE-REWERSE service functionalities they would be offered and
must be seen as a significant first attempt to align the technological locks they encountered, and
the ambitious cutting edge work on Semantic eventually how they expected that Semantic Web
Web technologies done by leading researchers technology could resolve those locks. As a result,
in Europe and the real world business problems this analysis has provided an overview of:
encountered by the European industry which may
find a potential solution in those same Semantic • Types of business or service problems
Web technologies. Given a continued rise in where the knowledge-based technology is
awareness among Semantic Web researchers of considered to bring a plausible solution.
the applicability of their work to industry and the • Types of technological issues (and the cor-
continued rise in awareness among industry of the responding research challenges) which
potential of the work of Semantic Web research- knowledge based technology is expected
ers, which has been begun in IST-NoEs, in IST- to overcome.
R&D projects, but also clearly in industry (SMEs
and large companies), the technology transfer is Figure 2 shows a breakdown of the areas in
gradually evolving. which the industry experts thought Semantic
Web technology could provide a solution. For
example, for nearly half of the collected use cases,
data integration and semantic search were areas

883
Semantic Web Take-Off in a European Industry Perspective

Figure 2. Breakdown of use cases by industry sector

where industry was looking for knowledge-based knowledge extraction, i.e., populating ontologies
solutions. Other areas mentioned, in a quarter of by extracting data from legacy systems; and (3)
use cases, were solutions to data management and ontology mapping, i.e., resolving semantic het-
personalization. erogeneity among multiple ontologies.
Figure 3 shows a breakdown of the technology Below, an illustration is given, with the help
locks identified in the use cases. There are three of a use case from the collection, how a concrete
technology locks which occur the most often in business problem can be used to indicate the
the collected use cases. These are: (1) ontology technology locks for which knowledge-based
development, i.e., modeling of a business do- solutions potentially might be useful. This use
main, authoring, reusing existing ontologies; (2) case addresses the problem of intelligent search of

Figure 3. Preliminary vision of technology locks in use case

884
Semantic Web Take-Off in a European Industry Perspective

documents in the corporate data of an international multiple ontologies (e.g., by taking the
coffee company. union of the axioms) with respect to evolv-
The company generates a large amount of ing business case requirements, see (Dou
internal data and its employees encounter difficul- D. et al., 2005) (McGuiness D. et al.,
ties in finding the data they need for the research 2000) (Protégé8), OAEI-2007 Ontology
and development of new solutions. The aim is to Alignment Evaluation Initiative9, NeOn10
improve the quality of the document retrieval and (Networked Evolving Ontologies) and
to enable personalization services for individual Ontology Matching survey site11.
users when searching or viewing the corporate • Ontology matching, matching results anal-
data. As technology locks, the expert mentioned ysis, producing explanations and match
here the corporate domain ontology development manager. These tasks and component are
and maintenance, and semantic querying. in charge of (on-the-fly and semi-automat-
Eventually, this analysis (by experts estima- ic) determination of semantic mappings
tions) has provided with a preliminary under- between the entities of multiple schemas
standing of scope of the current industrial needs and ontologies, see (Rahm E. et al., 2001)
and the current concrete technology locks where (Shvaiko P. and Euzenat, 2005), (Euzenat J.
knowledge-based technology is expected to pro- and Shvaiko P., 2007). Mappings are typi-
vide a plausible solution. However, to be able to cally specified with the help of a similar-
answer specific industrial requirements, there is ity relation which can be either in the form
the need to conduct further a detailed technical of a coefficient rating match quality in the
analysis of the use cases, thereby associating (0,1] range (i.e., the higher the coefficient,
to each technology lock a concrete knowledge the higher the similarity between the enti-
processing task and a component realizing its ties, see (Billig A. et al., 2002) (Ehrig M. et
functionalities. al., 2004) (Euzenat J. et al., 2004) (Do H.
H. et al., 2002) (Zhong J. et al., 2002) or in
Knowledge Processing the form of a logical relation (e.g., equiva-
Tasks and Components lence, subsumption), see (Giunchiglia F. et
al., 2003) (Giunchiglia F. et al., 2004). The
Based on the information processing needs mappings might need to be ordered accord-
identified during the technical use cases analysis ing to some criteria, see (Di Noia T. et al.,
(Shvaiko P. et al., 2004), a typology of common 2003) (Do H. H. et al., 2002).
knowledge processing tasks and a library of high
level components for realizing those tasks, was Finally, explanations of the mappings might
built, see Table 1. The first tentative typology be also required, see (Dhamankar R. et al., 2004)
includes twelve processing tasks. Let us discuss (Shvaiko P. et al., 2005). Matching systems may
knowledge processing tasks and components of produce mappings that may not be intuitively obvi-
Table 1 in more detail: ous to human users. In order for users to trust the
mappings (and thus use them), they need informa-
• Ontology management, ontology merg- tion about them. They need access to the sources
ing and ontology manager. These tasks that were used to determine semantic correspon-
and component are in charge of ontology dences between terms and potentially they need to
maintenance (e.g., reorganizing taxono- understand how deductions and manipulations are
mies, resolving name conflicts, browsing performed. The issue here is to present explanations
ontologies, editing concepts) and merging in a simple and clear way to the user.

885
Semantic Web Take-Off in a European Industry Perspective

• Data translation and wrapper. This task should result in re-combinations of busi-
and component is in charge of automatic nesses the technology provides and so will
manipulation (e.g., translation, exchange) have a profound impact on business and
of instances between heterogeneous infor- economic workflows.
mation sources storing their data in differ- • Content annotation and annotation man-
ent formats (e.g., RDF, SQL DDL, XML ager. This task and component is in charge
…), see (Hull R. 1997) (Petrini J. et al., of automatic production of metadata for
2004) (Velegrakis Y. et al., 2005) (Halevy the contents, see aceMedia12 for multime-
A. et al., 2006). Here, mappings are tak- dia annotation. Annotation manager takes
en as input (for example, from the match as input the (pre-processed) contents and
manager component) and are the support domain knowledge and produces as output
for generating query expressions that per- a database of content annotations. In addi-
form the required semantic and syntactical tion to the automatic production of content
manipulations with data instances coming metadata, prompt mechanisms offer the
from heterogeneous environment. user the possibility to enrich the content
• Results reconciliation and results reconcil- annotation by adding extra information
er. This task and component is in charge of (e.g., title, name of a location, title of an
determining an optimal solution, in terms event, names of people) that could not be
of contents (no information duplication, automatically detected.
etc.) and routing performance, for return- • Automated reasoning. This task and com-
ing results from the queried information ponent is in charge of providing logical
sources, see (Preguica N. et al., 2003). reasoning services (e.g., subsumption, con-
• Composition of Web services and planner. cept satisfiability, instance checking tests),
This task and component is in charge of see (Haarslev V. et al., 1999-2007). For
automated composition of Web services example, when dealing with multimedia
into executable processes (Orchestration). annotations, logical reasoning can be ex-
Composed Web services perform new ploited in order to check consistency of the
functionalities by specific on demand in- annotations against the set of spatial (e.g.,
teraction with pre-existing services that are left, right, above, adjacent, overlaps) and
published on the Web, see surveys from temporal (e.g., before, after, during, co-
(Chan et al., 2007) (Berardi et al., 2005) start, co-end) constraints. This can certify
(Hull et al., 2005) (Pistore et al., 2005) that the objects detected in the multime-
(Roman et al., 2005) (Traverso P. et al., dia content correspond semantically to the
2004) (Cardoso et al., 2003) (McIlraith et concepts defined in the domain ontology.
al., 2001). From a business viewpoint, it For example, in the racing domain, the
remains a key challenge to be overcome, automated reasoner should check whether
as the businesses react very positively to a car is located on a road or whether the
the need for a very effective integration grass and sand are adjacent to the road.
technology and for more agility in a very • Semantic query processing and query
competitive worldwide economy. In the processor. This task and component is in
meantime, reducing interoperability prob- charge of rewriting a query posed by a hu-
lems will open opportunities for easier man being or a machine, by using terms
innovative solutions and for the increase which are explicitly specified in the model
in cooperation between enterprises. This of domain knowledge in order to provide

886
Semantic Web Take-Off in a European Industry Perspective

Table 1. Typology of knowledge processing tasks & components

N Knowledge processing tasks Components


1 Ontology Management Ontology Manager
2 Ontology Matching Match Manager
3 Ontology Matching results Analysis Match Manager
4 Data Translation Wrapper
5 Results Reconciliation Results Reconciler
6 Composition of Web Services Planner
7 Content Annotation Annotation manager
8 Reasoning Reasoner
9 Semantic Query Processing Query Processor
10 Ontology Merging Ontology Manager
11 Producing explanations Match Manager
12 Personalization Profiler

semantics preserving query answering, see consistent delivery of the contents to a


(Mena E. et al., 1996) (Halevy et al., 2001) broad range user terminals (e.g., PDA, mo-
(Calvanese et al., 2002) (IST-IP aceMedia bile phone, portable PC).
2004). Examples of queries are “Give me
all the games played on grass” or “Give me Key Application Sectors and
all the games of double players”, in the ten- Typical Technology Problems
nis domain. Finally, users should be able to
query by sample content e.g. an image. In Healthcare and Biotechnologies
this case, the system should perform an in-
telligent search of images and videos (e.g., The medical domain is a favourite target for Se-
by using semantic annotations) where, for mantic Web applications just as the expert system
example, the same event or type of activity was for artificial intelligence applications 20 years
takes place. ago. The medical domain is very complex: medical
• Personalization and profiler. This task and knowledge is difficult to represent in a computer
component is in charge of tailoring servic- format, making the sharing of information even
es available from the system to the speci- more difficult. Semantic Web solutions become
ficity of each user, see (Antoniou G. et al., very promising in this context.
2004). For example, generation and updat- One of the main mechanisms of the Seman-
ing of user profiles, recommendation gen- tic Web - resource description using annotation
eration, inferring user preferences, and so principles - is of major importance in the medical
on. For example users might want to share informatics (or sometimes called bioinformat-
annotations within trusted user networks, ics) domain, especially as regards the sharing of
thus having services of personal metadata these resources (e.g. medical knowledge on the
management and contacts recommenda- Web or genomic database). Through the years,
tion. Also, a particular form of personal- the IR area has been developed by medicine:
ization, which is media adaptation, may medical thesauri are enormous (e.g., more than
require knowledge-based technology and 1,600,000 terms in Unified Medical Language

887
Semantic Web Take-Off in a European Industry Perspective

System, UMLS13) and are principally used for several databases. The mediator of NEUROBASE
bibliographic indexation. Nevertheless, the MeSh has been tested on decision support systems in
thesaurus (Medical Subject Heading) or UMLS epilepsy surgery.
have been used to provide data semantics with
varying degrees of difficulty. Finally, the Web Ontologies for Coding Systems
services technology allows us to imagine some
solutions to the interoperability problem, which is The main usage of ontologies in medical domain
substantial in medical informatics. Below, we will is as index of coding system: after using thesauri
describe current research, results and expected for indexing medical bibliography (PubMed with
perspectives in these biomedical informatics the Mesh17), the goal is to index Electronic Health
topics in the context of Semantic Web. records with medical concept in order to enhance
information retrieval or to allow epidemiological
Biosciences Resources Sharing studies. For that purpose, several countries intend
to use the SNOMED, an international classification
In the functional genomics domain, it is necessary of concepts organized in eight axes (Spackman
to have access to several databases and knowledge et al., 2002). Except the problem of languages,
bases which are accessible separately on the Web this classification exists in two versions: a clas-
but are heterogeneous in their structure as well sification of 160,000 concepts (SNOMED-I V3.5)
as in their terminology. Among such resources, and an ontology, which is the evolution of the
we can mention SWISSPROT14 where the gene preceding one, of 330,000 concepts, SNOMED
products are annotated by the Gene Ontology15, CT. The use of ontologies of such a size is dif-
Gen-Bank16, etc. When comparing these resources ficult. Some authors describe them as Reference
it is easy to see that they propose the same infor- Ontology which cannot be accessed without an
mation in different formats. The XML language, interface ontology (Rosenbloom et al., 2006).
which acts as a common data structure for the Notwithstanding, UK national health system
different knowledge bases, provides at most (NHS) is integrating SNOMED CT and it will be
a syntactic Document Type Definition (DTD) interesting to examine the results of this industrial
which does not resolve the semantic interoper- deployment18.
ability problem.
One of the solutions comes from the Semantic Web Services for Interoperability
Web with a mediator approach (Wiederhold G.,
1992) which allows for the accessing of different The Web services technology can propose some
resources with an ontology used as the Interlin- solutions to the interoperability problematic. We
gua pivot. For example and in another domain describe now a new approach based on a “patient
than that of genomics, the NEUROBASE project envelope” and we conclude with the implementa-
(Barillot C. et al., 2003) attempts to federate dif- tion of this envelope based on the Web services
ferent neuro-imagery information bases situated technology.
in different clinical or research areas. The proposal The patient envelope is a proposition of the
consists of defining an architecture that allows Electronic Data Interchange for Healthcare group
the access to and the sharing of experimental (EDI-Santé19) with an active contribution from the
results or data treatment methodologies. It would ETIAM20 society. The objective of the work is on
be possible to search in the various data bases filling the gap between “free” communication,
for similar results or for images with peculiari- using standard and generic Internet tools, and
ties or to perform data mining analysis between “totally structured” communication as promoted

888
Semantic Web Take-Off in a European Industry Perspective

by CEN (in the Working Group IV “Technol- be fundamental to maintain the existing links
ogy for Interoperability”21) or HL722. After the between elements, to transmit them together,
worldwide analysis of existing standards, the e.g., a scanner and the associated report, and
proposal consists of an “intermediate” structure to be able to prove it.
of information, related to one patient, and storing
the minimum amount of data (i.e. exclusively The interest of such an approach is that it pre-
useful data) to facilitate the interoperability be- pares the evolution of the transmitted document
tween communicating peers. The “free” or the from a free form document (from proprietary ones
“structured” information is grouped into a folder to normalized ones as XML) to elements respecting
and transmitted in a secure way over the existing HL7v3 or EHRCOM data types. In France, the
communication networks (Cordonnier E. et al., GIP-DMP23 retains such an approach (in conjunc-
2003). This proposal has reached widespread adop- tion with the Clinical Document Architecture of
tion with the distribution by Cegetel.rss of a new HL724) for the implementation of the exchanges of
medical messaging service, called “Sentinelle”, the Dossier Médical Personnel (a future national
fully supporting the patient envelope protocol electronic health record).
and adapted tools.
After this milestone, EDI-Santé is promoting What is Next in the
further developments based on ebXML and SOAP Healthcare Domain?
(Simple Object Access Protocol) in specifying
exchange (see items 1 and 2 below) and medical These different projects and applications highlight
(see items 3 and 4 below) properties: the main consequence of the Semantic Web being
expected by the medical communities: the sharing
1. Separate what is mandatory to the transport and integration of heterogeneous information or
and management of the message (e.g., patient knowledge. The answers to the different issues
identification from what constitutes the “job” are the use of mediators, a knowledge-based
part of the message. system, and ontologies, which should be based
2. Provide a “container” for the message, col- in the mid term on normalized languages such as
lecting the different elements, texts, pictures, RDF, OWL but also in addition to come OWL-S,
videos, etc. SAWSDL, WSML, SWRL, or RuleML. The work
3. Consider the patient as the unique object of of the Semantic Web community must take into
the transaction. Such an exchange cannot account these expectations, see for example the
be anonymous. It concerns a sender and an FP6 projects25,26,27. Finally, it is interesting to note
addressee who are involved in the exchange that the Semantic Web is an integrated vision of
and who are responsible. A patient can de- the medical community’s problems (thesauri, on-
mand to know the content of the exchange tologies, indexation, and inference) and provides a
in which (s)he is the object, which implies real opportunity to synthesize and reactivate some
a data structure which is unique in the form research directions (Charlet J. et al., 2002).
of a triple {sender, addressee, patient}.
4. The conservation of the exchange semantics.
The information about a patient is multiple in
the sense that it comes from multiple sources
and has multiple forms and supporting data
(e.g., database, free textual document, semi-
structured textual document, pictures). It can

889
Semantic Web Take-Off in a European Industry Perspective

Knowledge Management mantics, and are therefore machine-interpretable.


Although the first KM approaches and solutions
Leveraging Knowledge have shown the benefits of ontologies and related
Assets in Companies methods, a large number of open research issues
still exist that have to be addressed in order to
Knowledge is one of the key success factors for make Semantic Web technology a complete
enterprises, both today and in the future. There- success for KM solutions:
fore, company knowledge management (KM)
has been identified as a strategic tool. However, • Industrial KM applications have to avoid
if for KM, information technology is one of any kind of overhead as far as possible. A
the foundational elements, KM in turn, is also seamless integration of knowledge creation
interdisciplinary by its nature. In particular, it (i.e., content and metadata specification)
includes human resource management as well and knowledge access (i.e., querying or
as enterprise organization and culture28. KM is browsing) into the working environment is
viewed as the management of the knowledge required. Strategies and methods are need-
arising from business activities, aiming at le- ed to support the creation of knowledge,
veraging both the use and the creation of that as side effects of activities that are car-
knowledge for two main objectives: capitaliza- ried out anyway. These requirements mean
tion of corporate knowledge and durable innova- emergent semantics that can be supported
tion fully aligned with the strategic objectives through ontology learning, which should
of the organization. reduce the current time consuming task of
Conscious of this key factor of productivity building-up and maintaining ontologies.
in an ever faster changing ecosystem, the Eu- • Access to as well as presentation of knowl-
ropean KM Framework (CEN/ISSS29, Knowl- edge has to be context-dependent. Since
edgeBoard30) has been designed to support a the context is setup by the current busi-
common European understanding of KM, to ness task, and thus by the business process
show the value of this emerging approach and being handled, a tight integration of busi-
to help organizations towards its successful ness process management and knowledge
implementation. The Framework is based on management is required. KM approaches
empirical research and practical experiences can provide a promising starting point for
in this field from all over Europe and the rest smart push services that will proactively
of the world. The European KM Framework deliver relevant knowledge for carrying
addresses all of the relevant elements of a KM out the task at hand more effectively.
solution and serves as a reference basis for all • Conceptualization has to be supplemented
types of organizations, which aim to improve by personalization. On the one hand, tak-
their performance by handling knowledge in a ing into account the experience of the user
better way. and his/her personal needs is a prerequisite
in order to avoid information overload, and
Benefits of Knowledge-Based KM on the other hand, to deliver knowledge at
the right level of granularity and from the
The knowledge backbone is made up of ontolo- right perspective at the right time.
gies that define a shared conceptualization of
an application domain and provide the basis for The development of knowledge portals serving
defining metadata that have precisely defined se- the needs of companies or communities is still a

890
Semantic Web Take-Off in a European Industry Perspective

manual process. Ontologies and related metadata standards are needed to realize such integration.
provide a promising conceptual basis for generat- Knowledge management is obviously a very
ing parts of such knowledge portals. Obviously, promising area for exploiting Semantic Web
among others, conceptual models of the domain, of technology. Document-based portals KM solu-
the users and of the tasks are needed. The genera- tions have already reached their limits, whereas
tion of knowledge portals has to be supplemented semantic technology opens the way to meet KM
with the (semi-) automated evolution of portals. requirements in the future.
As business environments and strategies change
rather rapidly, KM portals have to be kept up-to- Knowledge-Based KM Applications
date in this fast changing environment. Evolution
of portals should also include some mechanisms In the context of geographical team dispersion,
to ‘forget’ outdated knowledge. multilingualism and business unit autonomy, usu-
KM solutions will be based on a combina- ally a company wants a solution allowing for the
tion of intranet-based functionalities and mobile identification of strategic information, the secured
functionalities in the very near future. Semantic distribution of this information and the creation
technologies are a promising approach to meet the of transverse working groups. Some applicative
needs of mobile environments, like location-aware solutions allowed for the deployment of an Intranet
personalization and adaptation of the presentation intended for all the marketing departments of the
to the specific needs of mobile devices, i.e., the company worldwide, allowing for a better division
presentation of the required information at an of and a greater accessibility to information, but
appropriate level of granularity. In essence, em- also capitalisation on the total knowledge. There
ployees should have access to the KM application are four crucial points that aim at easing the work
anywhere and anytime. of the various marketing teams in a company: (1)
Peer-to-peer computing (P2P), social net- Business intelligence, (2) Skill and team manage-
working (W2.0), combined with Semantic Web ment31, (3) Process management32 and (4) Rich
technology, will be a strong move towards getting document access and management33.
rid of the more centralized KM approaches that are Thus, a system connects the “strategic ontolo-
currently used in ontology-based solutions. W2.0 gies” of the company group (brands, competitors,
scenarios open up the way to derive consensual geographical areas, etc.) with the users, via the
conceptualizations among employees within an automation of related processes (research, classi-
enterprise in a bottom-up manner. fication, distribution, knowledge representation).
Virtual organizations are becoming more and The result is a dynamic semantic system of navi-
more important in business scenarios, mainly gation (research, classification) and collaborative
due to decentralization and globalization. Obvi- features.
ously, semantic interoperability between different At the end from a functional point of view, a
knowledge sources, as well as trust, is necessary KM system organises skill and knowledge man-
in inter-organizational KM applications. agement within a company in order to improve
The integration of KM applications with interactivity, collaboration and information shar-
e-learning (e.g., skill management in compa- ing. This constitutes a virtual workspace which
nies) is an important field that enables a lot of facilitates work between employees that speak
synergy between these two areas. KM solutions different languages, automates the creation of work
and e-learning must be integrated from both an groups, organises and capitalises structured and
organizational and an IT point of view. Clearly, unstructured, explicit or tacit data of the company,
interoperability and integration of (metadata) and offers advanced features of capitalisation

891
Semantic Web Take-Off in a European Industry Perspective

(Bonifacio M. et al., 2005) (Brunschweig B. et providers language(s) see (Guarino N. 1999) (Zyl
al., 2005) (Nordheim D. et al., 2005). J. et al., 200) (Lehtola A. et al., 2003) (Heinecke
Eventually, the semantic backbone makes J. et al., 2003) (Benatallah B et al., 2005).
possible to cross a qualitative gap by providing
cross-lingual data. Knowledge-Based E-Commerce
and E-Business Value
E-Commerce and E-Business
At present, ontology and more generally knowl-
Electronic commerce is mainly based on the edge-based systems appear as a central issue
exchange of information between involved stake- for the development of efficient and profitable
holders using a telecommunication infrastructure. e-commerce and e-business solutions. However,
There are two main scenarios: business-to-custom- because of the actual situation i.e. the partial
er (B2C) and business-to-business (B2B). standardization of business models, processes,
B2C applications enable service providers to and knowledge architectures, it is currently dif-
promote their offers, and for customers to find ficult for companies to achieve the promised
offers which match their demands. By providing ROI from knowledge-based e-commerce and
unified access to a large collection of frequently e-business.
updated offers and customers, an electronic Moreover, a technical barrier exists that is
marketplace can match the demand and supply delaying the emergence of e-commerce, lying
processes within a commercial mediation envi- in the need for applications to meaningfully
ronment. share information, taking into account the lack
B2B applications have a long history of using of reliability, security and eventually trust in
electronic messaging to exchange information the Internet. This fact may be explained by the
related to services previously agreed among two variety of e-commerce and e-business systems
or more businesses. Early plain-text telex com- employed by businesses and the various ways
munication systems were followed by electronic these systems are configured and used. As an
data interchange (EDI) systems based on terse, important remark, such interoperability prob-
highly codified, well structured, messages. A new lems become particularly severe when a large
generation of B2B systems is being developed number of trading partners attempt to agree and
under the ebXML (electronic business in XML) define the standards for interoperation, which is
heading. These will use classification schemes precisely a main condition for maximizing the
to identify the context in which messages have ROI indicator.
been, or should be, exchanged. They will also Although it is useful to strive for the adoption
introduce new techniques for the formal record- of a single common domain-specific standard
ing of business processes, and for the linking of for content and transactions, such a task is often
business processes through the exchange of well- difficult to achieve, particularly in cross-industry
structured business messages. ebXML will also initiatives, where companies co-operate and
develop techniques that will allow businesses to compete with one another. Some examples of
identify new suppliers through the use of regis- the difficulties are:
tries that allow users to identify which services a
supplier can offer. ebXML needs to include well • Commercial practices may vary wide-
managed multilingual ontologies that can be used ly, and consequently, cannot always be
to help users to match needs expressed in their aligned for a variety of technical, practi-
own language with those expressed in the service cal, organizational and political reasons.

892
Semantic Web Take-Off in a European Industry Perspective

• The complexity of a global description knowledge management tools are available, such
of the organizations themselves: their that the asset owner can readily enhance their data
products and services (independently or with semantic meaning, and client-side tools are
in combination), and the interactions be- available to enable machine interpretation of the
tween them remain a formidable task. semantic descriptions related to the products being
• It is not always possible to establish a prio- offered, such that the end user can benefit from
ri rules (technical or procedural) governing the available and mined knowledge. Examples of
participation in an electronic marketplace. some possible application areas were studied by
the Agent Cities project35.
Adoption of a single common standard may In the e-business area Semantic Web tech-
limit business models which could be adopted by nology can improve standard business process
trading partners, and therefore, potentially reduce management tools. One prototypical case is in the
their ability to fully participate in e-commerce. area of logistics. The application of knowledge
A knowledge-based approach has the potential technology on top of today’s business manage-
to significantly accelerate the penetration of elec- ment tools enables the automation of major tasks
tronic commerce within vertical industry sectors, of business process management36 see (Semantic
by enabling interoperability at the business level. Web Case Studies for eBusiness 2005).
This will enable services to adapt to the rapidly In one of the Knowledge Web Industry-
changing business ecosystem. Research co-operations, a number of scenarios
within the B2B integration scenario were iden-
Knowledge-Based E-Commerce tified, involving data mediation, discovery, and
and E-Business Applications composition of services. All of these use cases have
been evaluated according to a community-agreed
The Semantic Web brings opportunities to in- methodology defined by the SWS challenge meth-
dustry to create new services34, extend markets, odology with satisfying success levels defined by
and even develop new businesses since it enables the methodology. This is an important step when
the inherent meaning of the data available in the proving the added value of the Semantic Web
Internet to be accessible to systems and devices service technology applied to B2B integration
able to interpret and reason at the knowledge domain. In addition, the standardization process
level. This in turn leads to new revenue opportu- has been partially finalized within the OASIS
nities, since information becomes more readily Semantic Execution Environment Technical
accessible and usable. For example, a catering Committee (OASIS SEE TC) and W3C Semantic
company whose Web site simply lists the menus Annotations for WSDL and XML Schema (W3C
available is less likely to achieve orders compared SAWSDL WG). However, the standardization
to one whose menus are associated with related process in both groups is still ongoing, but under
metadata about the contents of the dishes, their business pressure has concluded respectively on
origin (e.g., organic, non-genetically modified, SAWSDL in 2007 and SESA framework early
made with local produce), links to alternative 2008.
dishes for special diets, personalised ordering The Industry-Research co-operation has de-
where a user profile can be established which monstrably solved a business case from the B2B
automatically proposes certain menu combina- domain. We have shown how the technology deals
tions depending on the occasion (e.g., wedding with requirements from B2B domain and how this
banquet, business lunch). The latter case assumes technology reacts to changes in back-end systems
that both provider-side knowledge generation and which might occur over the system’s lifetime.

893
Semantic Web Take-Off in a European Industry Perspective

Figure 4. Semantic Web services integration in B2B

The research is not yet ready for industry. It B2B integration when proprietary back-
must be shown how the technology is layered on end systems of one company needed to be
the existing infrastructure and how it interacts integrated with a partner using RosettaNet
with existing systems. For this purpose some parts standard. Whole scenario has been suc-
of the technology need to be standardized (such cessfully addressed.
as grounding mechanisms built on SAWSDL or • Discovery Scenario (http://sws-challenge.
the architecture). In particular, the grounding org/wiki/index.php/Workshop_Athens).
mechanism built on SAWSDL provides a “com- Addressing discovery scenario when a
mon interface” between semantic descriptions and supplier needed to be discovered and se-
non-semantic descriptions (in our case WSDL). lected from suitable ones. Whole scenario
The standardization is still ongoing while at the has been successfully addressed.
same time, the alignment of service semantics • Composition Scenario (http://sws-chal-
with this grounding mechanism must be further lenge.org/wiki/index.php/Workshop_
finalised. While it has been demonstrated how this Innsbruck). Addressing composition sce-
is possible to be done and what the added value nario when more services can satisfy the
of this approach is, the complexity of business user need. Whole scenario has been suc-
standards still needs to be addressed. cessfully addressed.
In addition, a prototype is available37 and has
been provided to NoE-Knowledge Web industry Work will continue and the co-operation plans
partners see Figure 4.The following scenarios to address additional scenarios of the SWS chal-
have been realised as part of the Semantic Web lenge, namely scenarios when services can be
Services Challenge: filtered based on non-functional properties (QoS,
financial, etc.). In addition, a tutorial was given
• Mediation Scenario (http://sws-challenge. on SWS in the context of business process man-
org/wiki/index.php/Workshop_Budva). agement at ICWS’07 conference, and the authors
Addressing the mediation scenario for co-organize the workshop on service composition

894
Semantic Web Take-Off in a European Industry Perspective

and SWS challenge held at the Web Intelligence all their personal devices, in the home, at work,
conference38 (Vitvar T. et al., 2007a) (Vitvar T. et in the car and on the go.
al., 2007b) (Hasselwanter T. et al., 2007). Semantic Web technologies can enhance
multimedia based products to increase the value
Multimedia and Audiovisual Services of multimedia assets such as content items which
are themselves the articles for sale (songs, music
Practical realisation of the Semantic Web vision videos, sports clips, news summaries, etc) or
is actively being researched by a number of ex- where they are used as supporting sales of other
perts, some of them within European collaborative goods (e.g. promotional images, movie trailers
projects, and others within company specific ini- etc). Value is added in search applications, such
tiatives. Earlier projects such as SEKT39 and DIP, that returned items more closely match the user’s
mainly focused on enhancing text based applica- context, interests, tasks, preference history etc,
tions from a knowledge engineering perspective. as well as in proactive push applications such as
Although significant benefits in unlocking access personalised content delivery and recommenda-
to valuable knowledge assets are realised via these tion systems, and even personalised advertising.
projects, in various domains such as digital librar- However, applications such as content personalisa-
ies, enterprise applications, and financial services, tion, where a system matches available content to
it was soon recognised that there was a challenging the user’s stated and learned preferences, thereby
and potentially highly profitable area of research enabling content offerings to be closely targeted
into the integration of multimedia and Semantic to the user’s wishes, rely on the availability of
Web technologies for multimedia content based semantic metadata describing the content in order
applications. Projects such as aceMedia, BOEMIE, to make the match. Currently, metadata genera-
and MESH are examples of initiatives aiming tion is mostly manual, which is costly and time
to advance the use of semantics and reasoning consuming. Multimedia analysis techniques which
for improved multimedia applications such as go beyond the signal level approach to a semantic
automated annotation, content summarisation, analysis have the potential to create automatic
and personalised content syndication. annotation of content, thereby opening up new
The drive for application of semantic tech- applications which can unlock the commercial
nologies in the multimedia and content domains value of content archives (Stamou et al., 2006)
comes from a proliferation of audiovisual devices (Stamou et al., 2005).
and services which have led to an exponential Automated multimedia analysis tools are
growth in available content. Users express dis- important enablers in making a wider range of
satisfaction at not being able to find what they information more accessible to intelligent search
want, and content owners are unable to make engines, real-time personalisation tools, and user-
full use of their assets. Service providers seek friendly content delivery systems. Such automated
means to differentiate their offerings by making multimedia analysis tools, which add the semantic
them more targeted toward the individual needs information to the content, are critical in realising
of their customers. Semantic Web technology can the value of commercial assets e.g. sports, music
address these issues. It has the potential to reduce and film clip services, where manual annotation
complexity, enhance choice, and put the user at of multimedia content would not be economically
the center of the application or service, and with viable, and are also applicable to users’ personal
today’s fast mobile data services and availability content (e.g. acquired from video camera or mobile
of wifi, such benefits can be enjoyed by consumers phone) where the user does not have time, or a suit-
and professional users in all environments using able user interface, to annotate all their content.

895
Semantic Web Take-Off in a European Industry Perspective

Figure 5. Automated semantic annotation in aceMedia

Multimedia ontologies are needed to structure for full exploitation of the potential of multimedia
and make accessible the knowledge inherent in the assets. Interoperability is essential in achieving
multimedia content, and reasoning tools are needed commercial success with semantic multimedia
to assist with identification of relevant content in applications, since it enables multiple manufac-
an automated fashion. Although textual analysis turers, content providers and service providers
and reasoning tools have been well researched, to participate in the market. This in turn enables
and despite the projects funded by the European consumer confidence to be achieved, and a viable
Commission in the 6th framework, fewer tools are ecosystem to be developed.
available for semantic multimedia analysis, since
the problem domain is very challenging. However, Knowledge Enhanced
automated multimedia content analysis tools such Multimedia Services
as those being studied within aceMedia40 are a first
step in making a wider range of information more In aceMedia the main technological objectives
accessible to intelligent search engines, real-time are to discover and exploit knowledge inherent
personalisation tools, and user-friendly content in multimedia content in order to make content
delivery systems. more relevant to the user; to automate annotation
Furthermore, interoperability of multimedia at all levels (see Figure 5) ; and to add functional-
tools is important in enabling a wide variety of ity to ease content creation, transmission, search,
applications and services on multiple platforms for access, consumption and re-use.
diverse domains. The W3C Multimedia Semantics Users access multimedia content using a
Incubator Group reported on interoperability is- variety of devices, such as mobile phones and
sues41 and it is clear that a common framework set-top-boxes, as well as via broadband cable
using Semantic Web languages tools is essential or wireless to their PC. Through exploitation

896
Semantic Web Take-Off in a European Industry Perspective

of Semantic Web tools, aceMedia has created a and (possibly extended) inference services for the
system which provides new and intuitive ways extraction of semantic descriptions from multi-
for users to enjoy multimedia content, such as media content. Additional aspects include among
intelligent search and retrieval, self-organising other scalability, logic programming and DL-based
content, and self-adapting content. For example, reasoning integration for non-standard inference
using aceMedia’s automatic metadata generation, support, and ontology evolution (Stoilos G. et al,
a user can annotate content taken with her mobile 2005) (Petridis K. et al., 2006) (Dasiopoulou S.
phone, then seamlessly upload it to her PC where et al., 2007).
further automatic metadata generation takes
place. aceMedia tools enables the content to be Leveraging Social Network Knowledge
automatically organised into thematic categories, for Movie Recommendations
according to the user’s preferences, and using
extensions to DLNA/UPnP (networked digital Another interesting reported multimedia experi-
home) standards, the content can be automatically ment is MediaCaddy (Garg S. et al., 2005) aiming
pushed to other users (as specified by the content at providing movie or music recommendations
owner) according to chosen rules. For example, based on published online critics, user experience
our user might automatically receive new pictures and social networks. Indeed, for the entertainment
of herself on her mobile phone or PC which were industry, traditional approaches to delivering
acquired and annotated on the device of one of meta-content about movies, music, TV shows,
her friends or family. etc. were through reviews and articles that were
The aceMedia use case highlighted a number done and published in traditional media such as
of future direction, issues and new challenges with newspapers, magazines and TV shows. With the
respect to semantic multimedia content analysis introduction of the Internet, non-traditional forms
and manipulation within a Semantic Web frame- of delivering entertainment started surfacing. The
work. Apart from the requirements with respect third quarter of 2003 in the U.S was the best ever
to formal uncertainty representations and more for broadband penetration bringing such services
effective reasoning and management tools sup- as content on-demand and mobile multimedia. As
port, two dimensions of significant importance of today more than 5000 movies and 2,500,000
include: songs are available on line. In the next couple of
years this figure is expected to grow in leaps and
• Cross-media analysis, where additional re- bounds. With such a phenomenal rise in content
quirements are posed due to the multimo- over IP, a new need for secondary metacontent
dality of knowledge considered, and their related to the movies/music emerged. Initially
semantic modelling and integration, and this was through movie reviews or music reviews
• Non-standard approaches to reasoning, as published on Web portals such as Yahoo, MSN
purely deductive reasoning alone proves and online magazine portals as well as entertain-
not sufficient ment sales sites such as Netflix.com and Amazon.
com.
Other projects which can use the results of this Most consumers today get information about
co-operation: particularly K-Space42, X-Media43, media content primarily from reviews/articles
BOEMIE44 and MESH45 constitute research con- in entertainment/news magazines, their social
sortiums working on the same topic. As, in the case network of friends (one user recommends a
of aceMedia, the main research directions focus song or movie to a friend), acquaintances and
on the exploitation of formal explicit knowledge advertisements. In most of the cases, one or all

897
Semantic Web Take-Off in a European Industry Perspective

Figure 6. Conceptual model of content navigation system from the MediaCaddy project

of the above influence user’s opinion about any and structured as an ontology using a semantic
content (s)he chooses to consume. In addition, a inferencing platform. Figure 6 illustrates the
new breed of customizable meta-content portal conceptual model of MediaCaddy
has emerged, which specifically targets the en- This provides multiple benefits, both allowing
tertainment industry. Examples of such portals for a uniform mechanism for aggregating disparate
include Rotten Tomatoes46 and IMDB47. However, sources of content, and on the other hand, also
these services today are typically accessed via allowing for complex queries to be executed in a
Web portals thereby limiting the interactions and timely and accurate manner. The platform allows
access to the information for a user in a non-PC this information to be accessed via Web Services
environment. APIs, making integration simpler with multiple
MediaCaddy is a recommendation and ag- devices and UI formats. Another feature that sets
gregation service built around a self-learning MediaCaddy apart is its ability to achieve a high
engine, which analyzes a click stream generated level of personalization by analyzing content
by user’s interaction and actions with meta-content consumption behaviour in the user’s personal
displayed through a UI. This meta-content (Music Movie/Music Domain and his or her social network
/Movies/ TV reviews/ article/ synopsis/ production and using this information to generate music and
notes) is accessed from multiple Internet sources movie recommendations.

898
Semantic Web Take-Off in a European Industry Perspective

Prominent Applications Conclusion and


Future Trends
Finally we list some excellent illustrations of
the applications of Semantic Web technology, as In 2000, three prominent authors in the Semantic
selected from a worldwide competition48 which Web activity expounded in a seminal Scientific
offers participants the opportunity to show the American paper (Berners-Lee T. et al., 2001) the
best of the art. Semantic Web vision. In the time since then, the
Semantic Web has become real. Currently, there
• Multimedia e-culture demonstrator, is to are hundreds of millions of RDF triples, on tens of
demonstrate how novel semantic-Web and thousands of Web pages, and thousands of ontol-
presentation technologies can be deployed ogy pages have been published using RDF schema
to provide better indexing and search sup- and OWL, with a growing level of industrial
port within large virtual collections of support. This very active support from industry
cultural heritage resources, 1st Prize 2006, is witnessed at worldwide key conference49 very
http://e-culture.multimedian.nl/demo/ focused on the applications of the Semantic Web
search Technology. Indeed, about 100 talks on industry
• CONFOTO, Essen, Germany. CONFOTO experience in testing and deploying the technol-
is an online service which facilitates brows- ogy and about 30 technology showcases and 10
ing, annotating and re-purposing of photo, workshops or tutorials were actively followed
conference, and people descriptions. 1st by hundreds of attendees (300 at STC 05, 700 at
Prize 2005: http://www.confoto.org/ STC 06, 730 at STC 07 and 210 at the 1st ESTC
• FungalWeb, Concordia University, Canada. 2007) mostly from the industry.
“Ontology, the Semantic Web, Intelligent However, the Semantic Web is still in its early
Systems for Fungal Genomics”.2nd days and there are many exciting innovations on
Prize 2005: http://www.cs.concordia.ca/ the horizon.
FungalWeb/ A keynote speech50 foresaw (Hendler J. & Las-
• Bibster – A semantics-based Bibliographic sila O., 2006) a “re-birth of AI” (or the end of the
P2P system. http://bibster.semanticweb. AI-winter) thanks to big-AI applications (Deep
org Blue, Mars Rover, Deep Space 1, Sachem-Usinor)
• CS AKTive space – Semantic data inte- and Web AI (IR, NLP, Machine Learning, Services,
gration. http://cs.aktivespace.org (Winner Grid, Agents, social nets) needed due to the tre-
2003 Semantic Web challenge) mendous amount of data continuously available on
• Flink: SemWeb for analysis of Social the Web and the emergence of new ways of doing
Networks. http://www.cs.vu.nl/~pmika things (loose coupling of distributed applications
(Winner 2004 Semantic Web challenge) or services, new business models, etc.).
• Museum Finland: Sem. Web for cultural From 2000 to 2007, three major endeavours
portal. http://museosuomi.cs.helsinki.fi have paved the way for the future: DARPA, W3C
(2nd prize 2004 Semantic Web challenge) and EU IST where DARPA and EU IST funded
• Also see Applications and Demos at W3C projects particularly were clearly forces towards
SWG BPD. http://esw.w3.org/mt/esw/ar- production of recommendations to W3C (RDF-S,
chives/cat_applications_and_demos.html OWL, Rules, …), for fast adoption in industry. In
the meantime, 2003 saw early government adop-
tion and emerging corporate interest, in 2005 the
emergence of commercial tools, lots of open source

899
Semantic Web Take-Off in a European Industry Perspective

software and even good progress in the problem • Enterprise Web 2.0 can be the catalyst for
of scalability (tractable reasoning over 10 million a more collaborative business environ-
triples has already been claimed by Oracle51).So, ment55. The BBC World Service had done
significant corporate activity is clearly noticeable a lot of work to try to create a more col-
today compared to 7 years ago: laborative work environment. As it turned
out, the BBC’s internal forums, which
• Semantic (Web) technology companies are only cost the company about 200 pounds,
starting and growing: Cerebra, Siderean, got the company to be more collaborative
SandPiper, SiberLogic, Ontology than the more formal initiatives did.
Works, Intellidimension, Intellisophic,
TopQuadrant, Data Grid, Software AG, Then, it is also witnessed that adding a few
OntoText, Autonomy, FAST, Exalead, semantics to current Web applications - meaning
iSoco, Jouve, Mondeca, Sirma, Pertim, “not harnessing the full picture at once but step by
Biovista, etc. step” – gives a significant push in applications:
• Semantic Web technology appears in richer metadata, data harvesting and visualization,
the strategic plans of large corporations: Web-based social network, digital asset manage-
Nokia, SAP AG, IBM, HP, Adobe, Cisco, ment, scientific portals, tools for developers, and
Oracle, Sun, Vodaphone, Renault, AGFA, so gradually closing the semantic gap.
Cable and Wireless, Novartis, JP Morgan
Chase Bank, Wells Fargo, Swiss Life, What has been Learned from AI?
Boeing, Audi, Elsevier etc.
• Outreach to industry is also demonstrated • Cross-breeding with AI succeeded, stand-
through a newly launched W3C initiative alone AI did not!
(2007): “Semantic Web Education and • Tools are hard to sell (needed too much
Outreach Interest Group - Case Studies skill and education)
and Use Cases”. Case studies include • Reasoners are a means, not an end (a key
descriptions of systems that have been component but not the end)
deployed within an organization, and • Knowledge engineering bottle-
are now being used within a production neck (Ontology development and
environment52. management)
• Government projects in and across agen-
cies: US, EU, Japan, Korea, China, FAO, What has been Learned
etc. from the Web?
• Life sciences/pharma is an increasingly
important market, e.g. the Health Care and • The magic word: Distribute, interconnect
Life Sciences Interest Group at W3C53 and Share Roadmap!
• Many open source tools are available: • PC era [1980-1990] – autonomous com-
Kowari, RDFLib, Jena, Sesame, Protégé, puting and Ethernet
SWOOP, Wilbur etc. see the W3C SWAD • Internet 1st generation [1990-2000] - Web
initiative54 1.0), “read-only Web”, Web sites and
• Related technologies are taking off: Companies’ portals
Google Base (taxonomies for resource de- • Social Networks [2000-2010] - Web 2.0,
scriptions), Web 2.0 initiatives for mash- corporate Knowledge Management and
up applications, etc. social nets

900
Semantic Web Take-Off in a European Industry Perspective

• Semantic Web [2007 – 2020] - Web 3.0 e.g. Rules in communicating applications, Rules
– merging social net with automated to describe / represent service process models,
Semantic Web Rules for policies and contracting, etc. (see e.g.
• Web OS [2020-2030] - Web 4.0 RuleML W3C56)
• Semantic Web Services and sevices
oriented computing (Papazoglou et al.,
However, it must be clear that there are still key 2006):
technology locks identified today that needs aca- ° Discovery: Automated service discovery,
demic research and R&D investments for a full reasoning on non functional parameters like QoS
uptake of the technology (Cuel et al., 2007): and cost.
° Composition: Business and industrial pro-
• Ontology and reasoning: cesses automated. I/O signature and behavioural
° The development of medium size to large composition (Inputs, Outputs, pre-conditions,
ontologies is a challenging task: e.g. modelling effects and message protocols). Robustness in a
of business domains, unified and industry-grade versatile and inconsistent context. Composition
methodology, best practices and guidelines, re-use driven by high level business needs (Business
of existing ontologies and simple tools to use. Rules).
° Automated update of ontologies and ° Management: Web services supervision,
knowledge bases: e.g. ontology maintenance by self-healing service, self-optimizing service, self-
extraction of new knowledge facts and concept configuring service, self-protecting service.
reasoning (abduction, learning), knowledge base ° Design tools: Unified design principles for
population from legacy databases, data warehouse engineering service applications, associating a
and data on the Web, consistency checking. service design methodology with standard soft-
° Ontologies interoperability: Overcome ware development and business process modelling
inevitable heterogeneity in spite of KR standards techniques, service governance, test and proof
via e.g. automated mapping (concepts, properties, checks.
rules, graphs, …) in particular in the open context ° Pilots and standard platforms. The most
of the Web and the social Web (Web 2.0). prominent (2007):
° Scalability: Be capable to process business WSMX57 (Fensel et al., 2005) probably the
and real applications needs e.g. approximate most complete architecture to date, experimented
reasoning, reasoning on inconsistent ontologies, on business cases and in transfer to OASIS
ontology and reasoning modularization, distribu- SAWSDL58: some running prototypes, indus-
tion and cooperation. trial pilots and transfer to W3C (Sivashanmugam,
° KR expressivity and tractability trade-off: 2003) (METEOR-S59)
Maintaining the just needed KR expressivity to OWL-S 60 (Ankolenkar, 2004) (OWL-S
avoid tractability and decidability issues, there MX61)
are many open problems in this area. Look for SWSF62
reasoning algorithm optimizations (!), measure In summary, the Semantic Web is “an interop-
experimental complexity and lastly may be relax erability technology”, “a architecture for inter-
the completeness property. connected communities and vocabularies” and
° Rules - Logic Programming and KR: “a set of interoperable standards for knowledge
Moving towards a deeper and broader automation exchange”63. Firstly, layman users facing the un-
of business process intra- and inter-enterprise manageable growth of data and information, and
require the addition of semantic rules technology. secondly companies facing the huge amounts and

901
Semantic Web Take-Off in a European Industry Perspective

volatility of data, applications and services, all academia and industry: ESWC, ISWC, ASWC,
require urgently automated means that master this ICWS, WWW, STC etc. Lastly, credits go also
growing complexity. In such de-facto context, no directly to the numerous people, in research labs
individual is able to identify knowledge patterns in in academia and in industry who are contributing
their heads, no company (and employees!) is able so strongly to make semantic technology a real
to always shorter its products and service lifecycle momentum in industry.
and self adapt rapidly enough to survive. IST-REWERSE is a NoE supported by the Eu-
The performance of semantic technologies ropean Commission under contract FP6-506779
clearly shows efficiency gain, effectiveness gain http://rewerse.net
and strategic edge. Those facts are based on a IST-Knowledge Web is a NoE supported by
survey64 of about 200 business entities engaged the European Commission under contract FP6-
in semantic technology R&D for development 507482 http://knowledgeweb.semanticweb.org
of products and services to deliver solutions and IST-aceMedia is an Integrated Project support-
also recently witnesses at the ESTC 2007 industry ed by the European Commission under contract
oriented major event. From an academic and tech- FP6-001765. http://www.acemedia.org
nological viewpoint, most things that have been
predicted have happened - the semantic chasm is
closing. Some things happened faster than antici- References
pated like – triple store scaling, cooperation tools,
enhanced SOA middleware, meta-data tagging Ankolenkar A., Paolucci M., Srinivasan N., and
tools, semantically-aware development environ- Sycara K., (2004). The owl services coalition, owl-
ments and last but not the least, the unpredicted s 1.1 beta release. Technical report, July 2004.
huge rise of Web 2.0 user-oriented technology65 Antoniou, G., Baldoni, M., Baroglio, C., Baum-
– and others still need to be realized: ontologies gartner, R., Bry, F., & Eiter, T. (2004). Reasoning
are there (but very little interlinking and the need Methods for Personalization on the Semantic Web.
is huge especially in the healthcare domain), Annals of Mathematics . Computing & Telefinfor-
public information sources and public re-usable matics, 2(1), 1–24.
ontologies (as RDF, OWL, etc.), standard Rules
(SWRL, WSML, etc.) and Logic Programming Barillot, C., Amsaleg, L., Aubry, F., Bazin, J.-P.,
integration to Ontology languages, scalable and Benali, H., & Cointepas, Y. (2003). Neurobase:
robust reasoners, technology transparency for Management of distributed knowledge and data
the final user and the practitioners, and these bases in neuroimaging. [New-York, NY.]. Human
technologies must mature into enterprise-class Brain Mapping, 19, 726–726.
products, etc.
Benatallah, B., Hacid, M. S., Léger, A., Rey, C.,
Pervasive computing is just emerging.
& Toumani, F. (2005). On automating Web ser-
vices discovery. The VLDB Journal, 14(1), 84–96.
doi:10.1007/s00778-003-0117-x
Acknowledgment
Berardi, D., Calvanese, D., De Giacomo, G.,
This work has been possible thanks to the three Lenzerini, M., & Mecella, M. (2005). Automatic
large European consortia REWERSE, Knowledge Service Composition based on Behavioral De-
Web and aceMedia. Acknowledgements are also scriptions. International Journal of Cooperative
for the large gathering of international confer- Information Systems, 14(4), 333–376. doi:10.1142/
ences mixing research results and prospects from S0218843005001201

902
Semantic Web Take-Off in a European Industry Perspective

Berners-Lee, T., Hendler, J., Lassila, O. (2001). Dasiopoulou, S., Saathoof, C., Mylonas, Ph.,
The Semantic Web. Scientific American Journal Avrithis, Y., Kompatsiaris, Y., & Staab, S. (2007).
(May 2001). Introducing Context and Reasoning in Visual Con-
tent Analysis: An Ontology-based Framework. In
Billig A. and Sandkuhl K. (2002). Match-making
Paola Hobson, Ioannis Kompatsiaris (Editors),
based on semantic nets: The xml-based approach
Semantic Multimedia and Ontologies: Theories
of baseweb. 1st workshop on XML-Technologien
and Applications. Springer-Verlag.
fur das Semantic Web, 39–51.
Description Logics for Information Integration.
Bonifacio, M., & Molani, A. (2005). Managing
(2002). Computational Logic: Logic Program-
Knowledge needs at Trenitalia. Proceedings of
ming and Beyond. LNCS; Vol. 2408, pp 41 – 60.
the Second European Semantic Web Conference
London, UK: Springer-Verlag.
ESWC 2005, Industry Forum proceedings.
Dhamankar, R., Lee, Y., Doan, A., Halevy, A.,
Brunschweig, B., & Rainaud, J. F. (2005). Seman-
& Domingos, P. (2004). iMAP: Discovering
tic Web applications for the Petroleum industry.
complex semantic matches between database
Proceedings of the 2nd European Semantic Web
schemas. Proceedings of the 2004 ACM SIGMOD
Conference, ESWC 2005 Industry Forum
international conference on Management of data,
Cardoso, J., & Sheth, A. (2003). Semantic e- pages 383 – 394.
workflow composition. [Technical description
Di Noia, T., Di Sciascio, E., Donini, F. M., &
of the composition of a workflow.]. Journal of
Mongiello, M. (2003). A system for principled
Intelligent Information Systems, 21(3), 191–225.
matchmaking in an electronic marketplace. Pro-
doi:10.1023/A:1025542915514
ceedings of the 12th international conference on
Chan, M. K. S., Judith, B., & Luciano, B. (2006). World Wide Web, WWW 2003, 321–330.
Survey and Comparison of Planning Techniques
Do, H. H., & Rahm, E. (2002). COMA - a system
for Web Services Composition. Technical Re-
for flexible combination of schema matching ap-
port.
proaches. In Proceedings of Very Large Databases
Charlet J., Cordonnier E., Gibaud B. (2002). VLDB 2002, 610–621.
Interopérabilité en médecine: quand le contenu
Dou, D., McDermott, D., & Qi, P. (2005). Ontol-
interroge le contenant et l’organisation. Revue
ogy translation on the Semantic Web. Journal on
Information, interaction, intelligence, 2(2), 37-
Data Semantics, 3360, 35–57.
62.
Ehrig, M., & Staab, S. (2004). QOM: Quick
Cordonnier, E., Croci, S., Laurent, J.-F., &
ontology mapping. Third International Seman-
Gibaud, B. (2003). Interoperability and Medical
tic Web Conference, ISWC 2004, LNCS 3298,
Communication Using “Patient Envelope”-Based
683–697.
Secure Messaging. Medical Informatics Europe
Congress, 95, 230–235. Euzenat, J., & Shvaiko, P. (2007). Ontology
Matching. Springer-Verlag.
Cuel R., Deleteil A., Louis V., Rizzi C. (2008).
Knowledge Web Technology Roadmap: The Euzenat, J., & Valtchev, P. (2004). Similarity-based
Technology Roadmap of the Semantic Web. To ontology alignment in OWL-lite. Proceedings of
appear early 2008. European Conference on Artificial Intelligence
ECAI 2004, 333–337.

903
Semantic Web Take-Off in a European Industry Perspective

Fensel, D., Kifer, M., de Bruijn, J., Domingue, J. Heinecke, J., & Cozannet, A. (2003). Ontology-
(2005). Web service modeling ontology (wsmo) Driven Information Retrieval. a proposal for
submission. w3c member submission. multilingual user requests. Workshop on Onto-
logical Knowledge and Linguistic Coding at the
Garg, S., Goswami, A., Huylebroeck, J., Jag-
25th annual meeting of the German Linguistics,
anathan, S., & Mullan, P. (2005). MediaCaddy
Feb. 26-28, 2003.
- Semantic Web Based On-Demand Content Navi-
gation System for Entertainment. Proceedings of Hibbard, J. (1997). Knowledge management-
the 4th International Semantic Web Conference, knowing what we know. Information Week
ISWC 2005. LNCS 3729, 858 – 871. (October 20).
Giunchiglia, F., & Shvaiko, P. (2003). Se- Hull R., (1997), Managing Semantic Heteroge-
mantic matching. Knowledge Engineering neity in Databases: A Theoretical Perspective,
Review Journal, 18(3), 265–280. doi:10.1017/ Tutorial at PODS 1997.
S0269888904000074
Hull, R., & Su, J. (2005). Tools for composite Web
Giunchiglia, F., Shvaiko, P., & Yatskevich, M. services: A short Overview. SIGMOD Record, 34,
(2004). S-Match: an algorithm and an imple- 2. doi:10.1145/1083784.1083807
mentation of semantic matching. In Proceedings
Lehtola, A., Heinecke, J., & Bounsaythip, C.
of the First European Semantic Web Symposium,
(2003). Intelligent Human Language Query Pro-
ESWS2004, LNCS 3053, 61–75.
cessing in mkbeem. Workshop on Ontologies and
Guarino, N., Masolo, C., & Vetere, G. (1999, Multilinguality in User Interface, in the Proceed-
May). OntoSeek: Content-Based Access to the ings of Human Computer Interface International,
Web. IEEE Intelligent Systems, 14(3), 70–80. HCII 2003, 4, 750-754.
doi:10.1109/5254.769887
McGuinness, D. L., Fikes, R., Rice, J., & Wilder,
Haarslev, V., Moller, R., & Wessel, M. (1999- S. (2000). An environment for merging and test-
2007). RACER: Semantic middleware for in- ing large ontologies. Proceedings of the Seventh
dustrial projects based on RDF/OWL, a W3C International Conference on Principles of Knowl-
Standard. edge Representation and Reasoning (KR2000),
Breckenridge, Colorado, 483–493.
Halevy, A. (2001). Answering Queries Using
Views: A Survey. The VLDB Journal, 10(Issue Mcllraith, S. A., Son, T. C., & Zeng, H. (2001).
4). doi:10.1007/s007780100054 Semantic Web services. IEEE Intelligent Systems
. Special Issue on the Semantic Web, 16, 46–53.
Halevy, A., Rajaraman, A., & Ordille, J. (2006).
Data Integration: The Teenage Years. 10-year best Mena, E., Kashyap, V., Sheth, A., & Illarramendi,
paper award. VLDB. A. (1996). Observer: An approach for query pro-
cessing in global information systems based on
Hasselwanter, T., Kotinurmi, P., Moran, M., Vitvar,
interoperability between pre-existing ontologies.
T., & Zaremba, M. (2006). WSMX: a Semantic
Proceedings of the First International Conference
Service Oriented Middleware for B2B Integration.
on Cooperative Information Systems CoopIS’96,
In Proceedings of the 4th International Conference
14–25.
on Service Oriented Computing, Springer-Verlag
LNCS series, Chicago, USA.

904
Semantic Web Take-Off in a European Industry Perspective

Nixon, L., Mochol, M., Léger, A., Paulus, F., Rahm, E., & Bernstein, P. (2001, Dec.). A survey
Rocuet, L., Bonifacio, M., et al. (2004). Proto- of approaches to automatic schema matching.
typical Business Use Cases. (Technical report Very Large Databases Journal, 10(4), 334–350.
Deliverable 1.1.2), Knowledge Web IST-NoE. doi:10.1007/s007780100057
Norheim, D., & Fjellheim, R. (2006). Knowledge Roman, D., Keller, U., Lausen, H., de Bruijn, J.,
management in the petroleum industry. Proceed- Lara, R., & Stollberg, M. (2005). Web Service
ings of the 3rd European Semantic Web Conference, Modeling Ontology . Applied Ontology Journal,
ESWC 2006 Industry Forum. 1(1), 77–106.
Papazoglou M., Traverso P., Dustdar S. and Rosenbloom ST, Miller RA, Johnson KB. (2006).
Leymann F. (2006). Service-oriented computing Interface terminologies: facilitating direct entry of
research roadmap. Technical report. clinical data into electronic health record systems.
Journal of the American Medical Informatics
Petrash, G. (1996). Managing knowledge assets
Association.
for value. Proceedings of the Knowledge-Based
Leadership Conference, Linkage Inc., Boston, Semantic Web Case Studies and Best Practices
MA, October 1996. for eBusiness (SWCASE). At ISWC2005, online
version
Petridis, K., Bloehdorn, S., Saathoff, C., Simou,
N., Dasiopoulou, S., Tzouvaras, V., et al. (2006). Shvaiko, P., & Euzenat, J. (2005). A survey of
Knowledge Representation and Semantic Annota- schema-based matching approaches. [JoDS].
tion of Multimedia Content. IEEE Proceedings on Journal on Data Semantics, 4, 146–171.
Vision Image and Signal Processing, Special issue doi:10.1007/11603412_5
on Knowledge-Based Digital Media Processing,
Shvaiko, P., Giunchiglia, F., Pinheiro da Silva,
Vol. 153, No. 3, pp. 255-262, June 2006.
P., & McGuinness, D. L. (2005). Web explana-
Petrini, J., & Risch, T. (2004). Processing queries tions for semantic heterogeneity discovery. In
over RDF views of wrapped relational databases. Proceedings of the Second European Semantic
In Proceedings of the 1st International workshop Web Conference, ESWC 2005, 303-317.
on Wrapper Techniques for Legacy Systems,
Shvaiko, P., Léger, A., Paulus, F., Rocuet, L.,
WRAP 2004, Delft, Holland, 2004.
Nixon, L., Mochol, M., et al. (2004). Knowledge
Pistore, M., Roberti, P., & Traverso, P. (2005). Processing Requirements Analysis. Technical
Process-Level Composition of Executable Web report Deliverable D 1.1.3, Knowledge Web
Services: “On-thefly” Versus “Once-for-all” Com- IST-NoE.
position. The Semantic Web: Research and Ap-
Spackman, K. A., Dionne, R., Mays, E., & Weis, J.
plications. Proceedings of ESWC 2005, Heraklion,
Role grouping as an Extension to the Description
Crete, Greece. LNCS 3532, Springer Verlag.
Logic of Ontylog Motivated by Concept Modeling
Preguica, N., Shapiro, M., & Matheson, C. 2003). in SNOMED. Proceedings of the AMIA Annual
Semantics-based reconciliation for collaborative Symposium 2002, San Antonio, Texas, p. 712-6,
and mobile environments. In Proccedings of the November, 9-13, 2002.
Eleventh International Conference on Coopera-
Stamou, G., & Kollias, S. (2005), Multimedia
tive Information Systems, CoopIS 2003, LNCS
Content and the Semantic Web: Standards, Meth-
2888, 38-55.
ods and Tools. Wiley.

905
Semantic Web Take-Off in a European Industry Perspective

Stamou, G., van Ossenbruggen, J., Pan, J., & Zhong, J., Zhu, H., Li, J., & Yu, Y. (2002). Concep-
Schreiber, G. (2006). Multimedia annotations tual graph matching for semantic search. In Pro-
on the Semantic Web. IEEE MultiMedia, 13(1), ceedings of the 10th International Conference on
86–90. doi:10.1109/MMUL.2006.15 Computational Science, 2393 (2002), 92-106
Stoilos, G., Stamou, G., Tzouvaras, V., Pan, J. Zyl, J., & Corbett, D. (2000). A framework for
Z., & Horrocks, I. (2005). A Fuzzy Description Comparing the use of a Linguistic Ontology in an
Logic for Multimedia Knowledge Representation. Application. Workshop Applications of Ontologies
In Proc. of the International Workshop on Mul- and Problem-solving Methods, ECAI’2000, Berlin
timedia and the Semantic Web, pp 12-19, ESWC Germany, August, 2000.
2005, Heraklion, Grece, June 2005.
Traverso, P., & Pistore, M. (2004). Automated
composition of Semantic Web services into ex- Additional Readings
ecutable processes. In Proceedings of the Third
International Semantic Web Conference, ISWC W3C http://www.w3.org/2001/sw/
2004, 380–394, 2004. Annual Semantic Web applications challenge:
Velegrakis, Y., Miller, R. J., & Mylopoulos, J. http://challenge.semanticweb.org
(2005). Representing and querying data trans- Journal of Web semantics (Elsevier)
formations. Proceedings of the 21st International
Conference on Data Engineering ICDE 2005, The Semantic Web: research and Applications
81-92. LNCS series: LNCS 2342 (ISWC 2002), LNCS
2870 (ISWC 2003), LNCS 3053 (ESWS 2004),
Vitvar, T. Zaremba M., Moran M., Haller A., LNCS 3298 (ISWC 2004), LNCS 3532 (ESWS
Kotinurmi P. (2007b). Semantic SOA to Promote 2005), LNCS 4011 (ESWC 2006), LNCS4273
Integration of Heterogeneous B2B Services. The (ISWC 2006), LNCS 4519 (ESWC 2007), LNCS
4th IEEE Conference on Enterprise Computing, 4825 (ISWC 2007).
E-Commerce and E-Services (EEE07), IEEE
Computer Society, July, 2007, Tokyo, Japan.
Vitvar, T., Zaremba, M., & Moran, M. (2007a). Endnotes
Dynamic Service Discovery through Meta-Inter-
actions with Service Providers. In Proceedings 1
http://www.web2con.com
of the 4th European Semantic Web Conference 2
http://www.ontoweb.org
(ESWC2007), Springer-Verlag LNCS series, 3
http://Knowledge Web.semanticweb.org
Innsbruck, Austria. 4
http://rewerse.net
Wiederhold, G. (1992). Mediators in the archi-
5
http://cordis.europa.eu/fp7/home_en.html
tecture of future information systems. IEEE
6
Knowledge Web Deliverable D 1.1.4v3
Computer, 25(3), 38–49. http://knowledgeweb.semanticweb.org/
semanticportal/deliverables/D1.1.4v3.pdf
Wiig, K. (1997). Knowledge management: where 7
http://knowledgeweb.semanticweb.org/
did it come from and where will it go? Journal o2i/
of Expert Systems with Applications, 13(1), 1–14. 8
http://protege.stanford.edu/index.html
doi:10.1016/S0957-4174(97)00018-3 9
http://oaei.ontologymatching.org/2007/
10
http://www.neon-project.org

906
Semantic Web Take-Off in a European Industry Perspective

11
http://www.ontologymatching.org/ 32
See for example in the Petroleum industry
12
http://www.acemedia.org (Nordheim D. et al., 2005)
13
http://www.nlm.nih.gov/research/umls/ 33
See for example Use of Ontology for pro-
umlsmain.html duction of access systems on Legislation,
14
http://us.expasy.org/sprot/ Jurisprudence and Comments (Delahousse
15
http://obo.sourceforge.net/main.html J. et al., 2006) http://www.eswc2006.org/
16
http://www.ncbi.nlm.nih.gov/Genbank/ industry.html
index.html 34
E.g. see the EU Integrated project “DIP
17
http://www.ncbi.nlm.nih.gov/sites/entrez Data, Information, and Process Integration
18
http://www.connectingforhealth.nhs.uk/ with Semantic Web Services”, http://dip.
19
http://www.edisante.org/ semanticweb.org/
20
http://www.etiam.com/ 35
agentcities RTD project http://www.agentci-
21
http://cen.iso.org/ and http://www.tc251w- ties.org/EURTD/
giv.nhs.uk/ 36
Semantic Business Automation, SAP, Ger-
22
http://www.hl7.org/ many http://www.eswc2006.org Industry
23
http://www.d-m-p.org/docs/EnglishVersion- Forum
DMP.pdf 37
http://sws-challenge.org/2006/submission/
24
http://www.hl7.org/Special/Committees/ deri-submission-discovery-phase3/ http://
structure/index.cfm#Mission sws-challenge.org/2006/submission/deri-
25
http://www.cocoon-health.com submisson-mediation v.1/ http://sws-chal-
26
http://www.srdc.metu.edu.tr/webpage/proj- lenge.org/2006/submission/deri-submisson-
ects/artemis/index.html mediation v.2/
27
http://www.simdat.org 38
http://events.deri.at/sercomp2007/
28
Some of the well-known definitions of 39
Semantically Enabled Knowledge Technolo-
KM include: (Wiig 1997) “ Knowledge gies http://www.sekt-project.com/
management is the systematic, explicit, and 40
http://www.acemedia.org
deliberate building, renewal and application 41
http://www.w3.org/2005/Incubator/
of knowledge to maximize an enterprise’s mmsem/
knowledge related effectiveness and returns 42
http://www.kpace-noe.net
from its knowledge assets”; (Hibbard 1997) 43
http://www.x-media-project.org
“Knowledge management is the process of 44
http://www.boemie.org
capturing a company’s collective expertise 45
http://www.mesh-ip.eu
wherever it resides in databases, on paper, 46
http://www.rottentomatoes.com
or in people’s heads and distributing it to 47
http://www.imdb.com/
wherever it can help produce the biggest 48
Annual Semantic Web applications chal-
payoff”; (Pettrash 1996) “KM is getting the lenge: http://challenge.semanticweb.org
right knowledge to the right people at the 49
Semantic Technology Conference http://
right time so they can make the best deci- www.semantic-conference.com/ ; European
sion”. Semantic Technology Conference http://
29
http://www.cenorm.be/cenorm/index.htm www.estc2007.com/
30
http://www.knowledgeboard.com 50
SemWeb@5: Current status and Future
31
Semantic Web, Use Cases and Challenges at Promise of the Semantic Web, James Hen-
EADS, http://www.eswc2006.org Industry dler, Ora Lassila, STC 2006, 7 March 2006,
Forum. San José, USA

907
Semantic Web Take-Off in a European Industry Perspective

51
Oracle Database 10g using RDF natively In January 2007 a research programme
supported by the 10g Enterprise Edition conducted by the Economist Intelligence
52
http://www.w3.org/2001/sw/sweo/public/ Unit and sponsored by FAST gauged the
UseCases/ relevance of Web 2.0 to large corporations
53
http://www.w3.org/2001/sw/hcls/ throughout the world and across a wide
54
Semantic Web Advanced Development range of industries. The research, which
for Europe http://www.w3.org/2001/sw/ consisted of an online survey plus follow-up
Europe/ interviews with senior executives at large
55
Forester, Erica Driver, October 2007 public corporations, found that Web 2.0 now
56
http://www.w3.org/2005/rules/wg/wiki/ has significant implications for big business
RuleML across a wide range industry sectors. By
57
http://www.oasis-open.org/committees/ 2006, and even earlier at some companies,
semantic-ex/faq.php, http://www.wsmx.org the world’s multinationals began to see
http://sourceforge.net/projects/wsmx many Web 2.0 technologies as corporate
58
http://www.w3.org/TR/sawsdl/ tools. In fact, according to the survey, 31%
59
http://lsdis.cs.uga.edu/projects/meteor-s/ of companies think that use of the Web as a
60
http://www.w3.org/Submission/OWL-S/ platform for sharing and collaboration will
61
http://www-ags.dfki.uni-sb.de/~klusch/ affect all parts of their business (Economist
owls-mx/index.html Intelligence Unit (2007): Serious business.
62
http://www.w3.org/Submission/SWSF/ Web 2.0. goes corporate. A report from the
63
ESTC 2007 Keynote speech from Susie EIU sponsored by FAST.Also to mention two
Stephens (Oracle) majors initiatives: MySpace with 300 million
64
Semantic Wave 2006, Part-1 Mills Davis users (Dec 2007) http://www.myspace.com
65
Web 2.0 and its related phenomena becomes and Facebook with 60 millions users (Nov
increasingly interesting for businesses. 2007) http://www.facebook.com

This work was previously published in Semantic Web for Business: Cases and Applications, edited by R. García, pp. 1-29,
copyright 2009 by Information Science Reference (an imprint of IGI Global).

908
909

Chapter 4.3
A Strategic Framework
for Integrating Web 2.0
into the Marketing Mix
Samantha C. Bryant
Philip Morris, USA

Abstract Introduction

Marketing strategy set by the marketing mix has Marketing has gone through a number of evolu-
remained fundamentally the same through years tions and technology has revolutionized a number
of other business disciplines being significantly of disciplines. New generations of consumers are
disrupted by emerging technologies. Emerging Web consuming media in a different fashion than before.
2.0 technologies such as wikis, blogs, YouTube, and Gone are the days of the 30-second Super Bowl
virtual worlds are not only affecting how compa- advertisements and here are the days of Facebook,
nies tactically approach marketing, but also their Flickr, and MySpace. Collaborating and participat-
marketing strategies. This chapter will explore the ing on the Internet is the preferred entertainment.
impact of Web 2.0 technologies on marketing and Web 2.0 technologies such as wikis, blogs, You-
brand management and how companies can lever- Tube, and Second Life are changing the behavior
age these technologies to strengthen relationships of consumers like never before.
between their brands and consumers through a Web
2.0 marketing mix. This new Web 2.0 marketing • Empowering them with knowledge from a
mix supplements the traditional four-p marketing myriad of sources
mix (price, product, promotion, and placement) • Enabling them to self-organize around
with a new “p” lens: participation. The focus of brands and share their passion (or dissatis-
this analysis is on B2C marketing of products and faction) for a brand
services only. • Enabling them to act as marketers of brands

Marketing strategy set by the marketing mix has


remained fundamentally the same through years
of other business disciplines being significantly
DOI: 10.4018/978-1-60566-122-3.ch003 disrupted by emerging technologies. However, Web

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

2.0 is not only affecting how companies tactically marketing theory. In 2008, the AMA hosted its first
approach marketing, but also their marketing conferences on Web 2.0 (social media) and market-
strategies. ing. The conferences focused on how companies
This chapter will explore the impact of Web need to recognize the impact social media has on
2.0 technologies on marketing and brand man- their brands and how they can benefit from using
agement and how companies can leverage these social media as new marketing tools (American
technologies to strengthen relationships between Marketing Association, 2007a).
their brands and consumers through a Web 2.0 Not surprisingly, the most written about this
marketing mix. The focus of this analysis is on relationship between Web 2.0 technologies and
business-to-consumer (B2C) marketing of prod- marketing has been via Web 2.0 technologies
ucts and services only. themselves, particularly blogs. Blogs, also known
as weblogs, are shared online journals or diaries
where people can post entries via the Web. Live-
Background: Web 2.0 Journal, Blogger, and WordPress are examples of
Technologies and Marketing online blog services where users can post their
thoughts with an emphasis on user interaction
Web 2.0 technologies, also known as social within the community. LiveJournal is one of
software technologies, are a second generation the most popular, with currently over 14 million
of Web-based communities and services that journals and communities. Numerous individuals
facilitate collaboration and sharing between us- blog daily about what they are witnessing and
ers. Web 2.0 does not refer to an update to any hearing about in the marketing world. These same
technical specifications of the Web but to changes individuals then harness the power of Web 2.0 by
in the way it is being used. These technologies linking to each other’s blogs and commenting on
are built on an architecture of participation (“Web the author’s thoughts. What results is insightful
2.0,” 2007). speculation about trends emerging as a result of
Little has been published about Web 2.0 Web 2.0’s impact on marketing and brand man-
technologies, particularly about their effect on agement.
marketing. In 2004, High Intensity Marketing
explored the effects of a new emerging stream of
networked technologies on marketing. However, The Traditional Marketing
these technologies were the mere beginning of Mix (Four Ps)
what was to explode a few years later as Web
2.0. (Mootee, 2004). Wikinomics: How Mass Col- With such revolutionary technologies disrupt-
laboration Changes Everything discussed Web 2.0 ing consumers’ lives, many businesses wonder
technologies, with examples of companies’ use to what the implication is to the marketing of their
interact with their consumers and improve their products and brands.
product offerings (Tapscott & Williams, 2006). The marketing mix, invented in the 1950s, is
The American Marketing Association (AMA) the mix of controllable marketing variables that a
is one of the largest professional organizations company uses to pursue the desired level of sales
for marketers and is trusted to provide relevant in the target market. The most common model
marketing information to experienced marketers. of these factors is the four-factor classification
One channel through which it educates its mem- called the four Ps. Optimization of the marketing
bers is conferences. The AMA has just begun to mix is achieved by assigning the amount of the
acknowledge how Web 2.0 is disrupting traditional marketing budget to be spent on each element

910
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

of the marketing mix so as to maximize the total Issues and Examples:


contribution to the firm (American Marketing The Effect of Web 2.0
Association, 2007b). Technologies on Marketing
The four Ps of the traditional marketing mix and Brand Management
are product, pricing, promotion, and placement.
Numerous companies are integrating Web 2.0
• Product: The product aspects of market- technologies into their portfolio of marketing chan-
ing deal with the specifications of the ac- nels. Let us explore some Web 2.0 technologies
tual goods or services, and how it relates to and how they are affecting marketing strategy.
the user’s needs and wants. Generally, this
also includes supporting elements such as Second Life (Virtual World)
warranties, guarantees, and support.
• Pricing: Pricing refers to the process of A virtual world is a computer-based simulated
setting a price for a product, including dis- environment intended for its users to inhabit
counts. The price need not be monetary; it and interact, usually represented in the form of
can simply be what is exchanged for the graphical representations of avatars. Second Life
product or services, such as time, energy, is an Internet-based virtual world developed by
psychology, or attention. Linden Research, Inc. that is a massive multi-
• Promotion: Promotion includes advertis- player online game. Second Life enables its us-
ing, sales promotion, public relations, and ers to interact with each other through motional
personal selling, and refers to the various avatars, providing an advanced level of a social
methods of promoting the product, brand, network service. Residents can explore, meet other
or company. residents, socialize, participate in individual and
• Placement: Placement or distribution group activities, and create and trade property and
refers to how the product gets to the cus- services with one another.
tomer, for example, point-of-sale place- Second Life has received much media atten-
ment or retailing. This p is also the place, tion regarding its benefits to the business world,
referring to the channel by which a product most significantly from a cover story in Business
or service is sold (e.g., online vs. retail), Week in April 2006 that brought the virtual world
to which geographic region or industry, to the attention of the masses, including a number
and to which segment (young adults, fami- of business leaders. The unique avatar population
lies, professionals; American Marketing in the virtual world Second Life topped 7 million
Association, 2007b). in 2007, with about 4 million distinct individuals
participating in the online world (Rose, 2007).
While effective for setting marketing strategy Second Life has been recognized to have so
since its creation, the traditional marketing mix much marketing potential that new companies
lacks relevancy when it comes to Web 2.0. A new have been established to assist companies with
term coined Marketing 2.0 is being used to describe establishing their presence. For example, Millions
the impact Web 2.0 has had on the discipline of of Us, an agency specializing in virtual worlds,
marketing. Companies are finding many uses of designs and measures marketing programs for
Web 2.0 technologies to successfully connect clients across a wide spectrum of platforms,
with their consumers. These are discussed in the especially Second Life (Millions of Us, 2007).
next section. Another company, The Electric Sheep Company,
is the largest company in the world dedicated to

911
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

designing experiences and add-on software for are able to develop their own homes and
3-D virtual worlds and has implemented major businesses around the Scion dealership
projects in both Second Life and There (another • A simultaneous real and virtual launch
virtual world) technologies (Carter, 2007). of Scion’s 2008 line at the Chicago Auto
In addition to Millions of Us and The Electric Show, meaning the new vehicle was
Sheep, REPERES has formed as the first market launched in both Second Life and the real
research institute in Second Life. REPERES per- world at the same time
forms quantitative research by surveying avatars • Free expert-led customization classes in
and qualitative research via private interviews Second Life for consumers to learn how to
with avatars. REPERES assists companies in personalize their virtual Scions; consumers
the development of their products and offers in can then showcase their designs in a Scion
Second Life using the following: gallery in Second Life

• A panel of avatars that are representative The Scion xB launch received substantial
of the overall population of Second Life in media attention, and the buzz only grew with
terms of nationality, gender, and age; this each subsequent release and event. Scion City has
panel is called upon to address issues faced organically developed its own culture and loyal
by brands seeking to establish themselves base of residents; one even created a MySpace
or develop an offer on Second Life page chronicling the evolution of this community,
• An understanding of behaviors, innova- which shows the convergence and power of Web
tions, and trends in Second Life 2.0 technologies working together (Millions of
• A space for tests and cocreation projects to Us, 2007).
be tried and evaluated (REPERES Second While Scion was the first major automobile
Life, 2007) manufacturer with a significant presence in Second
Life, numerous car companies have followed suit,
In mid-2006, Scion sensed an opportunity including Pontiac, Mercedes, BMW, Nissan, and
to engage the technology and design-oriented Toyota. Mercedes-Benz operates a car dealership
communities of Second Life. Millions of Us and that sells virtual cars and gives away branded rac-
Scion collaborated to create Scion City, a Second ing suits to avatars. BMW even allows avatars the
Life island that housed the first virtual-world car opportunity to test drive their vehicles.
dealership, representing the first time a major At the 2007 Food Marketing Institute Super-
auto manufacturer created a presence in a virtual market Convention and Educational Exposition,
world. Second Life residents had the opportunity Kraft Foods, Inc. unveiled more than 70 new
to not only purchase all three Scion models in products. Kraft chose to showcase these new of-
the dealership, but also customize them to make ferings at a virtual supermarket in Second Life.
them their own. This is consistent with Scion’s Online, consumers and convention attendees could
approach to allow buyers to customize numerous interact with Kraft’s latest products and take part
aspects on real-world cars as well. This launch in online forums with Kraft experts. By having
was such a success that Scion and Millions of Us new products introduced online, consumers could
have continued to work together to extend Scion’s see, “touch,” and learn about the product before
presence in Second Life through the following: it was able to hit physical supermarket shelves
(Kraft Foods, 2007).
• The expansion of Scion City into a full- In addition to Scion and Kraft, numerous other
fledged urban environment where residents brands have built a presence in Second Life, includ-

912
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

ing Adidas, Dell, Reebok, Sony BMG, Vodaphone, Consumers can watch full episodes directly on the
Sun, Sears, AOL, and Circuit City. widget, get short mini clips of some of the stars,
Reebok opened a virtual store that is an exten- send the widget to a friend, link out to download
sion of its real-world RBK custom campaign. The ring tones, and more. The widget is designed to
store sells plain white sneakers by size and then stand out on any site it is placed on (“CBS Mo-
features coloring machines for avatars to custom- bile Launches Widget to Promote New Animated
ize their shoes. Reebok grounds also contain a Series,” 2007).
basketball court for avatars to play on. Sony promoted the film Zathura via widgets
Adidas owns an island that features branded vs. a traditional online option like banner ads.
video clips and billboards. Avatars who purchase Widgets enabled Sony to provide an application
Adidas shoes at the store may then bounce on a related to the movie and more importantly allowed
trampoline next to the store, demonstrating Adidas’ users to interact with it. The widget was offered
shoes’ bounciness. via Freewebs, a Web site that enables consumers
Wired magazine reports that at least 50 major to easily create their own Web sites at no cost.
companies have a Second Life presence (Rose, The 11 million Web site owners of Freewebs
2007). YouTube is a video-sharing Web site where were able to embed the widget within their Web
users can upload, view, and share video clips. So sites, and 11,000 Web sites took this up within 6
many brands and products have moved in that a weeks. The widgets were viewed 600,000 times,
YouTube video of brands in Second Life has been and long after the original movie was out of the-
developed (Hayes, 2007). aters, thousands of widgets were still delivering
content (Jaokar, 2006).
Widgets To enable PC users to experience a game of-
fered on a different gaming system (Xbox), inter-
A widget is a small application that can be embed- active agency AKQA created a weather widget to
ded on different Web pages. Widgets represent promote the Microsoft Flight Simulator X game
a new indirect marketing channel that enters for the Xbox. The widget allows users to virtually
consumers’ lives via other Web 2.0 technologies fly and find out the weather at any airport through
and effectively promotes a brand, product, or a live feed from the National Weather Service. In
service and generates awareness. A widget can the first 2 months, users downloaded the widget
be placed by users onto their personalized home more than 150,000 times, spending an average of
pages, blogs, or other social networking pages. The 23 minutes with the flight simulator, the agency
widget is not intrusive advertising as a consumer says (Steel, 2006).
must actively choose to add the widget. Numerous other companies are using widgets
The content on widgets can include blogs, to market their products and services. Reebok
live discussions, bookmarks to other Web sites, created a widget that allows users to display cus-
webcasts, video, games, and more. Companies tomized pairs of RBK shoes for others to critique.
are already capitalizing on widgets to market their Radio stations are offering widgets that stream a
brands, products, and services. On its traditional station’s broadcast live. Airlines including Ameri-
Web site, CBS lets consumers select widgets to can Airlines and Air France offer a ticket purchas-
embed in their social networking profiles or blogs ing widget that allows a consumer to purchase a
to allow them (and everyone else who views ticket through the widget without having to visit
their pages) to see advertising about the shows. the airlines’ Web sites. (Guiragossion, 2007).
For example, CBS launched a constantly updat- Having a user add a widget to a personalized
ing rich-media widget to promote a new series. site that is visited by individuals who share similar

913
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

interests is an effective way to share your market- ate interest about the Eurostar brand and promote
ing message with those who would find it most a significant event in the brand’s history. It also
relevant. By adding the widget to his or her page, recognized Eurostar consumers who had taken
the user is showing endorsement for the brand, photos of Eurostar and shared them with others
making others trust the brand. on Flickr (Terret, 2007).

Flickr (Tagging) Social Networks

Tagging is the assignment of descriptive contex- Social networks are another Web 2.0 technology
tual tags to data, such as Web links stored with disrupting traditional marketing strategy. Social
memorable words for easy future access. An networks are Web sites designed to allow multiple
example is Flickr, an online photo-management users to publish content. Users are able to con-
and -sharing application that allows users to col- nect with those sharing similar interests and to
laboratively organize their photos by tagging them exchange public or private messages. Facebook
with descriptors that are searchable. and MySpace are popular examples. Facebook
On Flickr, users can post pictures and tag them is a social utility that connects users with other
with words that describe the photo so others can people by enabling them to publish profiles with
search and view the photos. One quick search on text, photos, and videos; to review friends’ profiles;
Flickr reveals 88,923 pictures tagged with Coke, and to join a network. MySpace allows users to
4,650 with Crayola, 3,472 for Vitamin Water, and create a community and share photos, journals,
32,367 for Heineken. One user posted a picture and interests with a growing network of mutual
entitled “Just been to the Heineken Experience,” friends. Some marketers are making their own
described as a photo taken of the photographer’s social networking sites such as Coke’s mycok-
friend after an exciting trip to the Heineken Ex- erewards.com and USA Network’s Characters
perience tourist attraction housed in the former Welcome Web site.
Heineken brewery on the Stadhouderskade in For those brands that have not created their own
Amsterdam. The administrator for a group entitled social networking site, existing social networks
Got Heineken? contacted the user and invited them enable consumers to create brands for themselves
to join the Heine group. The Heine group invites and assist others with creating their brands. At
members to post pictures of themselves holding MySpace, brands have member profiles and
a nice ice-cold Heineken or simply a photograph make friends with other MySpace members. At
of anything related to Heineken. There is no of- Facebook, members join Facebook brand groups,
ficial indication that Heineken is sponsoring it. just like they join Facebook fraternity or hobby
Currently, the group has 363 photos posted by groups, and can even display brand logos on their
212 members. personal profiles.
Eurostar, the United Kingdom’s high-speed Social branding may prove to be the ultimate
train, recognized that moving from Waterloo to product placement strategy. Companies can create
St. Pancras International Station was a significant a page for their brand or product. For example,
event in the brand’s life. Eurostar set up an account Big Sky Brands, maker of Jones Soda, created its
with Flickr, EurostarForTomorrow, to promote the own page. In addition to company information
move to a new station. Eurostar posted a number of and a list of retailers carrying the Big Sky Brand’s
photos of the new station, and collected a number products, its MySpace page contains a blog with
of photos other users have taken and saved them discussions about the new Jones Activated Energy
as favorites. This was an effective way to gener- Boosters, the latest extension to the brand family.

914
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

The blog also talks about promotions, such as the Idris Mootee (2004) in his book High Intensity
Hot Topic and Jones Soda Carbonated Candy Marketing examines the role of strategic market-
Summer Promo. Currently, Big Sky Brands has ing in the network economy, which is relationship
495 friends who, in the comments section of the driven, network centered, technology enabled,
page, engage in discussions about how much and information intensive. Mootee developed an
they love the brand and which products are their analytical model supplement to the traditional
favorites. The site also links to Big Sky Brands marketing ps called “The New 4P’s.” Mootee’s
Music MySpace page, another site dedicated new four ps are participation, peer-to-peer commu-
to empowering their consumers with Big Sky nities, predictive modeling, and personalization.
Brands’ enthusiasm through the medium of music Participation focuses on allowing consumers to
(Big Sky Brands, 2007).  choose their products, providing valuable insights
In addition to this official Big Sky Brands Web to companies about consumers’ needs and wants.
site, a number of unofficial MySpace pages have Mootee cites Dell, Procter & Gamble, and Levi’s
been developed to support Jones Soda. The major- as examples. Mootee also briefly touches on
ity of these page owners have become MySpace consumers playing a role in defining and owning
friends of Big Sky Brands. In contrast, Swiffer brands, such as Burton snowboards.
WetJet also has a number of MySpace pages, but Participation is more relevant than ever with
none are sponsored by Procter & Gamble. the Web 2.0 revolution, but it is more than a
Virtual worlds, widgets, tagging, and social supplemental p to the marketing mix. Participation
networks are Web 2.0 technologies offering op- should now encompass all of the four ps as a lens
portunities for companies to integrate them into through which the ps should be approached. The
their marketing mix. The next section provides previous section explored a number of examples of
a simple framework to help leverage Web 2.0 B2C companies successfully leveraging Web 2.0
to supplement and enhance their marketing ef- technologies to market their products. Applying a
forts. participation lens to the marketing mix can lead
similar companies to similar success.
To be more successful at marketing using Web
Solutions and 2.0 technologies, the traditional four-p marketing
Recommendations: A mix discussed earlier (price, product, promo-
Revised Marketing Mix to tion, and placement) should be approached with
Maximize the Benefits of a new p lens: participation. This is the Web 2.0
Web 2.0 Technologies marketing mix.
In this new marketing mix, each of the four
A characteristic of Web 2.0 is an architecture of elements is approached by enabling consumers to
participation that encourages users to add value participate in it. Let us explore how consumers are
to the application as they use it. Tapscott and currently participating in each p of the marketing
Williams (2006) argue in their book Wikinomics: mix and the steps a company can take to capitalize
How Mass Collaboration Changes Everything on that interaction and encourage more.
that the economy of Web 2.0 is based on mass Web 2.0 technologies provide a medium for
collaboration that makes use of the Internet. consumers to directly participate with companies
Companies can leverage the collective power of and their brands. Consumers can write blogs about
their consumers by leveraging Web 2.0 technolo- their favorite brands, define themselves and create
gies to enable the consumers to participate in the communities in social networks like Facebook and
marketing of the brand. MySpace by associating themselves with brands,

915
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

develop their own commercial advertisement for reasoning or decision making by a group that often
brands and post them on YouTube, and tag pictures occurs in traditional focus groups. Companies can
they have uploaded of their brands on Flickr. also learn more about the total product life cycle
In addition, Web 2.0 technologies enable com- beyond just the physical product, and identify
panies to more readily identify brand evangelists: additional uses for it. Companies can gain valu-
those consumers that are devoted to a brand or able customer insight and interaction via Web 2.0
product and preach their devotion to the world. in a quicker and cheaper fashion than traditional
Companies can make it even easier for consum- market research.
ers to participate in their marketing by evaluat- Flickr can be an avenue for trend spotting.
ing each p in their marketing mix and fostering Trend spotting is a relatively new consumer
involvement by their consumers via Web 2.0 research methodology that seeks to anticipate
technologies. Let us specifically examine each what consumers will desire in the future and to
marketing mix element. keep existing products relevant. For example, on
Flickr, a company can search for what consumers
Product are carrying in their purses. Companies can ask
consumers to participate in identifying product
The product aspects of the marketing mix are the trends by sending out a call to action to post
specifications of the actual goods or services, and photos of what is in their purses. Maybe Capital
how they relate to the user’s needs and wants. One with their “What’s in your wallet?” campaign
The scope of a product generally includes sup- could ask consumers to post Flickr photos of
porting elements such as warranties, guarantees, what is in their wallets, including a Capital One
and support. credit card. While a company may not know a lot
In Wikinomics: How Mass Collaboration about the consumer, Web 2.0 enables it to reach
Changes Everything, Tapscott and Williams into consumers’ lives and learn more about their
(2006) explore a new generation of “prosumers” behaviors (Brighton, 2005).
who treat the world as a place for creation, not Lego fostered an early prosumer community.
consumption. These prosumers can be engaged Lego’s Mindstorms enables users to build work-
to participate in all aspects of a product, from ing robots out of programmable bricks. Users
identifying consumers’ unmet needs to the de- reverse-engineered the products and shared
velopment of a new product to supporting the feedback with Lego. Lego developed a Web site
product once in a consumer’s hands. Companies for users to share their discoveries and inventions
need to encourage consumers to participate in their with other enthusiasts and the company. Users
products and recognize the contributions that are can even virtually develop their own models and
made. There is an allure of prestige and sense of then order the bricks to physically build it. This
social belonging that develops within prosumer enables Lego’s consumers to become a decen-
communities (Tapscott & Williams). tralized virtual design team, far larger than the
Web 2.0 provides immediate feedback about number of the in-house designers (Tapscott &
products. Companies do not need a Web site to Williams, 2006).
harness the power of the numerous discussions Procter & Gamble is also successfully leverag-
of products on Web sites beyond the control of ing the powers of Web 2.0 to enhance its research
the company. Blogs are a great tool to identify and development efforts for new products. The
latent consumer needs and wants. Communities research and development team had a success
are free focus groups of very raw, unscripted rate of less than 20%, below industry standards
feelings not tainted by groupthink, or the act of of 30%. Via the Web, Procter & Gamble turned

916
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

to the outside world for new and better ideas, and Pricing
now more than 35% of the ideas come through
the Web, resulting in success for 80% of Procter The pricing p of the marketing mix refers to the
& Gamble’s new product launches. process of setting a price for a product, including
One example was the discovery of a way to put discounts. The price need not be monetary; it can
edible ink pictures on potato chips. The solution simply be anything exchanged for the product or
came from an Italian professor at the University services, including time, energy, psychology, or
of Bologna who had invented an ink-jet method attention.
for printing edible images. This technology helped Companies can leverage Web 2.0 to learn more
the company get the new Pringles Prints potato about value that consumers perceive about brands
chips out in a single year, about half the normal and what consumers perceive they are paying for
time for such a process (Stephens, 2007). a product beyond just the monetary value. For
Cadbury brought back the discontinued Wispa example, online Web services such as eBay and
chocolate bar after a campaign on Web sites like craigslist let the consumer (or market) determine
Facebook, MySpace, and YouTube demanded the price for products.
its return. The chocolate company says that it is Consumers are able to discuss prices online via
frequently contacted by consumers asking for old blogs or social networking sites for plane tickets,
favorites to be reintroduced, but said the numbers HDTVs, furniture, cars, and so forth. They can
that had joined the Internet campaign to relaunch even let other consumers know where to find the
Wispa were unprecedented (Wallop, 2007). best discounts.
In the support stage of the product life cycle, A mashup is a Web page or application that
brands have an opportunity to engage with con- integrates complementary elements from two or
sumers during a crisis using Web 2.0. Consumers more sources. The most popular mashups include
are pretty responsive to companies and brands Google Maps as a source to identify certain things
who engage, especially when there is a problem. on a map. For example, one mashup shows secret
Speaking at Nielsen Business Media’s Next Big fishing holes in the United States via a Google
Idea Conference, EVP of strategic services at map. Mashups help locate the best price, such as
Nielsen Online Pete Blackshaw focused his re- CheapGas.
marks on brand intelligence as he explained how Beyond monetary value, companies can also
companies can “defensively” market their prod- get see what consumers are forfeiting other than
ucts by turning negative trends in the marketplace money by choosing their brands or products. For
to their advantage. Blackshaw noted the surge example, a number of loyal Starbucks consumers
in the blog traffic surrounding consumer crises, are concerned about the impact they are making to
such as the Mattel toy lead paint and poisonous the environment by enjoying Starbucks coffee in
pet food recalls from brands including Alpo and a new cardboard cup everyday with a cardboard
Mighty Dog, are opportunities for brands to have sleeve. These consumers found a Web site that sold
a touch point with consumers. Companies, said inexpensive reusable cloth holders to protect hands
Blackshaw, should learn to “manage around the from the hot coffee cup and were elated to share it
spikes, listen, react,” and move money out of mass with one another. Starbucks could learn about this
media and into online channels, asking consumers concern of its consumers and address it, possibly
to participate in the support of a product, even by offering this reusable cloth sleeve product in
during a recall (Kiley, 2007). stores or developing one of their own.

917
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

Promotion check for accuracy but not stifle the participation


by their consumers to define the brand.
Promotion is the element of the marketing mix A key to enabling consumers to participate in
where consumers can most readily participate promotion is to make it easy to search and find
and add value to the marketing of a brand. This areas where other consumers talk about your
includes advertising, sales promotion, public brand, product, or service, or create your own
relations, and personal selling, and refers to the site where consumers can talk about what you
various methods of promoting the product, brand, are interested in learning more about. Procter
or company. & Gamble developed a social networking site
Consumers can develop an advertisement for for women called Contessa. The intention of the
a product or brand and then publish that ad for social network is not about selling products, but
free on YouTube, tagged with keywords to inform for P&G to learn more about its women consum-
others of the ad. ers and learning about their needs and habits
Companies may also use Web 2.0 to promote (Ives, 2007).
their brands with an advertisement. Smirnoff de-
veloped a viral marketing video for the launch of Placement
their new Raw Tea product. The video was placed
on YouTube and at this time had over 4 million Placement or distribution refers to how the
views (Iamigor, 2006). The Web 2.0 technol- product gets to the customer, for example, point-
ogy enabled consumers to access the video on of-sale placement or retailing. This fourth p of
YouTube and then participate in disseminating the marketing mix has also sometimes been
the information by informing their friends about called place, referring to the channel by which a
the video. The launch of Raw Tea also included a product or service is sold (e.g., online vs. retail),
Web site with the videos and a sharing capability. to which geographic region or industry, and to
By placing the advertisement on a Web 2.0 site, which segment.
Smirnoff made it easier for their consumers to Companies can distribute their products via
participate by rating and sharing the video on a Web 2.0. American Apparel operates a clothing
site they already frequent. store in Second Life that sells virtual clothing for
A wiki is any collaborative Web site that users avatars. American Apparel has run promotions
can easily modify via the Web, typically without in Second Life where after purchasing a cloth-
restriction. A wiki allows anyone using a Web ing item in Second Life, a consumer receives a
browser to edit, delete, or modify content that coupon for a discount on the same or similar item
has been placed on the site, including the work of at an American Apparel store in the real world
other authors. One popular wiki is Wikipedia, a (Jana, 2006).
free encyclopedia that anyone can edit. Wikipedia Coke established the Virtual Thirst Pavilion
is updated every second by thousands of active in Second Life. It sponsored a contest to develop
contributors, making it an up-to-date reference a Virtual Thirst vending machine. The winning
source vs. a printed encyclopedia that is updated vending machine will be rolled out throughout
monthly or yearly. Second Life, making this real-world distribution
Wikis capture the knowledge of the collective channel as ubiquitous in Second Life as it is in the
whole. Consumers can define your product or real world. Coke did not intend for consumers to
brand on Wikipedia. Consumers define what a merely replicate an existing real-world vending
brand stands for or describe a product, including machine but to create a portable device for Second
its intangible intrinsic value. Companies should Life’s in-world digital society that unleashes a

918
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

refreshing and attention-grabbing experience, on What does it say about a brand to have a pres-
demand. Our goal is to enable individual creativ- ence or be communicated via these technologies?
ity in pursuit of a “vending” machine that can Before embracing participation in Web 2.0, a
exist only in your wildest imagination. Virtual company must determine if a brand is compatible
worlds make it possible for such innovations to with the spirit of Web 2.0. Web 2.0 is more about
occur, and we selected Second Life as the most spirit, concepts, and principles than definition. It
conducive to this experiment, is imperative that a brand be in accordance with
says Michael Donnelly, director of global that spirit before launching an initiative. If a brand
interactive marketing (Coca-Cola, 2007). This is incompatible with this experience of openness
Second Life contest establishes a new distribu- and exchange, it is advised to create or use another
tion channel for the brand in a virtual world, but brand or subbrand as a workaround for Web 2.0
also taps into consumers’ creativity for ideas for initiatives to protect the integrity of the core brand
real-world distribution mechanisms. (Smagg, 2007).
On Flickr, in addition to tagging photos, a The best action for a company to take may be
user may add notes that are visible when some- no action at all as long as it is recognized that Web
one viewing the photo scrolls over a particular 2.0 technologies are having an effect on marketing.
aspect of it. For new product introductions, such In summer 2007, Wired magazine exposed Second
as a new pair of Nike shoes, a photo of the shoe Life for not achieving the commercial potential that
may be uploaded to Flickr with associated tags was initially expected. A trip to Starwood Hotels’
(Nike, new, shoe, blue, limited edition, New Aloft Hotel in Second Life was described as creepy
York store) and notes that can link to a Web site due to the entire place being deserted, compared to
where the shoe may be purchased. the movie The Shining, where a tenant at a deserted
Companies can even create a mashup using hotel goes psycho due to the isolation—not exactly
Google Earth to locate products that are not wide- the reputation Starwood wanted for its new hotel
ly distributed such as specific shoe sizes. Online chain brand. The NBA sought to capitalize on Web
retailers such as Amazon.com and Travelocity 2.0 for its marketing efforts by developing both an
are building widgets that can drive traffic to their island in Second Life and a channel on YouTube.
sites for consumers to make purchases. The YouTube channel saw over 14,000 subscrib-
ers with 23 million views, while the Second Life
island had a mere 1,200 visitors. Those numbers
Challenges are not as surprising when it is revealed that the
traffic in Second Life is slightly more than 100,000
The consumer defines the brand; a brand is not Americans per week (Rose, 2007).
what a company defines it as, but what a consumer In late 2007, IBM broadcast a television adver-
says it is. Via Web 2.0, consumers have numer- tisement featuring two employees of an unknown
ous avenues to add their interpretation of brands, company talking about avatars from a virtual world.
for better or worse. Brands need to relinquish The dialogue is as follows:
control to get influence. Two barriers to adoption
are regulation of content and security. “This is my avatar. It’s all the latest rage. I can
One of the reasons Web 2.0 technologies like do business, I even own my own island! It’s
YouTube and Flickr are successful is because innovation!”“But...can you make money?”“Um...
they are authentic: “The lack of corporate pol- virtual or real money?”“Real money. The point of
ish adds to the feeling that there are real people innovation is to make real money.”“Oh. My avatar
behind the idea” (Moore, 2006). doesn’t know how to do that.” (Vielle, 2007)

919
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

While the advertisement is mocking the value organizes blogs and other forms of user-generated
of Web 2.0 technologies, in a roundabout way, content (photos, videos, voting, etc.). Technorati
IBM is capitalizing on Web 2.0 to market its “Stop is currently tracking over 110 million blogs and
Talking. Start Doing.” campaign. Yet, in mid-2007, over 250 million pieces of tagged social media.
IBM established a presence in Second Life, the The home page will immediately update you on
virtual IBM Business Center staffed by real IBM “what’s percolating in blogs now,” but a quick
sales representatives from around the world. In the search of your brand’s name will share what is
press release for the launch of this virtual center, hot about your brand at the moment. It is an easy
IBM boasts it has over 4,000 employees active way to monitor the brand’s image (Technorati,
in Second Life. The question remains: Is IBM not 2007).
seeing the value from this investment in a Web Quick monthly visits to Web 2.0 Web sites
2.0 technology, or did it create the advertisement can also keep marketers up to do date about their
mocking avatars merely to generate interest for brands (and the competition). Visit Flickr and type
the IBM brand because avatars are popular at the in your brand name. Cheerios returns over 8,000
moment? It also could have been buzz for its own images, many showing how Cheerios plays a role
presence in Second Life (IBM, 2007). in the everyday life of a child. Marketers will be
happy to find this is consistent with the brand’s
image, values, and positioning.
Immediate Action Visit YouTube and type in your brand name. A
for Companies McDonald’s marketer may be interested to find that
some commercials are tagged as creepy and racist
Companies should use Web 2.0 to assist with and are not positively portraying the megabrand.
keeping a pulse on their brands’ and products’ However, a video containing the McDonald’s
involvement in Web 2.0 technologies. Most brands menu song has had over 1.5 million views. The
already have a Web 2.0 presence, and most likely, McDonald’s menu song was a song listing all of
it is not officially endorsed by the brand. McDonald’s menu items from a promotion in 1989.
Web 2.0 technologies make it very simple for Reviewing the thousands of comments about the
a marketer to quickly and easily learn about what video reveals that consumers are excited to view
consumers are saying about a brand or product. this video on YouTube and remember memorizing
RSS or real simple syndication is any of various the song in 1989 when it was advertised during
XML (extensible markup language) file formats television commercials. This is definitely some
suitable for disseminating real-time information valuable insight available for McDonald’s.
via subscription on the Internet. RSS has become a Visit MySpace and type in your brand name
popular technology for bloggers and podcasters to or visit http://www.myspace.com/brandname.
distribute their content. NewsGator is a free Web- Mountain Dew does not have an official page,
based RSS news reader that consolidates news and but a 21-year-old from Southern California owns
updates from the Web, blogs, premium content http://www.myspace.com/mountaindew. JC from
providers, and internal applications and systems Newport Beach, California, owns Nike. In addition
and automatically delivers them to users. to learning about consumers’ thoughts on your
RSS feeds can be set up to find and aggre- brand, this exploring can also result in insights
gate information about a brand or product from about your loyal consumers who add your brand
blogs, podcasts, and so forth, and be sent out via to their MySpace pages.
e-mail every morning to keep a marketer updated. Visit Wikipedia and type in your brand name.
Technorati is an RSS service that searches and Wikipedia’s American Express entry has a wealth

920
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

of historical information, including advertising. Conclusion


The entry is up to date with AmEx’s most recent
promotions, including its Member’s Project. Web 2.0 technologies will not replace traditional
Anyone can contribute to a wiki, so periodically marketing such as direct mail or TV advertise-
verifying information is correct and adding updates ments, but instead are new complementing mar-
can benefit a brand (“Web 2.0,” 2007). keting channels that many consumers will expect
With Web 2.0 technology, companies can their brands to communicate through.
quickly and inexpensively make things happen. Companies that feel Web 2.0 technologies are
You can have your advertising messages spread on a right fit for their brands should take action to
the Web like a wildfire with social bookmarking market their brands using the Web 2.0 marketing
sites, RSS, and other Web 2.0 methods, and without mix as a guide. Enabling consumers to participate
having to pay anything for it, have thousands of in each aspect of the marketing mix will help brands
people coming to your Web site in a matter of days. remain relevant in today’s changing world.
What would have cost you millions of dollars in
investment and a dedicated team of developers
may now be accomplished with these Web 2.0 References
tools by a couple of guys in a garage in just a few
days (Beaudoin, 2007). American Marketing Association. (2007a). Beyond
Web 2.0 technologies intermingle with each Marketing 2.0: Harnessing the power of social
other as well, so a presence in one Web 2.0 tech- media for marketing campaign results. Retrieved
nology can give you a presence in others. For December 28, 2007, from http://www.marketing-
example, MySpace allows its members to blog power.com/aevent_event1221988.php
on their pages and add YouTube videos. American Marketing Association. (2007b). Dic-
tionary of marketing terms. Retrieved October
15,2007, from http://www.marketingpower.com/
Future Trends mg-dictionary-view1876.php

The Web is not going anywhere and will only Beaudoin, J. (2007). Web 2.0 for marketers ebook
continue to evolve. More Web applications will review on Web 2.0 portals. Retrieved November
be developed and compete for consumers’ time 26, 2007, from http://www.web20portals.com/
and attention. As consumers become accustomed web-20-marketing/web-20-for-marketers
to communicating and collaborating on the Web,
Big Sky Brands. (2007). MySpace. Retrieved
they may choose to take marketing your brands
November 26, 2007, from http://www.myspace.
into their own hands. Companies should be de-
com/bigskybrands
lighted at this occurrence. It is best to identify
and monitor these actions as they transpire, but Brighton, G. (2005). Flickr kills trendspotting.
not take actions to impede them. If a consumer PSFK. Retrieved September 27, 2007, from http://
is passionate enough to devote energy toward psfk.com/2005/03/flickr_kills_co.html
positively marketing your brand, you can be sure
that once disgruntled they will also take action,
only this time it could be detrimental.

921
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

Carter, B. (2007, October 4). Fictional charac- Jana, R. (2006, June 27). American Apparel’s vir-
ters get virtual lives, too. The New York Times. tual clothes. Business Week. Retrieved September
Retrieved November 12, 2007, from http://www. 26, 2007, from http://www.businessweek.com/
nytimes.com/2007/10/04/arts/television/04CSI. innovate/content/jun2006/id20060627_217800.
html?_r=1&ex=1192161600&en=6994ec8ad htm
88ab2d8&ei=5070&oref=slogin CBS mobile
Jaokar, A. (2006, October 25). Ajit Jaokar’s mobile
launches widget to promote new animated series.
Web 2.0 blog: The widget widget Web. Retrieved
(2007). MuseStorm. RetrievedDecember 28,
October 12, 2007, from http://www.web2journal.
2007, from http://www.musestorm.com/site/jsp/
com/read/289798.htm
spotlight/1042.jsp
Kiley, D. (2007). Using bad news to a brand’s
Coca-Cola. (2007, April 16). Coca-Cola launches
advantage. Business Week. Retrieved November
competition to design online “virtual thirst”
14, 2007, from http://www.businessweek.com/
coke machine. Retrieved October 14, 2007,
the_thread/brandnewday
from http://www.virtualthirst.com/virtualthirst-
socialmediarelease.html Kraft Foods. (2007). Kraft foods goes digital to
unveil more than 70 new products in first-ever
Guiragossion, L. (2007, October 29). A thousand
virtual supermarket in Second Life. Retrieved
ways to widget in the age of Web 2.0 market-
October 12, 2007, from http://www.kraft.com/
ing. Web 2.0 Marketing. Retrieved December
mediacenter/country-press-releases/us/2007/
28, 2007, from http://web2pointzeromarketing.
us_pr_04192007.htm
blogspot.com
Millions of Us. (2007). Retrieved October 12,
Hayes, G. (2007). A video comp of major
2007, from http://www.millionsofus.com
brands in Second Life. Retrieved Decem-
ber 28,2007, from http://www.youtube.com/ Moore, J. (2006). Five reasons why… Retrieved
watch?v=tEGHJuCbGdo October 16, 2007, from http://brandautopsy.
typepad.com/brandautopsy/2006/08/five_rea-
Iamigor. (2006, August 2). Tea partay. Retrieved
sons_wh.html
October 15, 2007, from http://www.youtube.com/
watch?v=PTU2He2BIc0 Mootee, I. (2004). High intensity marketing.
Canada: SA Press.
IBM. (2007, May 15). Live IBM sales people to
staff new virtual IBM business center. Retrieved Rose, F. (2007). How Madison Avenue is wasting
October 15, 2007, from http://www-03.ibm.com/ millions on a deserted Second Life. Wired Maga-
press/us/en/pressrelease/21551.wss zine. Retrieved November 12, 2007, from http://
www.wired.com/techbiz/media/magazine/15-08/
Ives, B. (2007, December 1). More Web 2.0
ff_sheep?currentpage=all
stories, part two: Procter and Gamble embraces
the wisdom of the Web for new product ideas. Second Life, R. E. P. E. R. E. S. (2007). Retrieved
The FASTForward Blog. Retrieved December October 12, 2007, from http://www.reperes-
28, 2007, from http://www.fastforwardblog. secondlife.com
com/2007/12/01/more-web-20-stories-part-two-
Smagg, C. (2007). 15 golden rules for Web 2.0.
proctor-gamble-embraces-the-wisdom-of-the-
Retrieved October 12, 2007, from http://visionary-
web-for-new-product-ideas
marketing.wordpress.com/2007/07/03/web20

922
A Strategic Framework for Integrating Web 2.0 into the Marketing Mix

Steel, E. (2006). Web-page clocks and other Vielle, T. (2007). IBM avatar commercial. SL Uni-
“widgets” anchor new Internet strategy. Wall verse. Retrieved December 31, 2007, from http://
Street Journal. Retrieved November 12, 2007 sluniverse.com/php/vb/showthread.php?t=2797
from http://marcomm201.blogspot.com/2006/11/
Wallop, H. (2007). Cadbury plans Wispa re-
marketing-widgets-you-saw-it-here-first.html
vival. Telegraph. Retrieved November 28, 2007,
Stephens, R. (2007). P&G Web 2.0 success story. from http://www.telegraph.co.uk/news/main.
Retrieved December 27, 2007, from http://www. jhtml?xml=/news/2007/08/18/nwispa118.xml
rtodd.com/collaborage/2007/11/pg_web_20_suc-
Web 2.0. (2007). Wikipedia. Retrieved Septem-
cess_story.html
ber 15, 2007, from http://en.wikipedia.org/wiki/
Tapscott, D., & Williams, A. (2006). Wikinomics: Web_2.0, http://en.wikipedia.org/wiki/Ameri-
How mass collaboration changes everything. New can_Express
York: Portfolio.
Technorati. (2007). Retrieved December 28, 2007,
from http://technorati.com
Terret, B. (2007). Brand in good Web 2.0 project
shock. Noisy Decent Graphics. Retrieved Decem-
ber 28, 2007, from http://noisydecentgraphics.
typepad.com/design/2007/11/brand-in-good-w.
html

This work was previously published in Social Software and Web 2.0 Technology Trends, edited by P. Deans, pp. 29-43, copyright
2009 by Information Science Reference (an imprint of IGI Global).

923
924

Chapter 4.4
Applying Semantic Web
Technologies to Car Repairs
Martin Bryan
CSW Group Ltd., UK

Jay Cousins
CSW Group Ltd., UK

Abstract a ‘shared language’ for the project, a reference


terminology to which the disparate terminologies
Vehicle repair organizations, especially those in- of organisations participating in the project can be
volved in providing roadside assistance, have to be mapped. This lingua franca facilitates a single point
able to handle a wide range of vehicles produced by of access to disparate sets of information.
different manufacturers. Each manufacturer has its
own vocabulary for describing components, faults,
symptoms, etc, which is maintained in multiple Current Situation
languages. To search online resources to find re-
pair information on vehicles anywhere within the Repair scenarios for resolving a vehicle breakdown
European Single Market, the vocabularies used to are varied, and can take place in a garage (repair by
describe different makes and models of vehicles a qualified mechanic in a franchised or independent
need to be integrated. The European Commission workshop) or by the roadside (repair by a quali-
MYCAREVENT research project brought together fied mechanic working for a Road Side Assistance
European vehicle manufacturers, vehicle repair (RSA) organisation, or a repair by a vehicle driver).
organisations, diagnostic tool manufacturers and For legal liability reasons, ‘driver-assisted’ repair
IT specialists, including Semantic Web technolo- scenarios only cover minor or temporary repairs
gists, to study how to link together the wide range of the type covered in owner’s manuals, such as
of information sets they use to identify faults and changing a vehicle wheel or a fuse.
repair vehicles. MYCAREVENT has shown that In workshop scenarios, access to repair informa-
information sets can be integrated and accessed tion may be provided through online access to repair
through a service portal by using an integrated information systems. Information may be provided
vocabulary. The integrated vocabulary provides publicly by a manufacturer for all users, or specifi-
cally to franchised dealers who are provided with
DOI: 10.4018/978-1-60566-066-0.ch002

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Applying Semantic Web Technologies to Car Repairs

access to information systems that are specific to be carried out before the cause of the problem
to the makes and models they retail. can be identified.
Access to repair information in roadside sce-
narios is more complicated. A vehicle driver may
not have access to the vehicle’s owner’s manual. Problem Statement
In the context of a roadside repair by a mechanic
working for an RSA, the mechanic might have With the ever increasing use of electronics in
access to repair information through a computer vehicle components, identifying and correct-
located in their van. RSA organisations, however, ing faults at the roadside or in an independent
rely heavily on the detailed knowledge of their workshop is becoming a challenge. While the
highly trained staff to diagnose faults without use of on-board diagnostic tools to report faults
the aid of documentation. RSA mechanics aim to electronically via dashboard messages can assist
repair as many vehicles as possible at the roadside, mechanics, identifying the cause of a fault from
but need to identify as early as possible if a car such messages is not always a simple process.
will need to be taken to a garage for repair. If When faults are reported over the phone from
the repair requires specialist equipment the RSA remote locations sufficient diagnostic informa-
must be able to identify the nearest garage with tion may only be obtainable if the vehicle can
suitable equipment that the car may be taken to be connected directly to the call centre informa-
for repair. tion centre using tools such as personal digital
Fault diagnosis precedes vehicle repair in assistants (PDAs) or mobile phones that can be
both repair scenarios. Details of the type of fault connected to the vehicle’s diagnostic ports.
are ascertained at the point of contact with the A roadside assistance vehicle cannot contain
customer, be this through direct conversation the wiring schematics for all models of vehicles.
with the vehicle owner at a service centre, or Although, under European Union Block Exemp-
by conversation through a call centre operator tion Regulation (European Commission, 2002),
when a motorist initially reports a problem. manufacturers provide access to all their repair
When contact is made through a phone call it is information, repairers at the roadside are not
important that call centre operators analyze the always easily able to find the repair information
customer’s situation in as much detail as pos- that they need, particularly if this is related to a
sible. They have to be able to identify whether previously unreported fault, while physical and
the problem is one that might be repairable at the business constraints impose restrictions on the
roadside or whether a recovery vehicle is likely set of spare parts, tools, etc, that can be available
to be needed from the responses received to an in the workshop or repair van at any one time.
ordered set of questions. Consequently, the following problem areas can
Customer contacts rarely lead to a detailed be identified:
fault diagnosis because vehicle owners typically
have insufficient knowledge of their vehicles to • Practical limitations exist on the level of
identify the cause of a problem. At best they can information that can be provided in any
describe the symptoms produced by the fault and repair context. There is variability in the
the conditions in which the symptoms manifest amount and quality of information that is
themselves (e.g. won’t start when it is too cold). available to describe a fault and its associ-
In many cases these descriptions can be used to ated symptoms and conditions in order to
identify the type of diagnostic tests that may have support fault diagnosis.

925
Applying Semantic Web Technologies to Car Repairs

• Environmental variables such as geo- Solution Description


graphical location, repair equipment, and
spare part availability may combine to The Mobility and Collaborative Work in European
constrain the speed with which a repair can Vehicle Emergency Networks (MYCAREVENT)
be affected, and determine the location at research project was sponsored by the IST
which the repair takes place. (Information Society Technology) program of
• Logistics and supply chain management the European Commission. The 3-year project
and facilitation can provide advance warn- brought together leading manufacturers from
ing of required spare parts or repair equip- the automotive sector, academic researchers and
ment at the point of initial fault diagnosis, commercial IT suppliers to develop facilities for
supporting decision processes such as the the provision of repair information to remote
direction of a vehicle to an appropriate users in the automotive aftermarket. Remote
repair location or the triggering of inven- access, for example a roadside repair, is enabled
tory supply processes to pre-order required by the use of mobile services. Research focused
parts and arrange their delivery to the re- on service development, process and organiza-
pair location. tion management, e-business, communication
• Maintenance of acceptable response times networks and human-computer interaction.Work
that meet customer expectations. in MYCAREVENT has been organized in nine
work packages:
The MYCAREVENT project addresses these WP 1: Project Management
issues by facilitating the diagnosis of faults and WP 2: Use Case and Business Model
the provision of repair information at the loca- WP 3: Ontologies
tion where the fault is first described, be this in a WP 4: Mobile Communication
workshop or at the roadside. WP 5: Remote Services
The MYCAREVENT project provides a single WP 6: Service Portal
point of entry – a portal – through which a user WP 7: Mobile Applications
can access services to support the description and WP 8: Training
diagnosis of a fault, and to search for and retrieve WP 9: Demonstration and Dissemination
repair information from a variety of content The relationships between these packages are
providers. For this to be achievable, however, it illustrated in Figure 1.
must be possible to associate the terms used by WP1 supported project management, estab-
the vehicle owner to describe the problem that lishing and monitoring the administration and
has occurred with the terms used by the content leadership of the MYCAREVENT project. It
provider to describe content or how to detect and needs no further discussion here.
solve the fault causing the problem. It should be WP2 developed the fundamental business
noted that content can be of variable quality and model and use cases that scope the project re-
scope – for example, repair information from a quirements and identify the customers, actors,
vehicle’s manufacturer will typically apply to spe- processes, and the constraints on the legal and
cific makes of vehicle, whereas information from organisational environment within which the
third parties like technical specialists working for project solution operates.
RSAs, or technical data for an automotive part or The project scope can best be understood
content from a third party information provider, by looking at the project’s three pilot scenarios,
may be more generic in application. which demonstrate the functionality provided by
the solutions and identify their targets:

926
Applying Semantic Web Technologies to Car Repairs

Figure 1. MYCAREVENT work package relationships

• Pilot I was designed to demonstrate possi- exchange of fault codes and repair information
ble solutions for original equipment manu- from remote locations, such as those required to
facturer (OEM) workshops and OEM road- carry out roadside repairs.
side technicians who require remote access The remote services work package (WP5) al-
to the MYCAREVENT service portal to lows a driver to search for self- help information,
obtain instructions for specific repairs. such as that provided in the owner’s handbook,
• Pilot II was designed to demonstrate how using standard Web browser software that resides
the concept of access to car repair informa- on their smart-phone or PDA.
tion via the MYCAREVENT service portal The service portal work package (WP6) defines
could be extended to help mechanics work- the core project portal, the gateway for accessing
ing in independent workshops and roadside repair information.
assistance services to identify faults. The mobile applications work package (WP7)
• Pilot III was designed to demonstrate the allows the MYCAREVENT Service Portal to
concept of “Driver Self Help” in those sce- be used to deliver automated diagnostic tests to
narios where the driver can carry out sim- trained mechanics. This requires the application
ple repairs using advice provided by the of additional access security and other middleware
MYCAREVENT service portal. services within the portal interface.
The role of WP3 Ontology and WP6 Service
The ontology work package (WP3) defines the Portal work packages is explained further in the
information structures that enable cross-system following sub-sections.
interoperability and the integration of content from
disparate sources and heterogeneous databases.
The mobile communications and devices work Objectives
package (WP4) provides a secure and reliable
communication service between users (roadside The MYCAREVENT Service Portal acts as a
assistants, drivers, and mechanics) and the service gateway to technical information on automotive
portal. These services are intended to enable the diagnosis, repair and maintenance that is available

927
Applying Semantic Web Technologies to Car Repairs

from automotive manufacturers and independent • A Generic and integrated information ref-
organisations supporting the aftermarket. erence model (GIIRM) (MYCAREVENT,
To ensure high user acceptance, the MY- 2005), providing a high-level conceptual
CAREVENT work packages use innovative model of the MYCAREVENT mobile ser-
state-of-the-art technologies to find the ‘right’ vice world.
information for user needs. To make this pos- • A set of W3C XML Schemas derived from
sible the service portal includes the following the GIIRM, which are used for the repre-
subsystems: sentation of data in messages, metadata
and interfaces.
• Core e-business infrastructure for the flex- • Terminology for populating the GIIRM,
ible implementation of workflow and busi- enabling repair information, symptoms
ness processes. and faults to be described in a generalized
• Service data backbone providing secure way.
links to services as well as manufacturer • A W3C Web ontology language (OWL)
and third party information repositories. (McGuinness, 2004) ontology, derived
• An ontology-based advanced query ser- from the GIIRM and the terminology, in
vice (AQS) for guided navigation through which data sources can be registered for
different data resources and terminologies. access by MYCAREVENT applications.
• Expert system hub combining the capabili-
ties of distributed (specialised) expert sys- Details
tem nodes.
• Authoring tools for specific types of tech- Since the publication of Tim Berners-Lee’s fu-
nical information, such as the interactive turistic paper on The Semantic Web in Scientific
circuit diagrams (IACD) used to identify American in May 2001 (Berners-Lee, 2001) the
faults in electronic systems. concepts that form the backbone of a system that
can add semantics to Web resources has begun to
The remainder of this section explains how form. As was pointed out in that paper:
the ontology-based advance query service applies
Semantic Web technologies to identify solutions For the Semantic Web to function, computers must
to repair problems. have access to structured collections of informa-
tion and sets of inference rules that they can use
to conduct automated reasoning.
Overview
The goal of the MYCAREVENT ontology
The MYCAREVENT Ontology work package was work was to link together collections of informa-
responsible for the development of the models, tion created by different vehicle manufacturers,
data structures and terminology sets used to sup- component suppliers and repair organizations in
port the work carried out by the service portal. such a way that we can use the collected informa-
The work package drew on the expertise of data tion to conduct automated reasoning wherever
modelling specialists, implementers and content possible.
providers (including OEM and RSA organisations) The start point for the work package was the de-
to build an agreed set of ‘information artefacts’ to velopment of a formal model that could record the
be used across all MYCAREVENT services.The relationships between the information components
Ontology work package developed: used to identify and repair faults. This top-level

928
Applying Semantic Web Technologies to Car Repairs

model was designed to be generalized enough to tion provides a generic model independent of any
apply to any type of repairable product. detail specific to implementation. It is, therefore,
Figure 2 shows a diagrammatic representation a platform-independent model.
of the MYCAREVENT Generic and Integrated To simplify the process of identifying relevant
Information Reference Model (GIIRM) which information sources, the model includes the con-
was developed to manage the inter-relationship cept of a Term. Terms can be used to describe a
between information message components ex- vehicle instance, its build specification, a system
changed between information suppliers and the or subsystem used in the build specification, a
service portal. The diagram is expressed in the condition under which a problem was detected,
Object Management Group’s Unified Modeling a symptom of the problem or a detected fault.
Language (UML) (Object Management Group, Terms can be grouped into Terminologies that
2007). can be applied by different manufacturers within
The information required to populate the their documentation and information delivery
classes defined in this model are supplied by systems.
information providers in the form of Informa- Users of the MYCAREVENT Service Portal
tion Bundles that conform to the ISO/IEC 14662 may or may not be aware of the terms used by
Open-EDI Reference Model (ISO/IEC 14662, manufacturers within their documentation. Us-
2004). In this standard Information Bundles are ers need to be able to enter terms that they are
defined as: familiar with for describing problems, etc. The
MYCAREVENT Advanced Query Service (AQS)
The formal description of the semantics of the needs to be aware of the relationships between the
recorded information to be exchanged by parties terms applied by a particular user community and
in the scenario of a business transaction. The the terms applied by a particular manufacturer.
Information Bundle models the semantic aspects Terminology, and the mapping of terms in one
of the business information. Information bundles terminology to terms in another terminology, is
are constructed using Semantic Components. central to the AQS as it enables the querying of
disparate information sources. The parameters
of a search query can be established by a user
A unit of information unambiguously defined in using their preferred terminology, which may
the context of the business goal of the business be manufacturer-specific or generalized using a
transaction. A Semantic Component may be atomic MYCAREVENT specific term set. This search
or composed of other Semantic Components. context is then passed to the AQS, which trans-
lates the MYCAREVENT terminology into the
The model allows, therefore, for simple (i.e. terminology used to describe the content that will
‘atomic’) and composite (i.e. ‘non-atomic’) at- be searched to retrieve information. This level of
tribute values, represented in the model using indirection allows disparate corpora of information
the concept of the ‘representation class’ which to be searched using their own terminology. From
can be either an atomic datatype or a non-atomic an integration perspective, the AQS enables con-
composite data type as defined by the ‘Naming tent integration using metadata describing repair
and design principles’ established in Part 5 of the information content created by OEMs or third
ISO Metadata Registries (MDR) standard (ISO/ parties, or through direct interface to a vehicle
IEC 11179-5, 2005). The GIIRM has foundations information source system.
in abstract concepts and existing standardisation The AQS has been developed using open-
work. This design philosophy and layer of abstrac- source technology to enable the latest develop-

929
Applying Semantic Web Technologies to Car Repairs

Figure 2. MYCAREVENT generic and integrated information reference model

930
Applying Semantic Web Technologies to Car Repairs

ments in Semantic Web technology to be adopted. Searching and Retrieving


It uses the Jena RDF triple store (“Jena”, n.d.) to Repair Information
record OWL classes and individual occurrences of
these classes. The Jena 2 Database Interface allows The key to the success of the MYCAREVENT
W3C Resource Description Framework (RDF) portal is to allow users to ask questions using
(Klyne, 2004) triples to be stored in MySQL, terms that they are familiar with and to use these
Oracle, PostgreSQL or Microsoft SQL databases, questions to generate alternative versions of the
on both Linux and WindowsXP platforms. question that OEM and other information provi-
An important feature of Jena 2 is support for sion systems associated with the portal can answer.
different kinds of inference over RDF-based The workflow steps used to establish a search query
models (for RDFS, OWL, etc). Inference models are described in the following sections.
are constructed by applying reasoners to models MYCAREVENT queries are implemented in
(Dickinson I, 2005). The statements deduced by a controlled, context sensitive, manner to provide
the reasoner from the model can appear in the guidance to users as they enter information into
inferred model alongside the statements from the service portal. Figure 3 shows the information
the model itself. RDF Schema (RDFS) (Brickley, components used to identify the type of vehicle
2004) reasoning is directly available within Jena: to be repaired within the portal.
for OWL an external reasoner needs to be linked Users complete each field in turn. As they do
to the Jena engine through a Reasoner Registry. so the options available to them in the next field
The Pellet reasoner (Clark & Parsia, 2007) is used are restricted using the ontology. So, for example,
within MYCAREVENT to ensure that all inferred as soon as the user identifies the make of vehicle
relationships are identified prior to searching. Jena to be repaired, the set of options that can be used
includes an OWL Syntax Checker that can be used to complete the model field is reduced to the
to check that OWL files are correctly formed. set of models appropriate for the entered make.
Jena includes an implementation of the SPAR- Completing the model field restricts the range of
QL query language (Prud’hommeaux, 2007) called years that can be entered in the Year field, select-
ARQ. SPARQL has been developed as part of the ing a year restricts the set of Series that can be
World Wide Web Consortium (W3C) Semantic selected, and so on.
Web activity to provide a transportable technique For countries such as the UK where a vehicle
for RDF data access that serves a similar purpose registration number decoder is accessible it is
to the structured query language (SQL) used to possible to enter the vehicle registration number
access information held in a range of relational (VRN) into a field displayed under the Decoder
databases. The MYCAREVENT AQS generates tab and have the entered number return informa-
SPARQL queries, based on the objects in the tion that has been recorded about the year of
GIIRM, which are used to identify the concepts manufacture, engine type, fuel type, etc, by the
being referred to by terms entered by users. Be- vehicle registration authority. Alternatively the
cause the query service is based on RDF it can manufacturer’s vehicle identification number
query the contents of any OWL data property used (VIN) can be used to automatically identify sys-
to record information about an individual class tem components. Where a decoding service is not
member within the MYCAREVENT ontology, or available each field in the vehicle description has
any language-specific RDF label associated with to be completed in turn.
a class or individual, irrespective of whether or When as much information on the vehicle as
not it is a term that has been specifically declared is available has been recorded the user can be
within a terminology. shown a list of available services, which can range

931
Applying Semantic Web Technologies to Car Repairs

Figure 3. MYCAREVENT Repair Information Form

in complexity from a set of online technical tips The Role of the


through the use of the advanced querying system MYCAREVENT Ontology
to identify relevant information resources to the
use of an expert system to diagnose problems. The ontology used by the AQS consists of a series
By reducing the set of selectable options at of specializations of the concepts in the GIIRM.
each stage to that recorded in the ontology we not These concepts reflect the core classes and prop-
only refine the search criteria but also reduce the erties of the information required to describe and
likelihood that subsequent queries will be rejected. assert associations between faults, symptoms,
Only those services that are relevant to the type conditions, vehicles, vehicle systems, diagnostic
of vehicle to be repaired and the service that is tests, terminology, repair information documents
being employed to repair it are offered to users so and human language. These classes are further
that, for example, users are not prompted to carry sub-classed or specialised to refine concepts from
out tests using diagnostic equipment that is not the abstract and generic model level to a lower and
fitted to the vehicle or accessible to the repairer, more ‘concrete’ or ‘real world’ level that aligns
and are prompted to carry out all tests that must with the business requirements. To illustrate, if
be performed to identify a specific fault. business analysis shows that a particular informa-
tion provider has circuit diagram, repair procedure,

932
Applying Semantic Web Technologies to Car Repairs

Figure 4. Specialization of generic model classes

and owner manual types of repair information ties associated with the Vehicle class as they are
document, then the terminology used to describe displayed using the Protégé ontology editor used
document types is enhanced to include these new to maintain the MYCAREVENT ontology.
kinds of document, and within the ontology new As a value is assigned to each of these
document sub-classes are created to reflect these properties within the MYCAREVENT System
document types. The development process keeps Identification form, the set of related properties
the terminology and ontology in alignment with that can be found by querying the triple store
each other, and with the GIIRM which defines the is reduced. Only those entries that are used in
base model and so ensures interoperability across matched terms need to be displayed to users when
the portal and database(s) accessed by it. they are required to select options for a currently
Figure 4 shows how the Vehicle System class empty field. The order in which responses are
is specialized into two levels of sub-class. Each requested can be optimized to ensure that the
class has associated with it multilingual labels, minimum set of options is provided at each
and a set of properties. Figure 5 shows the proper- response point.

933
Applying Semantic Web Technologies to Car Repairs

Figure 5. Properties of a vehicle

In MYCAREVENT users never see the under- Alternatives


lying ontology while completing the basic Repair
Information form shown in Figure 3. They do On-line access to information for vehicles is avail-
not need to browse the class tree, or know which able from a variety of sources, including individual
properties they are dealing with. The ontology OEMs and third party automotive data suppliers.
is simply used as a Web service by applications Subscription models vary, but typically documents
that need to request data from users, so that they are only available to paying subscribers and may
can restrict the set of choices offered to users require the installation of specialist software or
to that appropriate for the currently identified training in the use of a specialized information
processing context, which is shown in the top retrieval system. For an individual user requiring
right-hand window on the display. The role of access to information for a variety of vehicles,
the ontology is, therefore, to reduce information the availability of a single portal through which
overload for users. multiple sources of information can be accessed
In later stages of the process, when accessing and queried is an attractive option. Rather than
data using the Symptoms and Title tabs, users paying multiple subscriptions and having the
are prompted to enter keywords that are searched burden of maintaining integration with multiple
against lists of symptoms or the titles of documents. information access points, the portal provides
In these scenarios matches can only be made in a central point of access from which a user can
those situations where the title contains relevant search for and retrieve information.
wording, or where the symptoms entered match Each manufacturer and third party information
symptoms recorded in one of the terminologies. provider maintains their own vocabulary of terms
for describing components, faults, symptoms, etc,
in multiple human languages. If a trained mechanic
knows which terms a manufacturer has applied

934
Applying Semantic Web Technologies to Car Repairs

to the fault they have identified then it is possible single European market, irrespective of language
to search the manufacturer’s database using a and affiliation, however, centralized diagnostic
free text search. But if a mechanic is not familiar services of the type that can be supplied using the
with the terms used by a specific manufacturer, MYCAREVENT service portal provide a viable
there are unlikely to be any synonyms for them alternative once agreement can be obtained from
provided by information suppliers. OEM and other content providers for making the
The issue of understanding the meaning of a information available online through the service
particular terminology is not confined to metadata portal.
describing different types of vehicle systems, but
applies to data recorded as codified values as
well. On-board and plug-in diagnostic devices Cost and Benefits
report faults using a set of codes. Some codes
are internationally agreed, and will identify the In 2004 the European Commission reported, in
same fault in a number of vehicles. Other codes their European Competitiveness Report (European
are manufacturer specific. As with terminology, Commission, 2004a) that the automotive industry,
descriptions and names for fault codes may be as one of Europe’s major industries, contributes
maintained in a variety of human languages. For about 6% to total European manufacturing em-
manufacturer specific codes the same code may ployment and 7% to total manufacturing output.
be used to identify different problems in different Total value added produced in the motor vehicles
makes of cars. Without access to a decoder that industry in the EU-15 in 2002 was roughly the
turns the codes into meaningful descriptions of same as in the US, some €114 billion.
faults for a specific vehicle build specification, With 209 million passenger cars in use in 2002
knowing the code does not necessarily help the the European Union (EU-25) is by far the largest
repairer. single market for cars in the world. It accounts
Manufacturers of diagnostic tools need to for roughly 38% of all cars on major international
ensure that their decoders are always up-to-date, markets. On average, four out of ten EU inhabitants
providing the correct interpretation of codes for own a car. According to the ACEA, the European
the latest models. Users of such tools need to be Automobile Manufacturers Association, 15 mil-
able to update their tools regularly to ensure the lion new passenger cars were registered in the EU
correct analysis of faults. While many OEMs and EFTA in 2006.
already provide an up-to-date tool set for their According to the EU report, it is expected that
vehicles through their BER portals, the extent 90% of all future innovation in vehicle manu-
of this is not consistent across manufacturers, facturing will be driven by IT. This affects both
and so a central point for updating information the electronics dominated spheres of multimedia
online provides one approach to addressing any entertainment and navigation systems and the
gap which may arise between the release of new traditional mechanical components such as the
vehicle types and the availability of tools to di- chassis, body, engine or brakes. For instance, the
agnose their problems, other than those provided percentage of electronics in the chassis is expected
by the vehicle manufacturer. to increase from 12% to 40% in the next decade.
By providing a central point for searching Similar developments are expected for safety fea-
for repair information, issues such as the need tures, e.g. pedestrian protection, traction control,
to ensure this content is up-to-date, and legal backward driving cameras, night-view display
liability for any claim arising from errors, need on the windscreen, sensor controlled brakes or
to be taken into account. For access across the fuel economy regulation. Product differentiation

935
Applying Semantic Web Technologies to Car Repairs

will be increasingly driven by electronics: for By using ontologies to establish relationships


example, performance tuned variants of the same between the terms used by vehicle owners and
engine will differentiate suppliers. The value of repairers to describe faults and the terms used
electronic components in vehicles could rise from by manufacturers to classify and describe faults,
its current 20% today to 40% by 2015. the MYCAREVENT Advanced Query Service
Since October 2002 motor vehicle distribution can provide a more flexible solution to finding
and servicing agreements within the EU have information, resulting in a higher likelihood of
come under the new Block Exemption Regula- mechanics being able to find the information they
tion (BER). Under the new regulations repairers need in a timely manner.
cannot be required to use original spare parts.
Only if repair costs arise which are covered by
the vehicle manufacturer, for example warranty Risk Assessment
work, free servicing and vehicle recall work,
can the vehicle manufacturer insist on the use of Each manufacturer produces thousands of models,
original spare parts. Other than that, matching each of which can have many build specifications.
quality spare parts of other manufacturers or of Model details and build specifications have to
independent suppliers can be used. be defined prior to manufacture, but cannot be
The automotive aftercare market had a turn- used within the service portal until the product
over of around €84 billion per annum at the end is released. Unless the release date is known in
of the 20th century; automotive replacement parts advance, data relating to vehicle models, build
account for around half of this figure, some 45% specifications, system components, etc, cannot be
of which is supplied by independent aftermarket added to the ontology at the time they are captured
(IAM) suppliers (European Commission, 2004b). by the manufacturer, but need to be made avail-
The 210 million motorists in the EU spend on able at the point when relevant documentation
average €400 each per year and approximately is released.
€5,000 during the average vehicle lifetime on Manufacturers are naturally reluctant to main-
repair and maintenance. tain two sets of information, which could get out
Major service providers in the automotive of step with one another. It must be possible to
industry are franchised dealers (120,000 deal- automatically convert information in local systems
ers employing 1.5 million people in 1999) and into the form that can be used by the AQS. As an
independent repair shops (160,000 garages alternative it should be possible to turn a query
employing about 600,000 people). In addition, to the AQS into a query to the manufacturer’s
18,000 roadside service vehicles fulfil 14 million product database.
missions a year. Another area of risk is in the level at which
By providing a single access point through data is described, and the equivalence of termi-
which details of electronically available informa- nology at different levels. A constraint on the
tion can be searched, using generic Semantic Web MYCAREVENT ontology is that it currently only
technologies rather than manufacturer-specific recognizes two levels in the system component
solutions, the MYCAREVENT Service Portal hierarchy, system and sub-system. If a manu-
simplifies and speeds up the task of finding in- facturer uses a multi-level system hierarchy this
formation on how to repair vehicles with specific needs to be flattened into a two level hierarchy
problems. With over a million potential customers for reference within the service portal. This means
it provides a cost-effective solution to information that entries at lower levels in the hierarchy have to
distribution. become members of the appropriate higher level

936
Applying Semantic Web Technologies to Car Repairs

sub-system, thus restricting the level of refinement • Identify other terms that include the largest
that can be applied to queries. This restriction is identifiable substring of the entered term
necessary because otherwise it would not be easy in compound nouns such as those used in
to convert AQS queries to queries that could be German.
applied to manufacturer-developed services that
can only handle two levels of querying. Expanding terminologies to cover all European
Where diagnostic information is a requirement languages, particularly agglutinative languages
for identifying build specifications and associated such as Finnish and Hungarian, where there are
repair information, obtaining the necessary infor- many compound words that could be derived from
mation without access to OEM-provided diagnostic a term, will make identifying potential matches
equipment can be a problem, especially in roadside much harder. For such languages it will be vital to
breakdown scenarios. Unless the repairer can send be able to define relationships between alternative
appropriate information to the portal it will not be references to a term within terminologies.
possible to retrieve relevant repair instructions. For Identifying which terms have significance in
this reason, other MYCAREVENT work packages which documents is another problem area. Unless
have concentrated on how to get information from the sub-systems that documents refer to are unam-
on-board diagnostic devices to a portal, diagnostic biguously recorded in either the data or metadata
tool or expert system capable of identifying the associated with a document, and the faults that
cause of the problem. can be solved using a document are recorded in
Not all concepts are applicable to all makes, or to the ontology or manufacturer’s information base,
all models made by a specific manufacturer. Where refining queries down to the level of identifying
a feature is specific to a particular manufacturer it documents that are specific to a particular problem
is not to be expected that other content providers with a given sub-system will not be possible. The
to the portal (be they manufacturers or third party best that can be achieved is to identify the set of
information providers) will have equivalent terms documents that refer to a particular sub-system and
in their terminology. If a user requests informa- allow users to determine from metadata describ-
tion on this subject for another make of vehicle ing the type of document, etc, whether it may be
the system will not be able to match the term. In suitable for solving the problem.
such cases a number of strategies can be adopted Relying on user selection of suitable documents
to find appropriate terms, including: introduces another risk. Manufacturers want to be
paid for preparing and supplying data. Users only
• Identifying the term in the terminology of want to pay for information that they know will
another manufacturer and informing the solve their problem more efficiently than alterna-
user that this term is manufacturer-specific tive solutions. If the cost of information is too high
• Identifying the sub-system with which the users will not risk purchasing something that may
term is associated by the originating manu- not solve the problem. If the cost of information is
facturer and offering a set of terms associ- too low manufacturers, or third party document-
ated with the same sub-system that are used ers, supplying the information will not be able to
by the manufacturer of the vehicle being recover the cost of preparing the information for
repaired distribution. Because of legal liability concerns,
• Identifying other terms that include one and the requirements by the BER, manufacturers
or more of the words in the entered term, are reluctant to supply information units which do
which may or may not identify higher-level not contain all the legally required warnings, safety
concepts notices, etc, that can apply to the repair scenario.

937
Applying Semantic Web Technologies to Car Repairs

OEMs want to supply units of information that be used to identify specific faults. At present this
are known to contain all relevant details for the information is generally not available from manu-
sub-system(s) that are connected with the fault. facturers. Until it is systematically recorded the
The rate at which documents change is also a efficient identification of faults within the portal
concern to information suppliers. It is not possible will be difficult.
to maintain an up-to-date repository that includes The recording of the symptoms reported when
all repair information generated by all vehicle a particular fault has occurred is, however, also an
manufacturers, even if an efficient enough content opportunity for the service portal. By recording
management system was available to store and the symptoms reported by users, the conditions
access them. The best a service portal can expect under which they occur and the fault that was
is to receive metadata about which documents are eventually identified as the cause of the problem
available for which sub-systems, and the roles within the service portal it should become pos-
those documents serve. If the metadata supplied sible, over time, to generate statistics that can be
with each document fails to identify the type of used to predict the likelihood of a particular fault
faults that the document can help to correct, it being the cause of an exhibited symptom.
will not be possible to associate faults with the The size of the European automotive industry
documents that can be used to repair them. These is another major risk. If all vehicle suppliers ad-
risks were identified and confirmed by the OEMs opted the system, and it covered all cars in current
involved in the project, and in some cases would production, the potential user community could
prevent them from being able to integrate their be as many as 2,000,000 people. Several portals
content with MYCAREVENT. would be required to cope with such a load. To keep
The key strength of using an ontology-based the systems synchronised it would be necessary
approach to service portal management is that it to adopt a time-controlled update system, with
reduces information overload on users, who other- updates being scheduled for early in the morning
wise would find it difficult to find their way through when system use is low. A separate system would
the maze of specifications and information types have to be assigned the task of receiving informa-
supplied by different manufacturers. By minimiz- tion from manufacturers and accumulating them
ing the set of options provided at each stage in ready to carry out a single daily update of online
the process, the MYCAREVENT advanced query portals. The downside of these process integra-
service makes it possible to identify information tion issues would be that any changes made to
resources provided by a range of manufacturers documentation, build specifications, etc, would
through a single reference point. not be available on the day they were recorded by
Until manufacturers are able to provide in- the manufacturer. This risk can be managed using
formation as to which faults can be solved using Trading Partner Agreements and Service Level
which documents a weakness of the service portal Agreements between the portal and the informa-
is that it will necessarily rely on users making the tion providers, following standard practices for
final choice between a range of information re- managing business relationships.
sources that cover a particular component. Where Expanding the proposed system to cover all
diagnostic tests are available their results can be vehicles would also require significant expan-
used to narrow down the range of possibilities. sion of system functionality, because commer-
When diagnostics are implemented an associated cial vehicles have a much wider range of build
problem is that of identifying the relationships specifications. One vehicle manufacturer reported
between symptoms, the conditions they can oc- to have 93,000 build specifications for trucks.
cur under and the diagnostic test results that can Part of the reason for this is that there are more

938
Applying Semantic Web Technologies to Car Repairs

distinguishing features, such as type of steering, program can look for only those pages that refer
number of axles, body type, couplings, etc, used to to a precise concept instead of all the ones using
define the build of a commercial vehicle. For such ambiguous keywords. More advanced applications
vehicles it becomes important to use the unique will use ontologies to relate the information on a
vehicle identification number (VIN) rather than page to the associated knowledge structures and
its vehicle registration number (VRN) to obtain inference rules.
accurate details of the build specification.

The real power of the Semantic Web will be realized


Future Research Directions when people create many programs that collect Web
content from diverse sources, process the informa-
As Tim Berners-Lee pointed out in his seminal pa- tion and exchange the results with other programs.
per on the Semantic Web (Berners-Lee, 2001): ... The Semantic Web, in naming every concept
simply by a URI, lets anyone express new concepts
Traditional knowledge-representation systems that they invent with minimal effort. Its unifying
typically have been centralized, requiring ev- logical language will enable these concepts to be
eryone to share exactly the same definition of progressively linked into a universal Web.
common concepts such as “parent” or “vehicle”.
But central control is stifling, and increasing the
size and scope of such a system rapidly becomes The need has increased for shared semantics and a
unmanageable. Web of data and information derived from it. One
major driver has been e-science. For example,
life sciences research demands the integration of
Two important technologies for developing the diverse and heterogeneous data sets that originate
Semantic Web are already in place: eXtensible from distinct communities of scientists in separate
Markup Language (XML) and the Resource De- subfields. … The need to understand systems across
scription Framework (RDF). XML lets everyone ranges of scale and distribution is evident every-
create their own tags—hidden labels such as where in science and presents a pressing require-
<author> or <title> that annotate Web pages or ment for data and information integration.
sections of text on a page. ... Meaning is expressed
by RDF, which encodes it in sets of triples, each The need to integrate data across a range of
triple being rather like the subject, verb and object distributed systems is by no means restricted
of an elementary sentence. These triples can be to the scientific community. It is a fundamental
written using XML tags. characteristic of any e-business scenario that needs
to be linked to back-office systems or to systems,
such as those used for payment management, run
An ontology is a document or file that formally by other companies. OWL allows the UML-based
defines the relations among terms. ... We can ex- modelling techniques that are fundamental to the
press a large number of relations among entities design and maintenance of back-office systems to
by assigning properties to classes and allowing be swiftly integrated with the XML-based mes-
sub-classes to inherit such properties. ... Ontologies saging approach that has been widely adopted for
can enhance the functioning of the Web in many inter-system communication.
ways. They can be used in a simple fashion to The trend towards globalisation that character-
improve the accuracy of Web searches—the search ises today’s business environment is established

939
Applying Semantic Web Technologies to Car Repairs

and set to continue. Increasingly, demands are terminology to model the domain in a way that
placed upon organisations to integrate information reflects how they understand it and speak about
from diverse sources and to deliver new products it. Furthermore, they can now encode knowledge
and value propositions in narrower timescales. and logic about the data structure, moving it out
To meet these demands IT organisations need to of application logic.
evolve towards loosely coupled systems where In this chapter we have said nothing about how
services can be assembled to support the execution OWL’s limited set of description logic (DL) rules
of business processes in a flexible way. A service can be used to constrain the values assigned to on-
oriented architecture (SOA) is not the only answer, tology properties or to infer membership of a class
though – for an SOA to be effective, a common from the presence or absence of property values.
view on to the data of the organisation needs to MYCAREVENT has not identified any points at
be available, so that data can be provided when which rules more complex than those needed to
and where needed to the processes consuming constrain cardinality or to ensure that all object
that data. properties are members of a given class or set of
RDF-based data integration has a lot to offer classes need to be applied to repair scenarios. But
because it provides a way to access information in many business scenarios more complex rules,
held in disparate systems without imposing a new including access control rules and permissions
structure on source data. If metadata is available, management, will be needed to ensure that busi-
or can be generated, a metadata-based approach ness constraints can be met. The presence of an
provides a framework structuring, processing and alternative, expert-system based approach to rule
querying information sources. definition and application within MYCAREVENT
The use of OWL and Semantic Web technolo- has meant that the service portal team has not
gies moves us beyond simple metadata to structures fully investigated the role that inferencing rules
where additional rules that determine the logical might play in the development of e-business ap-
meaning of the data can be layered on top of existing plications, though a number of possibilities have
data by the application of an ontology. Ontologies been identified, including ones related to digital
allows rules to be specified which can be reasoned rights management and skill-based access control
over using methodologies such as description to information resources.
logics, allowing inferred models of data to be Work began in 2006 on a W3C Rule Inter-
constructed. Not only is the explicit meaning of change Format (RIF), an attempt to support and
the data recorded, but the implicit meaning of the interoperate across a variety of rule-based formats.
data can also be inferred and exposed by applying RIF (see www.w3.org/2005/rules for details) will
such rules to data. eventually address the plethora of rule-based for-
The use of ontology-based approaches is an- malisms: Horn-clause logics, higher-order logics,
other step along the path from the computer to the production systems, and so on. Initially, however,
conceptual world used by humans. Now program- the Phase 1 rule semantics will be essentially Horn
ming has progressed from binary assembler lan- Logic, a well-studied sublanguage of first-order
guages to 4th generation object-oriented paradigms, logic which is the basis of logic programming.
ontologies allow knowledge to be encoded as data Among the deliverables scheduled from the RIF
structures in a way that reflects the understanding Working Group for the end of 2007 is:
and semantics of the domain and of human users.
Data can be modelled in a manner that is more A W3C Recommendation on using this rule inter-
intuitive and conceptually closer to the way humans change format in combination with OWL. This
think. Ontologies allow humans to use their own document is needed to help show implementers and

940
Applying Semantic Web Technologies to Car Repairs

advanced users how these technologies overlap Due to the EU Block Exemption Regulation,
and the advantages and limitations around using service providers have the right to access differ-
them together. This document must clearly state ent kinds of repair information, training material
which features of OWL can be mapped to (or oth- and tools.
erwise interoperate with) Phase 1 rules and which The MYCAREVENT project gathered part-
cannot, and software using this mapping must be ners from across Europe to establish a model of
demonstrated during interoperability testing. The excellence leveraging innovative applications and
document may also discuss rule language exten- state-of-the-art technologies, to offer a way for
sions to cover the excluded OWL features. making the market more transparent, competitive
and lucrative. It developed and implemented new
A second phase, scheduled to be completed in applications and services which could be seam-
2009, will extend rule processing to provide full lessly and securely accessed by mobile devices
first-order logic, negation, scoped negation-as- deploying Semantic Web technologies. These tools
failure and locally closed worlds. allow us to provide manufacturer-specific car re-
Until RIF tools are readily available, OWL pair information that matches problems identified
users will have to make use of proposals such as by Off/On-Board-Diagnostic systems.
that for a Semantic Web Rule Language (SWRL) Breakdown information is presented in dif-
(Horrocks, 2004) that extends OWL’s built-in set ferent languages. Mobile workers in different
of simple rule axioms to include Horn-like rules. countries can interact with service portals of
While MYCAREVENT has not currently identi- independent service suppliers as well as those of
fied any rules it needs to deploy within the AQS car manufacturers. Using the MYCAREVENT
which cannot be implemented using predefined Service Portal it becomes possible to provide a
SPARQL queries, it is anticipated that there will single point of access to information for any make
be other applications based on the GIIRM for of car, so ensuring that any car manufactured in
which more complex queries of the type provided Europe can be repaired in any European workshop
by SWRL may be needed. It will be interesting or by any European roadside assistance organisa-
to see, as RIF develops, whether the additional tion, irrespective of the preferred language of the
functionality offered by adopting Horn-clause or owner or the mechanic.
higher-order logics provides a simpler solution to
the type of reasoning currently being performed
by the expert system currently used to identify the References
causes of problems within MYCAREVENT.
Berners-Lee, T. (2001). The Semantic Web. Sci-
entific American, (May): 2001.
Conclusion Brickley, D., & Guha, R. (2004). RDF Vocabulary
Description Language 1.0: RDF Schema, W3C.
The automotive market has become one of the
most important and complex industries in the Clark & Parsia LLC. (2007). Pellet: The Open
EU, due to the rapid development and change Source OWL DL Reasoner.
in electronics, electrics, software and hardware.
Deliverable, M. Y. C. A. R. E. V. E. N. T. D3.2
Economically, it is a major contributor to the EC
(2005) Generic and Integrated Information Refer-
economy, accounting for circa 6% of total Euro-
ence Model. L.3.2_Generic_and_Integrated_In-
pean manufacturing employment and 7% of total
formation_Reference_Model_DT_v01.00.pdf
manufacturing output.

941
Applying Semantic Web Technologies to Car Repairs

Dickinson, I. (2005). HOWTO use Jena with an ISO/IEC 14662:2004. Information technology
external DIG reasoner. -- Open-edi reference model.
European Commission. (2002). EC Regulation Jena – A Semantic Web Framework for Java.
1400/2002; Application of Article 81(3) of the
Klyne, G., & Carroll, J. (2004). Resource Descrip-
Treaty to categories of vertical agreements and
tion Framework (RDF): Concepts and Abstract
concerted practices in the motor vehicle sector.
Syntax, W3C.
European Commission. (2004). The European Au-
McGuinness, D., & van Harmelen, F. (2004). OWL
tomotive Industry: Competitiveness, Challenges,
Web Ontology Language Overview, W3C.
and Future Strategies, European Competitiveness
Report 2004, Object Management Group. (2007) Unified Mod-
eling Language (UML).
European Commission. (2004). Proposal for a
Directive of the European Parliament and of the Prud’hommeaux, E., & Seaborne, A. (2007)
Council amending Directive 98/71/EC on the SPARQL Query Language for RDF, W3C.
Legal Protection of Designs. Extended Impact
Shadbolt, N., Berners-Lee, T., & Hall, W. (2006).
Assessment.
The Semantic Web Revisited. IEEE Intelligent
Horrocks, I., et al. (2004). SWRL: A Semantic Web Systems, 21, 96–101. doi:10.1109/MIS.2006.62
Rule Language Combining OWL and RuleML.
ISO/IEC 11179-5:2005. Information technology
-- Metadata registries (MDR) -- Part 5: Naming
and identification principles.

This work was previously published in Semantic Web for Business: Cases and Applications, edited by R. García, pp. 31-49,
copyright 2009 by Information Science Reference (an imprint of IGI Global).

942
943

Chapter 4.5
The Web Strategy Development
in the Automotive Sector1
Massimo Memmola
Catholic University, Italy

Alessandra Tzannis
Catholic University, Italy

Abstract a desk analysis focused on the strategic positioning


of the current businesses in the automotive sector
Especially in recent years, a transformation is ongo- (i.e., complexity evaluation of the presence on the
ing: the Web, besides being a means of information Internet, strategic architecture, quality, and effec-
sharing (internal-external), becomes a powerful tool tiveness of this presence).
for saving costs, reducing the distribution structure,
initiating distance transactions, and ever more, be-
comes a mechanism of integration with the external Introduction
environment and a catalyst of experiences for all
stakeholder. Starting from the identification of the I need a new car!
key elements, potentialities, and of the impact of Centy, as I affectionately call my old car, is gasp-
the Internet on firms’ performance, competitiveness, ing its last breaths. The “poor thing” has really had it,
effectiveness, and efficiency, this chapter is focused but then, it is really old; 10 years have passed from
on the changes in the automotive sector due to the that happy day it made its entry in our family.
integration between business strategy and Web However, I am not so sure I like the idea of
strategy. Therefore, starting from the consideration searching for a worthy successor. I am not an au-
of a clear identification and subsequent sharing tomobile fan, and my knowledge of them is rather
need of strategic goals, a research work will be limited. I am, however, sensitive to environmental
presented exploring, on the basis of an interpreta- problems, and I try my best to reduce the impact
tive model, the Internet potential in the automotive of my own person on the ecosystem as much as I
sector in order to achieve the identification of an can, considering its balance has already been seri-
optimal path definition and development of Web ously damaged.
strategy. This objective will be developed through What bothers me most is the idea of the long
search ahead of me. I do not have any particularly
DOI: 10.4018/978-1-60566-024-0.ch009 difficult requests, but I want to consider such an

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
The Web Strategy Development in the Automotive Sector

important purchase very carefully, avoiding hurried I am starting to question my beliefs. Some sites
decisions. I need a sensible family vehicle, with a only offer general information, deferring the sale
good balance between performance, consumption, to a moment of real interaction. There are some
and above all cost. I need to acquire information, sites though that are true virtual car dealerships,
but just the idea of going from one dealer to another and allow you to get through the whole buying
puts me in a bad mood. I can not stand having to procedure on their site, with even a home delivery
waste time on this, and above all, having to pollute service of the newly purchased car. So after all
for a car I do not even own yet! I would not spend too much time, as I did with
Also, interaction with the dealers is more Centy.
often than not rather uncomfortable. At times— However, I do not feel ready for an online
especially the more prestigious brand dealers— purchase. I have identified the model that seems
they treat you with a presumptuousness as if you right for me, but I need some extra information.
were a nuisance, a matter that needs to be settled Through the faithful search engine, I reach the Web
as quickly as possible. Other times, for the more site of the car manufacturer. I am welcomed into
sporty brands, they give you so many figures and a very sophisticated ambient. I click on “Product
acronyms, that they make you feel like “you did Range” and I find myself in a virtual car dealer
not do your homework” as in your school days. showroom.
Lastly, with some there is feeling of a lack of The models are well-presented, with clear
transparency, something like “tell ‘m everything 3D images. The technical features provided are
but not really the truth.” exhaustive for each model. I discover the “car
Of course I do not want to demonize all car configurator,” a very interesting tool that allows
dealers. These are only impressions, and on top you to configure your car by choosing the color,
of that, only personal ones. But I do believe that, the interior, the optional features, and, once the
at least once, we have all felt the uneasiness I final price has been determined, to have access
described above, either completely or in part. to another service that allows one to request per-
I am holding the yellow pages in my hands, sonalized financial options.
ready to copy my city’s car dealers addresses on It is also possible to print out a customized
a piece of paper, but then a question just comes brochure showing the specific car model with the
to mind, popping up almost unconsciously: “Why chosen color, interior, and optional extras. I find
not use the Internet?” out that the site offers me the possibility to not
I do most of my work nowadays through the only get to know my local dealer, but also book
Web: I keep in touch with my friends scattered a test-drive at my pleasure.
around the world, I keep my bank account, book And there is more! The site offers a whole se-
my holidays, buy music, and lately I have also ries of services that I would never have imagined
taken care of my physical well-being using the that go well beyond the usual general information
telemedicine services offered by my city’s health about the manufacturer. I realize that the contents
department. But then I say to myself: “But for a vary according to the phase of the purchasing
car it is different!” But is it really? Why would the process: when the customer is in the process of
Internet not have brought about the same changes choosing a car, the customer can benefit from tools
in life style, in the way of thinking, of buying, in that compare the models and optional extras, and
the automotive sector as it did in other areas? that allow people to create the car that is closest
I start my Web search by typing the word to their expectations. Afterwards, some gadgets
“car” into the search engine. Promptly a long list that make the car look closer and more real, can
appears with sites specialized in online car sales. be downloaded (pictures and videos). Once the

944
The Web Strategy Development in the Automotive Sector

purchase has been completed, the customer can automotive sector during our research work.
monitor the delivery process to see when it will We had to use our creativity simply because we
arrive at the customer’s door and after that, book imagined an ideal Web site that should hold all
the assistance and maintenance services, activating the solutions, information, and services that in
an SMS reminder service. reality the car manufacturing companies activate
Often the Web becomes an occasion to create a only partially in their portals.
community of customers of the same brand. Vari- It has rightly been observed that the Web
ous services have been created to this end, from the represents “a new space, a new territory, made of
normal forum to the more sophisticated blog, that computers, connections, software and above all,
allow an almost one-to-one interaction with the information…inhabited by individuals, compa-
manufacturer, offering the possibility to interact in nies, and by organizations; a place made of games,
various ways and degrees with the managers, and commerce and exchange” (Porter, 2001).
to know the reasons behind some project choices, Comparing the results of our research with
or to be informed about new models before they similar ones in other branches, such as banking
are launched on the market. (Frigerio, in this book), tourism (Yaobin, Zhaohua
I also find an online magazine that I can virtu- & Bin, 2007), retail business (Duke, Chul, Sang-II
ally read and that, very coincidentally, not only & Soung, 2006), and the health sector (Baraldi
contains car related articles, but “talks” about & Memmola, 2007), it seems that the automotive
environment, style, and sports. I realize I am un- sector is, generally speaking, running far behind
derstanding, really entering the brand’s life style. the others. The feeling that we get is that the
Once I own the car, through the Web site, I can companies belonging to this branch, even though
request a loyalty card that, apart from working as they have been developing a “public presence” on
a credit card on the main international circuits, the Web for quite a time, have hardly exploited its
allows me to obtain discounts not only for the potential, limiting themselves in various cases to a
car’s assistance (i.e., maintenance, repairs etc.), low profile Web strategy, structured as a source of
but also in many shops in line with the brand’s information rather than as a service. The Internet
style and features. has therefore played a role up to now that is not
At this point I am ready to leave the virtual dissimilar from any other media that allows a
world to get back to the real one. I can calmly company to develop a one-way communication
think about my visit to the dealer where I booked with the customers, and more generally, with the
my test drive. I have got all the information I need company’s stakeholders.
and I have made my choice. I could even make the This delay and lack of sensitivity toward “evo-
purchase online. I have seen that some sites offer luted usage forms” of the Internet’s potentialities
this possibility, but entering the dealer’s showroom can be partly justified by the strong emotional
will allow me to satisfy the emotional value of the value linked to the purchase of a car, leading
purchase, that is, to hold tight the steering wheel the customer to the dealer’s showroom to touch
of the long-desired vehicle, breathe in the “smell the product that the customer is going to buy, to
of new,” touch the dashboard. The Internet is not verify its qualities, thereby trusting tactile senses
able to satisfy these senses, at least not yet! to guide and reassure. And it must be said that
The lines above are obviously fantasy, but they past attempts to develop forms of e-commerce
are not science fiction, as they describe a situation in the automotive field, although they requested
that is very likely in the light of the information the purchase to be finalized at the dealer’s, did
and service contents that we have found on the not have much success.
Web analyzing the companies’ Web sites of the

945
The Web Strategy Development in the Automotive Sector

Over the past years however, this trend has Potentials of the Internet
been changing dramatically. According to a re- for the Automotive Sector
search carried out by McKinsey (2003), during
the next 10 years we will see the third revolution Why does the Internet continue to receive so much
in the automotive sector, after the creation of attention? Why are the Internet and its potenti-
the Ford mass production factory and the lean alities always being discussed? What makes the
production of the Toyota Production System. Internet a better technology than so many others
Customers of the automotive branch, according that preceded it? Does the Internet really change
to the research, will be expecting “hightest per- the way “to do business”? Does the Internet cre-
formance” for the same price, generating strong ate new business models or does it change the
pressure on cost reduction and innovation capac- existing ones?
ity in the automotive companies. These factors Quite a few years after the boom of the New
will, without a doubt, lead to a whole new set-up Economy and the subsequent enlivenment of the
of these companies’ supply chain. managerial literature at the turn of the century,
In fact, various companies are starting to grasp perhaps the question will receive more sensible
the idea that the Web can become an essential answers and above all, be sustained by a longer
component in the so-called “low cost car strategy” experience. The Internet surely is a surprising
to which many companies (e.g., Renault, Tata, technology through which it is possible to com-
General Motors, Fiat, Volkswagen, and Toyota) municate, interact, sell or distribute products and
are decidedly heading. Through the Web the services, and above all, create a powerful tool to
distribution network, which currently represents bind customers through communities in the form
an important part of a car’s final price, can be of forums, chat groups, blogs, user groups, and
removed or reduced to a minimum. so forth.
Not only! The Web becomes the privileged What mainly distinguishes the Internet from
instrument of interaction with the so-called “iPod previous technologies is that it sums up all their
generation,” who through the company’s portal features and all their potentials in one low-cost
or a specific product site, can contribute to the standard, that is, through the Web people can
various steps of the product’s project or image watch TV, listen to the radio, and talk over the
definition (e.g., accessories, internal and external telephone. Therefore it may make more sense to
graphic features, etc.). In this sense Fiat’s experi- talk about Media strategy rather than Web strat-
ence with its Web site www.500wantsyou.com egy: to underline the necessity, from a company’s
is surely significant. viewpoint, to govern the various interactive chan-
Finally the Web can play a vital role in support- nels, on and off the Web, in a harmonious and
ing customer relationship management (CRM) coordinated way.
policies, or in the broadening of the product The Internet, therefore, changes or can change
system with a series of complementary services the way to do business. It has rightly been observed
linked to the automotive world (e.g., assistance, that the brick and mortar companies, that is,
insurance, loans, etc.). which existed before or which were not founded
This chapter presents the results of a com- in function of the Internet, have to develop their
plex research aiming to define in what measure own business model in order to seize opportuni-
and with which modes the companies of the ties, taking into account the limits (but also the
automotive sector are using the potentialities strong points) of the current situation.
of the Internet to create value for their main The most important thing is not to succumb
stakeholders. to the charms of the Internet, which is only a

946
The Web Strategy Development in the Automotive Sector

tool after all, a dependent variable of the model The Internet as a Means
of analysis of the company’s governance. It is to Rationalize Time
therefore necessary to align the choices that have
been made within this scope with those which The Internet enables efficient time use, as a result
have determined the business strategy, selecting of its power to reduce or dilate it. The first distinc-
the Internet’s best potentialities that better fit the tive feature appears when looking for information
purpose, thereby carefully studying realization about a particular car, the Web offers the pos-
time and modes. sibility to find it quite quickly. In the same way
the Internet allows a time dilation realization if
The Internet as a Media and services, streamlining the transaction, are imple-
a Geographical Distance mented (e.g., booking of test-drive, of assistance
Reduction Tool services, etc.).

The Web facilitates exchange based relations, Web Externalities at the


overcoming the obstacles of time and space. Base of the Internet
More than 10 years since the diffusion of the
Internet, this may seem very obvious. According Technologies or products present Web externali-
to the authors, this is the main reason behind its ties if their value is proportional to the amount of
success though. The Internet is certainly not the users, that is, the more the users, the higher the
only media available: the alternatives are the press, value (Katz & Shapiro, 1986). In order to grasp the
television, and radio. As said above, the Internet meaning of these words, think of the usefulness
is extraordinary because it summarizes them all; of a telephone that can be used to talk to only one
at the same time developing interaction and the person. The value of such a phone would most
possibility to realize a communication process certainly be lower than that of a phone which has
that is not just one way. a potential for worldwide communication. Clearly,
As in other areas, also in the automotive sector the more people connected to the telephone sys-
this potential allows re-examination of the devel- tem, the higher its value will be for its users. The
opment logic of the supply chain (e-procurement), Internet represents, in this sense, clear proof of
the distribution processes (relations with the deal- this: the higher the number of people connected,
ers), and of the relationship with and assistance the higher is the value of the Net.
to the customer. Therefore, it does not matter if The authors have had a clear demonstration
the dealer is located far from the manufacturer; of this potential, participating at a discussion fo-
through the company’s Intranet2 the dealer will rum about a new small Italian car manufacturing
be able to see if a particular model is available company. The continuously increasing number of
or, alternatively, how long the production will participants and their interactions is progressively
take. A small size, niche manufacturer will be creating a pressure group aimed at demanding
able to offer cars for sale to potential customers the satisfaction of the customers’ wishes on the
all over the world, even if the manufacturer has manufacturer’s part.
no dealerships at all.
The Internet as a
Distribution Channel

The Internet, in general, operates as a distribu-


tion channel for all products with high informa-

947
The Web Strategy Development in the Automotive Sector

tion content that can be digitized (e.g., software, The Internet as a Transaction
music, videos, images, plane tickets, various Cost Reduction Tool
services related to banking, insurances, and even
healthcare). In the automotive sector too, the Internet could
In the automotive sector, the Internet could reduce the costs of the transaction process, as
be an interesting substitute for the existing dis- a direct consequence of the above-analyzed
tribution channels. Through the Web, low-cost elements. Often in fact, companies have to do
cars could be sold, making it necessary to reduce research work to find suppliers that offer the
the cost of the value creation channel as much raw materials needed for the requested quality
as possible. And there is more! The Web could standards; the customers have to acquire infor-
act as the main distribution channel for all those mation on the products’ features, on the prices,
companies that have difficulties with traditional and, in general, on the alternatives offered by the
approaches, for example, small size or niche market; the dealers have to acquire information
manufacturers or companies of a collateral sector on the customer’s financial situation, or interact
such as car tuning. with the headquarters to evaluate availability
and delivery terms of each model. In the above
The Internet as a Tool to Reduce activities, surely the majority of the information
Asymmetrical Information necessary to the transaction’s completion can be
easily obtained on the Web.
Information is asymmetrical when, in a transac-
tion, one party holds information that is relevant to The Internet as a Tool to Support
the same transaction that the counterparty does not Corporate Governance and
have. A classic example, printed in all economics Personnel Training Courses
books, is that of the transaction of a used car: here
the asymmetry lies in the fact that the buyer does Finally, the Internet’s technology is revealed as
not only ignore the real condition of the car, but an irreplaceable tool sustaining the following
often does not even have the means to know if the activities:
asking price is congruous with the market value,
or to evaluate the technical features. The Web • The company’s analysis, control, and gov-
offers such information, for example, the prices ernance (business intelligence systems,
asked by different dealers for similar models, or ERP systems that are ever more oriented
consumers’ associations that certify the quality towards Web-based applications).
of the product one intends to buy. • Distance training of operators, with huge
The reduction of these asymmetries can work cost savings and significant benefits re-
both ways. Through the Web a car manufacturer garding the service level.
can get to know about the expectations, tastes, and • Management of the company’s infor-
opinions of its customers. A very good example mation service and integration (through
is Fiat’s blog quellichebravo and it’s Web site Intranet and management networks) of the
www.500wantsyou.com, gathering information company’s divisions, organizational units,
essential for the development of new products. production sites, and dealerships.

948
The Web Strategy Development in the Automotive Sector

Objectives and Methodology company’s management can understand what


of the Research to use the web for, through the Web strategy it
should be able to understand how to use it. The
Objectives of the Research colonization of the virtual space offered by the
Web should then be realized through a progressive
This research work, carried out through a general alignment and a systematical evaluation (Mem-
survey of the Web sites of the main companies mola, 2007) of:
of the automotive sector, aimed at verifying how
and to what extent these companies are currently • The company’s main strategic
exulting the potentials offered by the Internet orientations.
to create value for the major stakeholders. In • The degree of acceptance and awareness of
particular, the aim was to collect data that would the Internet’s potentials on the part of the
allow us to evaluate: company’s internal and external stakehold-
ers (i.e., customers, personnel, suppliers,
• The strategic positioning—obviously on dealers, etc.).
the Internet—of the companies of the au- • The possible impact on the organizational
tomotive sector. structure and internal processing.
• What results have been achieved, or rather • The company’s general usage logic of other
how much of the potential offered by the information and communication technolo-
technology, is being used by the companies gies sustaining its processing.
to define their own public presence on the • The costs and benefits linked to the service
Web. or information contents activated one by
• If and to what degree the strategic initia- one on the Web.
tives are successful in terms of visibility • The performance measurement mecha-
and registered traffic. nisms enabling the evaluation of the proj-
ect’s success.
The Methodology of the Research
However, the creation of a strategy (business
R. M. Grant (1998) defines the main objective strategy or Web strategy) consistent with the
of a company’s strategy as the aim of “guiding internal and external environment in which the
the management’s decisions to excellent results company operates is not sufficient. Porter (2001)
through the search of a competitive advantage; it observes that it is essential to also “act upon” the
is simultaneously a means of communication as strategic positioning, that is, on the ability to do
of co-ordination inside companies.” business in a different way than the competitors,
In a correlated way, the Web strategy could in order to create value for the customers, that
be defined as that systemic set of decisions and is, not imitable. Which means using the Internet
actions aimed at determining in what measure and and ICT solutions in order to apply an univocal
through which modes the company’s positioning characterization of the product system to the im-
on the Internet can constitute an advantage in the age, the quality, and to complementary services
pursuit of the company’s strategy. rather than to the distribution logistics in such
It is evident that business and Web strategies a way that it will be impossible, or at least very
need to undergo an integrated process of defini- difficult, for the competitors to copy.
tion, development, and implementation. On the It is subsequently necessary to define the
other hand, if through the business strategy the strategic profile of the Internet presence, in order

949
The Web Strategy Development in the Automotive Sector

to be able to continue with the positioning of the present their organizational structure, mis-
single company compared to the direct competi- sion, commitment to the environment, and
tors, and to do the necessary categorizations and linked services (e.g., assistance, mainte-
possible benchmarking actions. nance) or complementary products (e.g.,
According to Vittori (2004) and Buttignon secure driving courses etc.).
(2001), the evaluation of a company’s approach • VCS: Offers the possibility to create a
to the Web is normally based on the technological space on the Web to build relationships and
dimension (i.e., show-case or static site, interactive exchange ideas and opinions; in this case,
site), the recipients of the communication process the aim of automotive companies is that
(i.e., B2B, B2C, suppliers, customers, etc.), or on of customer retention through the creation
the goal(s) of the site (i.e., e-commerce, customer of virtual communities such as a forum, a
service, branding, database creation, etc.). chat, or a blog where customers and fans
Such orientations are not consistent with the may exchange information or express their
objectives and the set-up that we have put into views on products, brand, services, and so
practice in our research work. In order to proceed forth.
with the evaluation of the strategic positioning of • VDS: The Internet offers a new, efficient,
the companies of the automotive sector, we have and inexpensive distribution channel, suit-
used the model created by Angehrn (1997) that able for a variety of services and products.
goes under the acronym of ICDT, which stands All companies offering goods or services
for information, communication, distribution, that can be digitized (e.g., e-books, videos,
transaction. This approach explains in a simple, music, etc.), but also companies wanting
but effective way, the sort of strategic approach to offer services to support their traditional
used by companies for the definition of their own products (e.g., online assistance, training,
Web site, evaluating how and to what degree they etc.) can benefit. Obviously, the automo-
are using the potentials offered by the Internet. tive companies belong to the latter catego-
The model’s name (Figure 1) derives from the ry and they can, for instance, offer the cus-
segmentation of the Internet’s virtual space into tomer the possibility to download their car
four main spaces: the virtual information space instructions manual, to receive reminder
(VIS), the virtual communication space (VCS), services for car maintenance, and so forth.
the virtual distribution space (VDS), and the vir- • VTS: The space allows one to realize re-
tual transaction space (VTS). This segmentation mote transactions with different interlocu-
highlights the fact that “the Internet has extended tors (e.g., customers, suppliers, personnel,
the traditional market space by providing new etc.); this Internet scope has traditionally
spaces in which economic agents can interact by been identified as e-commerce, related to
exchanging information, communicating, distrib- the online sale of goods or a series of con-
uting different types of products and services and tents which in any case “assists” the trans-
initiating formal business transaction” (Angehrn, action process (e.g., booking of test-drives,
1997). In particular: maintenance and repair services, job of-
fers, etc.).
• VIS: The space that companies use to in-
troduce themselves to customers or other With Angehrn’s model, a “map” of the Internet
stakeholders; in essence, it is a “one way,” presence can be drawn up, aiming at offering a
low-cost communication channel that the synthetic and prompt view of the Web strategy
companies of the automotive sector use to pursued by the company (Figure 2).

950
The Web Strategy Development in the Automotive Sector

Figure 1. The ICDT model (Angehrn, 1997). © 1997. Albert A. Angehrn. Used with permission

To this end the Web site is ideally divided into “occupation” area of the Net’s virtual space. As
a series of minimum units of analysis (MUAs). such, it does not necessarily coincide with the
A MUA represents an area inside the site charac- single Web page, but can be spread over more
terized by the homogeneousness of the contents pages and can share the same page with other
(informative or service) representing a precise MUAs. In the VIS area the MUAs which can

Figure 2. Creation of the map of the Internet presence

951
The Web Strategy Development in the Automotive Sector

be activated are mainly informative ones, for the fact that the stakeholders do not value
example, those informing about the company’s the Internet very highly.
history, mission, product range and, so forth. In • The high profile strategy: In this case (ty-
the VDS area instead, a MUA does normally pology Z in Figure 3) the Web strategy
have the task to give a content service, such as starts to actively support the pursuit of the
a reminder service for the car maintenance. Each institutional strategy. The company, al-
single MUA will then be placed within its virtual though it started exploiting the Web as a
correspondent area (i.e., VIS, VDS, VCS, VTS). mere low-cost communication tool, is now
The more external the MUA, the higher the degree starting to significantly broaden and qual-
of technological sophistication (LOS) and the ify its presence on the Web, “aggregating”
level of content customization (LOC) measured other types of virtual space (e.g., VTS,
along the orthogonal axes in the figure. VDS, VCS).
The Internet presence map allows us to • The very high profile strategy: In a con-
evaluate and compare how and to what degree text as the one represented by Company K
the automotive companies are using the Web in Figure 3, the top management consid-
to pursue their business strategy. It is therefore ers the Web strategy an essential part of
possible to work out a taxonomy of the strategic the business strategy and the stakeholders
approaches to the Internet that we have divided accept and are aware of the opportunities
into four main scenarios (Figure 3): offered by the Internet. The occupation of
the virtual space offered by the Internet is
• The low profile strategy: In this scenario homogeneous and characterized by high
(represented by typology X in Figure 3) levels of technological sophistication and
the company, although present on the Net, of content customization; the company
is (almost) unable to exploit its opportu- also “lives” through, and on the Web. An
nities. Probably, the company’s top man- essential part of the relationship with the
agement does not consider the Internet’s customers (or the other stakeholders) is
technology as a means through which the lived through this means of interaction.
company can pursue its business strate-
gies, or the company has not clearly de- The “colonization” of these virtual spaces can
fined what the Web should be used for, or obviously be developed according to different
even, it may simply represent the com- approaches and modes. Subsequently, the evalu-
pany’s answer to their stakeholders’ low ation of the strategic positioning on the Internet
degree of acceptance and awareness of of the automotive companies has been carried
the Internet. out through the consideration of four analytical
• The medium profile strategy: The scenario viewpoints:
(typology Y in Figure 3) defines an opera-
tive context in which the company seems • The complexity of the Internet presence,
only interested in exploiting a few oppor- or the intensity with which a company uses
tunities that the Web is able to offer. Very the Internet to sustain its institutional ac-
likely, in this situation the company’s top tivities. An intensive and extensive usage
management is still not fully aware of the of this technology will very likely be real-
opportunities that the Web can offer, or in ized through a Web site that has got “plenty
any case it does not want to make substan- of” content able to give information about
tial investments in this technology, due to the company (VIS), furnish goods/services

952
The Web Strategy Development in the Automotive Sector

Figure 3. The taxonomy of the strategical approaches to the Internet

(VDS), create virtual places of communi- The Complexity of the


cation (VCS) and carry out transactions Internet Presence
(VTS).
• The strategic architecture of the Internet The aim of this analytical perspective is to mea-
presence, or the more or less balanced sure the consistence of the Internet presence in
modes through which the company has the automotive sector, not so much in terms of the
continued with the colonization of the dif- Web site’s number of pages, but content richness
ferent virtual spaces made available by serving the purposes of:
the Internet, privileging some stakeholder
typologies (e.g., private customers, sup- • Providing information about the VIS area.
pliers, dealers, rental companies, business • Distributing services in the VDS area.
customers, etc.) and, perhaps, neglecting • Creating spaces for virtual communication
other typologies. in the VCS area.
• The quality of the Internet presence, • Realizing transactions in the VTS area.
measured by the LOS and by the LOC of
the contents proposed by the Web site. The higher the number of contents offered in
• The effectiveness of the Internet pres- the various areas, the higher the degree of com-
ence, in terms of visibility (based on the plexity that can be attributed to the site, which is
main search engines) and success (visitors’ measured through two indicators:
count) of the company’s Web site.
• The amount of MUAs counted in the site.
• The content usability index (CUI).

953
The Web Strategy Development in the Automotive Sector

Table 1. Methodology for analysing the Internet presence

Step 1: Internet presence complexity • examine Web site contents and classify them in MUAs
• assess the existence of other structural features (e.g., foreign language version, native
language search engine, Web site map, Extranet)
Step 2: Internet presence strategic • position retrieved MUAs in corresponding virtual space areas (VIS, VDS, VCS, VTS)
architecture • verify Web site use of Internet potential
Step 3: Internet presence quality • assign the LOS of each MUA
• assign the LOC of each MUA
Step 4: Internet presence effectiveness • verify Web site visibility by means of the most common search engines
• measure average Web site traffic

The concept of MUA has been explained above, customer clubs). Therefore, although the Extranet
whereas the CUI allows one to measure, in complex is not a tool aimed at improving the legibility of
terms, the existence of a series of “structural char- the site’s contents, it actually increases the degree
acteristics” that have been activated within the site of complexity of the Internet presence.
to improve content usability. It is infact logical to
suppose that when the amount of content increases, The Strategic Architecture of
for a Web site user looking for information, the the Internet Presence
degree of complexity increases accordingly; the
presence of an internal search engine or a site map Building a Web presence allows a company to
can be very helpful to this end. We also wanted to use virtual space, which areas (i.e., VIS, VDS,
consider the site’s usability by a particular category VCS, VTS) can be used in various ways and to
of stakeholders, that is, people with poor eyesight, various extents, depending on the company’s Web
verifying the possibility of easy modification of strategy.
character size, background color, or the existence of This analytical perspective allows one to high-
a voice describing the contents. The CUI is thus the light the Web strategy followed by the company in
outcome of the ratio between the number of struc- the “colonization” of such space in terms of:
tural characteristics deemed noteworthy according
to the methodology of the research, that is: • Number of occupied virtual spaces, in order
to evaluate the size of the presence built on
• Site map. the Internet.
• Search engine. • Exploitation degree of the Net’s potential in
• Site accessibility by poor sighted people global terms and of each typology of virtual
(character size and background modifica- space (VIS, VDS, VTS, VCS).
tion, etc.). • Stakeholder typologies that the company is
• Reserved area/Extranet. privileging (or, possibly, neglecting).
The indicators used to evaluate these aspects
An Extranet area can be defined as a company’s are respectively:
nonpublic presence on the Internet. Whereas the • The degree of “colonization,” resulting from
use of an Intranet is totally private, within the com- the simple enumeration of the Internet’s vir-
pany and by personnel only, an Extranet may give tual space areas occupied by the company
information and contents outside the company’s (VIS, VCS, VDS, VTS).
scope in a reserved way. In fact, customers can use
it to share information or services (the so-called

954
The Web Strategy Development in the Automotive Sector

• The coverage index, or the degree of cov- • The number of MUAs found in the site that
erage of the virtual space considered as a relate to a particular stakeholder category.
whole and of each single area. • The total number of MUAs that relate to
• The coverage index in respect to the dif- the same stakeholder category found dur-
ferent stakeholder categories, that is, the ing the research.
degree of coverage of the demand for in-
formation and services expressed by the The Quality of the Internet Presence
site’s users (i.e., customers, dealers, sup-
pliers, etc.), who for whatever reason are With regard to the qualitative dimension of the
interested in the company. Web contents, the objective of the research work
aims at using a measurement process that would
Defining the degree of colonization of the In- be able to take into account the technical aspects
ternet’s virtual space is a simple but effective way as well as the traceability of the contents within the
to measure the evolutionary phase of a company’s site, and that would respond as much as possible
Web strategy. However, since an area can also be to people’s informative or service needs.
“colonized” through only one MUA, it is clear Among the many alternatives, the drivers that
that this indicator furnishes a very approximate have been adopted by the research methodology
indication that needs to be integrated by the study to evaluate the Internet presence are the LOS
of the mean coverage and the single area coverage and the LOC, presented with their mean value
index. If the degree of colonization indicates the through the MUAs found in the sites of the ana-
company’s presence in one of the Internet’s virtual lyzed companies.
areas, through the coverage index it is possible The evaluation of the level of sophistication
to assess how much the company is benefitting, (in the broadest sense of the word) is linked to
through its site, from the opportunities offered by the technology used. A progressive variable scale
the Net in that specific area. from 1 to 15 has been used to represent the various
In particular, the coverage index for each area intermediate situations that can occur through:
(VIS, VDS, VCS, VTS) is fixed according to the
ratio between: • A MUA based on static pages realized
with programs that were not specifically
• The number of MUAs found in the com- ideated for the Internet (e.g., MS Word,
pany’s site for each area. Powerpoint, etc.).
• The total number of MUAs found in the • A MUA based on real pages with the most
same area during the research. modern tools of multimedial presentation
(e.g., Flash View animations, audios, vid-
The average coverage index is subsequently eos, images) and with a high degree of
calculated through the mean average of the val- interactivity.
ues calculated in each single area and supplies a
synthetic, though approximate measure, of how The level of customization is also based on
much the company is exploiting the advantages the possibility to “cut” the offered site contents,
offered by the Internet in general terms. according to the user’s specific customization
Finally, the coverage index regarding the dif- or service needs. In this case too, a progressive
ferent stakeholder categories is the outcome of evaluation scale from 1 to 5 has been used. These
the relation between: values represent the extremes of a continuum of
intermediate solutions that can turn up between:

955
The Web Strategy Development in the Automotive Sector

• A MUA based on static pages, without any • A traceability index resulting from the
possibility of customizing the content, major Italian and international search
and without any additional services. engines.
• A MUA based on pages that allow the • A ranking of the Web site’s registered traf-
creation of dynamic, customized contents fic supplied by “Alexa.com,” which draws
with accessory services, for example, to up an international “hit parade” based on
build your own customized home page. the registered traffic.

The Effectiveness of the The first indicator tells how easy it is to “reach”
Internet Presence the Web site through the major search engines (e.g.,
Yahoo.com, Google.com, Msn.com, Altavista.
We wanted to complete the evaluation process com, Virgilio.com, Arianna.com). As keywords,
of the Internet presence of the companies of the we have used, with an incremental approach, the
automotive sector through an analysis that would company’s name, the name of a product of the
express the degree of “success” obtained by the company, and lastly, a series of other informa-
Web sites. This intention implies, nevertheless, tion (i.e., the dealers and the promotions related
the solution of two methodological matters: to the company, a product’s accessories line).
The index is equal to the sum of the mean of the
• The first concerns the need to translate in positive results (search engines with positive
actual terms the very concept of “a Web outcome compared to the total number of search
site’s success.” engines) obtained for each level of measurement
• The second is closely linked to the first (i.e., company, product, and other information).
and concerns the finding of variables able A result is positive when the search engine has
to give a measurement as objectively as included the searched information in the first
possible. seven proposals (normally visualized on the first
page of the search results).
There are basically two sorts of goals a com- Alexa.com is a site that makes a classification
pany can have when setting up a Web site: of all Web sites worldwide, giving them points
according to the number of registered visitors.
• Maximizing its visibility on the Web and Alexa.com uses an evaluation model that takes into
therefore be reached by the highest num- account both the number of accesses to the site,
ber of potential users. and the number of pages visited within the site.
• Maximizing the use of the Web site by all The number of accesses quantifies the number of
the stakeholders. visitors that have accessed the site on a particular
day. The number of pages visited is calculated by
The higher the degree of the above goal the amount of internal URLs requested by the
accomplishment, the more effective (and suc- same visitor. Obviously the repeated request of
cessful) the Internet presence will be, thereby the same URL, on the same day, is counted as a
also solving the second critical point, that is, the single page. The site with the highest combina-
effectiveness level measurement of a company’s tion of accesses and visited pages will be given
Internet presence. The measurement has been a value equal to 1.
obtained through the combined consideration The evaluations furnished by Alexa.com relate
of: to all the active sites on the Web. Therefore, a
data normalization process had to be carried out

956
The Web Strategy Development in the Automotive Sector

in order to obtain a classification specifically The research sample examines 56 automotive


related to the analyzed companies. companies that roughly fall into two macro catego-
ries: generalist companies and niche companies.
The first category includes those automotive
Main Findings companies with differentiated offer (in terms
of price level and type/model of car and service
The Research Reference features), multiple targets, and different strategic
areas of market covered; while the second cat-
This project was developed to focus on the Ital- egory includes automotive companies that have
ian customer point of view. We took into account a strong brand and image, often part of a group,
all the car-maker operating in Italy at the time of offer a limited range of luxury cars and focus on
the research3 from the databases of the National a segment, target type, or special use. These tend
Association of Italian Automobile Manufacturers to present a higher level price range with many
(ANFIA) and National Union Foreign Car Dealers customized options (even if the variable is not
(UNRAE). The study took into consideration the discriminatory) (see Table 2).
Italian version of the Web site, or when this was For each dimension of analysis (e.g., complex-
not available, original language Web sites. This ity, strategic architecture, quality, effectiveness) a
choice was made following two criteria: positive and negative benchmark was created, de-
fining the average value of 5% of companies with
• Automotive companies, except for a few the best and worst performances, respectively.
cases, are global in nature. They develop
a different approach in terms of products Analysis of the Internet
and services offered, reflecting the specific Presence Complexity
characteristics and the importance of the
market, depending on each geographical The complexity of the Internet presence was
area (state or continent). We decided, there- evaluated through the combined consideration
fore, to evaluate their strategic positioning of the following indicators:
in a specific market area: Italy, in our case.
• The international sites of larger sized com- • Number of MUA activated within the Web
panies (singled out by domain name_com- site.
pany.com) are simply “entry systems” • CUI, calculated on presence or absence,
through which they are readdressed to the within the site, of fixed structural charac-
site of the country or language (singled out, teristics such as: a) site map, b) internal
for example, by domain name_company.it search engine, c) accessibility to site for
or name_company.uk). poor sighted people, and d) reserved area
(Extranet).
Obviously, this decision may have an impact
on the results of the research, which are in fact The research identifies a total of 56 MUAs
influenced by the strategic choice which the variously distributed in the different areas where
company decides to follow in that specific mar- it is possible to articulate the virtual space of
ket. Nevertheless, research does not imply any Internet (VIS, VDS, VCS, VTS). Three distinct
change or adjustment to the framework which clusters (see Table 3) stand out from the Web sites
remains the same, regardless of the geographic of theAnalyzed companies:
area considered.

957
The Web Strategy Development in the Automotive Sector

Table 2. Macro categories of reference of the analyzed companies sample

Generalist Manufacurers Niche Manufacturers


• Alfa Romeo • Abarth
• Audi • Aston Martin
• BMW • Bentley
• Cadillac • Corvette
• Chevrolet • Daimler
• Chrysler • Dr
• Citroen • Ferrari
• Dacia • Hummer
• Daihatsu • Lada
• Dodge • Lamborghini
• Fiat • Lotus
• Ford • Maserati
• Honda • Mayback
• Hyundai • Mini
• Jaguar • Pagani
• Jeep • Porsche
• Kia • Rolls-Royce
• Lancia • Santana
• Land Rover • Smart
• Lexus
• Mazda
• Mercedes
• Mitsubishi
• Nissan
• Opel
• Peugeot
• Renault
• Saab
• Seat
• Skoda
• Ssangyong
• Subaru
• Suzuki
• Tata
• Toyota
• Volkswagen
• Volvo

• Cluster A: Companies, using the Web, with one that activates just a few. A company’s Internet
a high number of informative or service strategy is nothing more than a “translation” of
contents, with over 30 activated MUAs. its business strategy on the Web. This explains
• Cluster B: Companies with an average how the activation of some MUAs, for example,
profile approach and between 25 and 30 those supplying informative content or provide
activated MUAs a booking service for their company museum,
• Cluster C: Companies with a low profile is typical of companies that focus on tradition
approach and less than 25 MUAs. and historical authenticity of their brands as an
added value element (e.g., Alfa Romeo, Ferrari,
The simple enumeration of MUAs must not Lamborghini, Maserati, etc.)
be misleading. We are not looking at performance From the above considerations, it results that
criteria and stating that a company that activates a the number of activated MUAs is not a satisfac-
higher number of MUAs inside its site is better than tory indicator of the complexity of the Internet

958
The Web Strategy Development in the Automotive Sector

Table 3. Cluster of 56 analyzed companies

Companies with less than 25 MUA Companies with between 25 and 30 MUA Companies with over 30 MUA
Daimler Honda Citroen
Lada Suzuki Ford
Dr Bmw Jaguar
Abarth Hyundai Mazda
Pagani Kia Saab
Rolls-Royce Seat Volkswagen
Mayback Chevrolet Volvo
Santana Ferrari Lotus
Jeep Smart Porsche
Chrysler Subaru Fiat
Dodge Audi Skoda
Lamborghini Cadillac Lancia
Tata Nissan Alfa Romeo
Dacia Land Rover Mercedes
Hummer Mini Renault
Daihatsu Mitsubishi Maserati
Toyota Opel
Bentley Peugeot
Ssangyong
Lexus
Aston Martin
Corvette

presence, which gives way to the introduction the greatest number of activated MUAs with a
of the second indicator, the CUI as previously poor usability level due to the absence of basic
defined. navigational instruments (i.e., site map, search
Overlapping the two parameters used to evalu- engine, accessibility for poor sighted people) on
ate the complexity of the Internet presence, certain which the CUI index was developed. It is usually
inconsistencies come to light: one would expect a the general manufacturers who offer the greatest
high number of CUIs to result from a high number number of contents, compared with the niche
of MUAs, due to greater usability of the numerous manufacturers. Lotus is an exception, which even
contents offered. In fact, in several situations this if it belongs to the second category, shows up in
hypothesis is denied by sites which, faced with a the higher part of the second quadrant.
rather large number of activated MUAs, present a
quite low CUI index (Alfa Romeo, Volkswagen, Analysis of the Strategic
Volvo, Saab). Architecture and Internet Presence
The good positioning of Maserati, Renault,
Mercedes, and Lancia stands out. Alfa Romeo’s This perspective of analysis recognizes the strate-
situation is worth mentioning since it presents gic profile of the Internet presence of companies

959
The Web Strategy Development in the Automotive Sector

Figure 4. The complexity of the Internet presence

more in the automotive sector than in other areas Nearly all of the analyzed companies, however,
of research. A company pursues a high profile adopt an Internet strategy which concentrates
Web strategy the moment it achieves a presence on the information space (VIS), in which the
that is: contents are expressed in information regarding
the company structure, the products and services
• “Widespread” in each of the areas in which offered by the company, the prices of the products
the use of the Internet virtual space has been or services, contacts, and so forth.
developed (degree of colonization). There are five distinct clusters of indexing
• Quantitatively important in terms of num- (Figure 5):
ber of MUAs activated compared to those
potentially activeable (coverage index). • Cluster A: Companies with a Web strat-
egy of medium/high profile that develop
The results of the research demonstrate a a widespread presence in all the areas
generally elevated degree of colonization (aver- and with an over average coverage in-
age value 3.57) as a consequence of the fact that dex. Companies nearest to the positive
the automotive companies have predisposed Web benchmark like Alfa Romeo, Maserati,
sites with at least one MUA in each virtual space. Fiat, Volkswagen, Renault, Lotus, but also
Nevertheless, the potential of the Internet is still Cadillac, Jaguar, Smart, Land Rover, and
only used to a limited degree, that is, less than half Audi, who show lower degrees of exploi-
of its potential. The average value of the coverage tation of the Internet opportunities, appear
index is just 37%. The fact that many companies in this cluster.
still only have a weak presence in the areas of • Cluster B: Companies with a medium pro-
greater value content such as VDS, VCS, and VTS file Web strategy which develop a wide-
(the coverage indices specific to these areas are spread presence in all the areas and with a
generally very low) contributes to this result. coverage index value in line with or lower

960
The Web Strategy Development in the Automotive Sector

Figure 5. The strategic architecture of Internet presence

than average, for example, Honda, Seat, evaluation of informative or service contents acti-
Chevrolet, Corvette, Hummer, Abarth, vated for each individual stakeholder category.
and so forth. Figure 6 shows the performance of the av-
• Cluster C: Companies with a Web strat- erage coverage index of the automotive sector
egy of average profile which develop a with regard to the Italian market compared with
presence limited to three areas, but with the informative or service contents activated for
an average coverage index value (e.g., each stakeholder; it is possible to develop the
Ford, Saab, Opel, Nissan, Aston Martin, following considerations:
and Toyota).
• Cluster D: Companies with an average/ • It is confirmed the trend highlighted in the
low Web strategy profile which develop a previous lines underlining that the compa-
presence limited to three areas, but with a nies in this sector are not yet able to fully
consistently lower than average coverage exploit the potential of the Internet. On av-
index value (e.g., Chrysler, Jeep, Pagani, erage, the index of coverage for each cat-
DR, and Lada). egory of stakeholder varies its range from
• Cluster E: Companies with a low profile a low of 41% to a maximum of 54%.
Web strategy which develop a presence • Certain categories of stakeholders (i.e.,
limited to only two areas and with a lower suppliers, business customers, rental com-
than average coverage index value (e.g., panies, public companies, new employees)
Tata, Santana, Rolls Royce, and Daimler are more privileged than others (e.g., part-
Jaguar). ners, employees, protected categories).
• The relatively low coverage for private
Another evaluation parameter of strategic customers shows how the potential of the
architecture of the Internet presence concerns the Internet, compared with those who may
be classified as the institutional recipients

961
The Web Strategy Development in the Automotive Sector

Table 4. Stakeholder considered in the research

Stakeholder Description
Providers Companies providing semifinished goods (e.g.m alloy wheels, plastic wheel cover, piston, connecting-rod, etc)
or providing services (financial companies, consulting, etc.) for the analyzed company.
Partner Companies that, participating in the design or development of the product, may be considered a strategic partner
of the analyzed company (e.g., Microsoft developing in collaboration with Fiat the Blue & Me service or Brembo,
a world leader in braking, involved with a decisive role in the design of a new model).
New Employee Those who are interested in starting a working relationship with the company analyzed (e.g., work with us...).
Employee Those who have an employment relationship with the company analyzed.
Dealers Companies that are holders of a concession of sales of the company analyzed (includes the car dealers and the
so called “authorized” and “retailers”).
Car Rental Companies The car rental societies.
Private Customers Actual customers or prospects of the analyzed company.
Business Customers This category is different from the previous one because it does not refers to private customers in a B2C logic,
but to companies in a B2B logic. Expectations at a content or service level are obviously different from the
previous stakeholder identified.
Persons With Disability Those who are actual or prospect customers of the analyzed company, carrying a form of disability.
Public Administration The public corporations (e.g., state, region, province, municipality) which may be interested in the territorial
development policy implemented by the analyzed company.

of the Web site remains high and largely Analysis of the Quality of
unexplored; this result is the consequence of the Internet Presence
a large number of MUAs related to this type
of stakeholders (see Figure 7) a still relatively The quality of the Internet presence is intended
modest coverage especially in more “ad- as the level of technological sophistication
vanced” areas such as the VDS, the VCS, and of the distributed contents and the possi-
the VTS. bility to enter in a personalized or assisted
way. The following parameters of evaluation
have been used to develop such a perspective
analysis:

Figure 6. The stakeholder coverage index

962
The Web Strategy Development in the Automotive Sector

Figure 7. MUAs per stakeholder

• Level of technological sophistication performances that are not far or around the
(LOS) average measured value.
• Level of customization (LOC) • Cluster C: A final group of “latecomers”
• Measured on a scale from 1 to 5. who present a considerable weakness in
respect to this dimension of analysis, con-
As a first approximation, Figure 7 shows the sisting of those companies that have values
indexing of automotive companies regarding the close to the negative benchmark, such as
qualitative dimension of their presence on the Aston Martin, Santana, Tata, Lada, and so
Internet. Although in this case the significant forth.
differences noted in other analysis perspectives
were not evident. Three distinct clusters however A further evaluation takes into consideration
arise: the balance between LOS and speed of access to
the contents in each individual Web site. Most of
• Cluster A: The first group basically shows the sites (41%) present a “strong preference” for
the presence of companies like Mini, LOC rather than LOS (with higher values of the
Ferrari, Maserati, and Alfa Romeo and as- first index than the second one). So the companies
sumes a “pioneering” position regarding tend to favor highly customized contents over
the qualitative characterization of their technology, which ends up being a little sacrificed.
presence on the Net. The more sophisticat- Only 11 of the 56 analyzed companies, in fact,
ed the technology, the higher the possibil- consider the level of sophistication more important
ity to allow customization of the Web site than the level of content customization.
content; this is far superior compared to the A final consideration concerns the alignment
average value of the two parameters. between business strategy and Internet strategy.
• Cluster B: The second group in the posi- Companies with greater evocative power, with a
tion of “chasers” is particularly notewor- middle to high position in terms of market level or
thy considering it combines the majority of which target the young or a niche, tend to present
the companies present on the market with sites that focus on aspects of emotional involve-
ment, full of images or interactive contents. The

963
The Web Strategy Development in the Automotive Sector

aim, in this case, is to create a bond with the brand differences in terms of traffic registered do
rather than the mere gratification of informative not exist in the surveyed Web sites.
demands. • The consistent distance which exists be-
In this sense, it can be noted: how Mini stands tween the average reference values and the
out thanks to a technologically advanced site companies which present a performance
and to highly customized contents; how Fiat has close to the negative benchmark (Figure
a good position, thanks, above all, to new ideas 7). The companies in question (i.e., DR,
introduced for the launching of the new Fiat 500; Hummer, Ssangyong) are relatively new
how Ferrari has one of the richest sites regarding on the Italian market, or have an extremely
service contents and high path diversification; and focused offer. In our view, these could be
how Abarth presents an extremely distinguished the main reasons for such different results.
image from a technological point of view, offering
the possibility to customize the site structure, its Focusing the attention on the companies pres-
aspect, colors, and music, almost achieving the ent in the quadrant II, it is possible to note the
type of tuning which is at the basis of this brand’s existence of four clusters:
philosophy.
• Cluster A: Companies that have an ef-
Analysis of the Effectiveness fective presence on the Web with a high
of the Internet Presence level of visibility and traffic registered.
Companies like Fiat, Nissan, Volvo, Alfa
The measure of success of the strategic choices by Romeo, Ford, Citroen, Mercedes, and so
companies in the automotive sector regarding Web forth, which are closest to the positive
positioning was achieved through the combined benchmark, are included in this first area.
consideration of the following parameters: • Cluster B: Companies that record high
levels of traffic, but show a lower visibility
• A tracing index of the site through a panel compared with the search engines we used
of search engines. (e.g., BMW, Mazda, Audi, Suzuki, Subaru,
• A ranking, created by Alexa.com, which Mini, etc.).
supplies an indication of the traffic regis- • Cluster C: Companies which present
tered by the Web site. high visibility levels, but at the same time
achieve a relatively worse performance in
Generally speaking, the Web sites of automo- terms of recorded traffic (e.g., Volkswagen
tive companies enjoy good visibility: 63% of them or Jaguar) who therefore present atypical
presents a high traceability index (more than 2), behavior.
23% presents average visibility (with an index • Cluster D: “Follower” companies with
between 1.60 and 2), and only 8 out of the 56 visibility levels and recorded traffic low-
companies presented a decidedly low visibility. er than the average (e.g., Jeep, Tata, and
Concerning registered average traffic, the Chrysler).
results of the research offer two considerations:
The previous arguments refer to the dynamics
• A substantial aggregation of performance of competitive interaction among companies in the
values corresponding to the positive automotive sector specific to the Italian market.
benchmark; with the exception of a few As in every other research area, also with regard
companies (e.g., Fiat, Nissan, Volvo) great to this perspective of analysis, the reference was

964
The Web Strategy Development in the Automotive Sector

Figure 8. The quality of the Internet presence

the Italian language Web sites of Italian automo- rather their brand names, have on the interna-
tive companies, whose contents, in certain cases, tional scene; Toyota, Mercedes, Chevrolet, Ford,
are not managed directly, but by a branch or the and Volkswagen are the players who achieve
Italian importer. It is evident that the results ob- the highest performance levels. Italian brands
tained in terms of visibility, but above all recorded undergo a decisive reduction, with the exception
traffic inside the Web site, are influenced by the of Ferrari which, although it is a niche producer,
competitive positioning that the company achieves has an international standing. Lancia’s case is
in the real world. In other words, the particularly noteworthy; the brand name is present on the
high performance rate of Fiat, Alfa Romeo, and Italian market, but not particularly diffused on the
Lancia is also a consequence of the fact that the international scene, which in this new evaluation
study deals with well-known national brands with presents performance levels that are a lot lower
high values of brand awareness. On the contrary, than the average.
the results of Chrysler or Jeep can be fit into the
logic of a minor diffusion of these brands, having
almost a niche presence on the Italian market. Limitations and Further
We have therefore tried to repeat the evalua- Research Areas
tion of this perspective of analysis using original
language sites of the different companies (singled In order to present a complete view of the phe-
out by the extension .com or from the extension of nomenon, the methodology used to analyze the
the country of origin as, for example, Lotus [www. presence of automotive companies on the Web
lotuscars.co.uk]). The results shown in Figure 8 inevitably experiences a few structural limits
are, evidently, more consistent compared to the which may be clarified as follows:
impact that different automotive companies, or

965
The Web Strategy Development in the Automotive Sector

Figure 9. The effectiveness of the Internet presence

• The Internet is an ongoing reality. The re- The research presented in this chapter is the
search project photographs the situation as first phase of a complex project, still being car-
it was at the time of the analysis of the Web ried out, which also includes the development
sites, which took place in the months of of an in-depth analysis through a series of case
March and April 2007. histories which help to fully explain the possible
• The aim of this research is to provide an definition paths of the Web strategy of an auto-
objective evaluation of the presence on motive company.
the Internet. Even though a precise scale The intent is to move from an analysis which
of evaluation of the LOS and of the pos- takes place “outside” of the company, which is
sibility to proceed to a customization of the resolved in the evaluation of the “finished prod-
contents offered by the Web site (LOC) had uct” of the Web strategy, through the site to an
been defined, there are inevitable elements analysis “within” the company. The aim is to have
of subjectivity present in this phase which a complete vision of the phenomenon object of the
cannot be eliminated. survey, studying the path of preparation, definition,
• The research examines the Italian user’s and successive implementation of the Internet
point of view. We have already discussed strategy, the critical aspects and the duration of
the reasons for this choice which, however, development, the impact on the organization, and
do not prejudice the possibility of applying the expectations and requirements of the persons
this methodology to any other geographical either directly or indirectly involved.
context. It was in our interest, furthermore,
to evaluate the dynamics of the competitive
positioning of the companies in the automo- Conclusion
tive sector in a specific geographical area.
This research would not have had much sense 15
years ago. More to the point, it would have been

966
The Web Strategy Development in the Automotive Sector

Figure 10. The effectiveness of the Internet presence (focus on 2nd quadrant)

Figure 11. The effectiveness of the Internet presence (international positioning)

967
The Web Strategy Development in the Automotive Sector

impossible to carry out. The raw materials on which tive sort and low technological level. Although it
our study is based would have been missing, or, cannot be said that automotive company sites are
anyway, the object phenomenon of the survey would merely digital transpositions of the brochures or
have had totally different levels and dimensions. product catalogues, wider areas of improvement
It is in this simple and obvious reflection, but could be made, both in terms of services offered
absolutely corresponding to the truth, that the and greater personalization of the proposed naviga-
extraordinariness of Internet must be found. It is tion routes. In general terms, it can be stated that
almost banal to say that Internet has overturned the focus is still basically on the product and on
the logics and paradigms of our daily life. This the image associated with the product.
same book has been written through the Internet! A further consideration examines the ap-
Its authors have interacted virtually over the Net, proach to the Net, which does not always seem
without ever having met physically. to occur strategically, but often with a logic tied
Inevitably, Internet also changes the company to the improvisation or an image of the Internet
logics and paradigms. Thanks to (or because of) as a mere communication tool. In several cases,
this technology, the way to run a company has the impression is that the Web site “frees” itself
changed. The Web “opens” new opportunities. from the logics of strategic positioning that the
And again, not even these are new concepts. Very company follows in the real world in order to
concisely it is possible to affirm that through a follow merely commercial or promotional logics,
redesigning of the company proceedings, the on one hand, or to present the product and the
Net paves the way for improved performance in company sometimes too quickly and superficially.
terms of lower costs, higher quality, and greater Companies rarely demonstrate awareness that the
capacity of innovation. The Internet is a remark- investment in this technology must come through
able instrument for interaction with the company an appropriate alignment with the institutional
stakeholders, while other media provide only strategy of the company (Minard, 2001), with a
one-way channels of communication. Through correct review of the main operative mechanisms
the Net it is possible to improve logistics and in- and company governance.
teraction. Parallel considerations can be made for If we were to retrace the results of our analy-
other phases of the company’s value chain, such sis to different steps of the purchasing process,
as the distribution and the relationships with the it can be stated that the initial stages of seeking
car dealers in the case of automotive companies. basic information and presentation of product
Finally, the Web could become the preferred in- alternatives are definitely well covered, with often
strument for interaction with customers (not just complete contents and of a multimedia sort. The
the private ones, but also business-to-business), phase of the actual transaction, and above all,
in order to fully understand their expectations the one linked to the management practices of
and requirements, gain their loyalty, and even post sales services (e.g., assistance, maintenance,
involve them in the design and launching of new etc.) tends to be poorly managed, with a few ex-
products (as Fiat did with the site 500wantsyou. ceptions. The “pioneeristic” practices allow the
com concerning the new Fiat 500). online management of accessory services such as
The feeling which emerges from the results of financing, purchase or ordering, and possible pay-
this research is that companies in the automotive ment for maintenance/repairs, or the activation of
sector are not fully exploiting the opportunities. a reminder service which advises the client when
Frequently, a limited use of the Web is wit- to proceed with maintenance.
nessed, focused on the less complex areas, such as The contents allowing companies to gain cus-
the allocation of contents usually of the informa- tomer loyalty both through playful components

968
The Web Strategy Development in the Automotive Sector

(e.g., screen savers, images, logos, wallpaper but how we can achieve them on the Web. By doing
also real videogames) and through mechanisms of this, the corporate strategy of our company will
online community (e.g., forums, chats, blogs and be realized.
also personalized credit cards) are beginning to
catch on. At times contents are increased without
adequate attention to their usability, causing confu- References
sion and a frequent absence of a profiling logic of
the user, of the user’s interests and needs. AA. VV. (2002). Stanford-Makovsky Web cred-
The road towards a mature use of the Web by ibility study 2002: Investigating what makes Web
the automobile companies is still a long one, at sites credible today. Retrieved June 23, 2008, from
least judging from the results we have gleaned www.webcredibility.org
from our research. The question, or rather, the Alemi, F. (2000, Fall). Management matters:
questions that must be asked at this point are: What Technology succeeds when management inno-
is it worth? Is it worth investing in the Web? How vates. Frontiers of Health Services Management,
can the Web, through either its more traditional 17(1), 17–30.
ways or from its more evolved declinations (like
the various blogs, Wikipedia, YouTube, and Fa- Angehrn, A. (1997, August). Designing mature
cebook) help to accomplish the corporate strategy Internet business Strategies: The ICDT model.
of a company? European Management Journal, 15(4), 361–369.
These are important questions, we realize, that doi:10.1016/S0263-2373(97)00016-9
could generate a specific trend of research. As far
Angehrn, A., & Meyer, J. F. (1997, Summer). De-
as we are concerned, having studied the Internet
veloping mature internet strategies (cover story).
phenomenon for several years, we have become
Information Systems Management, 14(3), 37–43.
more aware of the fact that it is not a question of
doi:10.1080/10580539708907058
how much to invest, but rather how to invest. The
success and choice effectiveness of the strategic Baraldi, S., & Memmola, M. (2006). How health-
positioning on the Web are influenced by the care organisations actually use the internet’s virtual
company’s brand, but they also depend heavily on space: A field study. International Journal of
the grade of originality of the informative contents Healthcare Technology and Management, 7(3/4),
or proposed services. 187–207.
We would like to conclude with a sentence
Buttignon, F. (2001). Strategia e valore nella net-
from one of our many “specialists” we met during
economy. Milan, Italy: Il Sole 24 Ore.
the planning phases of our research work: “The
point is that on the Internet we are all guests... and Duke, H. C., & Chul, M. K., Sang-II, K., & Soung,
as guests, we have to bring something!” What is H. K. (2006, December). Customer loyalty and
required, therefore, is to gain the attention, “bring- disloyalty in Internet retail stores: Its anteced-
ing something which is not normally present in ents and its effect on customer price sensitivity.
the home of our host, or rather, on the Net,” that International Journal of Management, 23(4),
is, contents that are new and original. But, in 925–942.
order to avoid that, this becomes an exercise for
Elango, B. (2000). Do you have an Internet
its own sake, it is necessary “to understand what
strategy? Information Strategy: The Executive’s
type of invitation we have received and how we
Journal, 16(3), 32–38.
intend to behave accordingly.” We must know
what the expectations of our stakeholders are and

969
The Web Strategy Development in the Automotive Sector

Glaser, J. P. (2002). The strategic application of Porter, M. E. (2001, March). Strategy and the In-
information technology in health care organiza- ternet. Harvard Business Review, 79(3), 62–79.
tions (2nd ed.). San Francisco: Jossey-Bass.
Robert, M., & Racine, B. (2001). E-strategy.
Grant, R. M. (2005). Contemporary strategy Milan, Italia: McGraw-Hill.
analysis 5/e. Oxford, UK: Blackwell.
Shapiro, C., & Varian, H. R. (1999). Information
Kaplan, R. S., & Norton, D. P. (2004). Strategy rules. Boston: Harvard Business School Press.
maps. Boston: Harvard Business School Press.
Vittori, R. (2004). Web strategy. Milano, Italia:
Katz, M. L., & Shapiro, C. (1986). Technology Franco Angeli editore.
adoption in the presence of network externalities.
Yaobin, L., Zhaohua, D., & Bin, W. (2007,
The Journal of Political Economy, 94, 822–841.
June). Tourism and travel electronic commerce
doi:10.1086/261409
in China. Electronic Markets, 17(2), 101–112.
Malcolm, C. (2001, February). Making a health- doi:10.1080/10196780701295974
care Web site a sound investment. Healthcare
Financial Management, 2(74), 74–79.
Malec, B., & Friday, A. (2001). The Internet and Endnotes
the physician productivity: UK & USA perspec-
tives. Paper presented at the EHMA Congress,
1
This chapter is the result of joint research
Granada. work done by all authors. Nevertheless,
Massimo Memmola is the author of section
Memmola, M. (2007). The development of a Web I, II and III; all other authors have equally
strategy in a healthcare organization: A case his- contributed to section IV, V and VI, the
tory. In L. Al-Hakim (Ed.), Web mobile-based ap- discussion and the conclusion.
plications for healthcare management (pp. 1-35). 2
The research was drawn up in February, 2007;
Hershey PA: Information Science Publishing. The analysis of the Web site took place in
Minard, B. (2002). CIO longevity and IT project March and April, 2007.
leadership. IT Health Care Strategist, 4(1), 3–7.
3
A net that can be accessed only by users of
a particular company/organization.

This work was previously published in Business Web Strategy: Design, Alignment, and Application, edited by L. Al-Hakim; M.
Memmola, pp. 170-197, copyright 2009 by Information Science Reference (an imprint of IGI Global).

970
971

Chapter 4.6
Using Semantic Web Services
in E-Banking Solutions
Laurent Cicurel
iSOCO, Spain

José Luis Bas Uribe


Bankinter, Spain

Sergio Bellido Gonzalez


Bankinter, Spain

Jesús Contreras
iSOCO, Spain

José-Manuel López-Cobo
iSOCO, Spain

Silvestre Losada
iSOCO, Spain

Abstract to define their own portfolio management strate-


gies regardless of the information provider. In deed
Offering public access to efficient transactional the semantic layer is a powerful way to integrate
stock market functionalities is of interest to all the information of many providers in an easy way.
banks and bank users. Traditional service oriented With due regard for more development of security
architecture (SOA) technology succeeds at provid- technological issues, research on SWS has shown
ing reasonable, good Web-based brokerage solu- that the deployment of the technology in commercial
tions, but may lack extensibility possibilities. By solutions is within sight.
introducing Semantic Web Services (SWS) as a way
to integrate third party services from distributed
service providers, we propose in this chapter an Introduction
innovative way to offer online real-time solutions
that are easy-to-use for customers. The combined When operating on the stock market, investors make
use of ontologies and SWS allows different users their decisions on the basis of huge amount of in-
formation about the stock evolution, economic and
politic news, third parties recommendation and other
DOI: 10.4018/978-1-60566-066-0.ch016

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Using Semantic Web Services in E-Banking Solutions

kind of sources. Thanks to the proliferation of the SOA will be extended in the proposed solution of
Internet banks the profile of an average investor this chapter. Next section will first exposes the
is changing from a financial expert to common current situation of brokerage applications based
people making small investments on the online on classical SOA.
stock market. In addition to the business generated
around the stock market operations, banks use their
online stock market application to attract new and Current situation:
to reinforce the customer commitment. Brokerage Application
Banks, as any other commercial organization, based on Web Services
needs to optimize the deployment of new products
and services to the market. The deployment time Banking companies have invested heavily in the
of new services or applications is an important last few years to develop brokerage solutions
issue in a highly competitive market, since it based on a new dominant paradigm in the IT
defines the future market share and revenues. World: service oriented architecture (SOA). The
Online banks are looking for technologies and concept of this paradigm is not new: propose a
architectural paradigms that would allow them loosely coupled distributed system architecture
to design, implement and deploy new services on where independent services provide functional-
a low cost basis and in a short time period. New ity, so that the difficulty is divided which leads
services often imply integration of many already to reduce the development cost and improve the
existing applications, some of them internal and reusability. But the technologies to implement this
others external to the organization. paradigm are relative new. Web Services are one
This is the case of online stock brokerage of the solutions that appeared a couple of years
solutions adopted by online banks. An online ago and that made the success of this paradigm.
stock brokerage application proposes to the user For this reason Web Services are often confused
to buy and sell its stock options via a computer- with the SOA paradigm.
ized network. Banks are willing to offer an easy In this section we first present in more detail
to use application including as much information the business case for the brokerage application that
and as many options as possible without incurring we propose. We will then explain why a service
large development costs. We will show that the oriented architecture implemented using Web
use of the Semantic Web technology, combined Services technologies is a suitable solution. The
with a service-oriented architecture (SOA), greatly solution properties will be detailed and it will be
reduces the cost and effort of developing and shown that this kind of architecture is suitable for
maintaining an online stock brokerage solution. brokerage application. We then present what the
A broker based on a semantic service oriented benefits of such an architecture are from both, a
architecture has all the advantages of a service technical and a business point of view.
oriented architecture (e.g. modularity, reusability)
combined with the advantages of Semantic Web Web-Based Brokerage Applications
technologies. Semantic Web technology main
advantage is to give a clear semantic inside (and Introduction
eventually outside) the enterprise which reduces
the communication confusions (technical or hu- As a major interface between the financial world
man). This also leads to higher maintainability of and the non-financial world, banks always try to
the products and to a better automatisation of the improve their services related to the stock mar-
system mechanisms. These advantages applied to ket. As the Internet represents one of the most

972
Using Semantic Web Services in E-Banking Solutions

interesting communication channels of recent However, the Internet medium, as a cheap and
years, banks are interested in using this channel universal way to perform banking operations is
to improve the quality of their service and thus the highlighted solution of this business case.
increase their image and revenue. Such banks or In current brokerage solutions banks usually
bank departments have been called eBanks or only offer the service of making the operations
online banks.We have identified three different (buy, sell) but rarely integrate the search of relevant
strategies for online Banking: information to make the transaction decision. The
delivery of this ‘hard to integrate’ functionality
• Technological leader profile: Banks that search, is, however, a useful to provide to the end
focus their strategy on technology and user, who wants to buy and sell stock knowing the
consider the Internet an opportunity to im- most relevant and current information. The user
prove their markets. Also, the Internet spe- usually searches this information on external pages
cialized banks, usually recently founded independently of the bank services. Some of these
banks (not subsidiaries) that have earn a are free services which propose information of
significant market share, even though they the stock market in real-time or with a minimum
do not offer their clients a wide range of delay. Web pages such as Yahoo Finance (http://
products. finance.yahoo.com/), Google Finance (http://fi-
• Follower banks profile: Banks that first nance.google.com), Reuters UK (http://uk.reuters.
considered the Internet as a threat. When com/home), Xignite (http://preview.xignite.com/),
the market has matured, they changed Invertia (http://www.invertia.com) are good ex-
their strategy from a defensive position to amples of financial information from different
a competitive attitude towards those who providers with different degrees of quality.
were the first leaders in Internet banking. However, the need to search the information
In some cases, subsidiary entities were on several Web pages and then navigate to the
created so as not to cannibalize their own brokerage application to execute the transaction
market share. is a waste of time and adds unwanted complex-
• Non “internetized” banks: Banks that did ity to the end user. The idea is to build an online
not invest in the Internet because of their broker that merges and provides a unified or single
small size, their strategy or other reasons. point of access to information and operational
However, they are a minority in terms of services. In that way, the user will have an uni-
market share. fied environment which integrates most of the
tools required to fulfil his/her wishes of buying
In these days, banks that already have Web- and selling stocks. A comparison of a traditional
based brokerage application in place choose the brokerage application interaction and the new
technological leader strategy, while those banks interaction we proposed is showed in Figure 1. In
that are only now considering developing their the traditional interaction, the user must retrieve
own applications follow the second strategy. Other the information from each Data Provider (Yahoo
mediums are also possible: Finance, etc.) independently and only then do the
brokerage action (i.e. buy or sell). In the unified
• Branches: Too expensive for Banks. Only environment we propose, all the interactions are
for high end clients. performed through the brokerage application and
• Phone banking service: Expensive for the brokerage application take care of showing the
banks. Only for selected clients best data information regarding the user context
• Mobile services (SMS, mobile phone ap- (user profile, portfolio of the user, etc.).
plications): Cheap for banks, usually free In the next section we define the functional
or at a small fee per service usage. and non-functional requirements of our business

973
Using Semantic Web Services in E-Banking Solutions

Figure 1. Brokerage application interaction comparison

case. The following sections present one relatively bank services: buy, sell.
new but already commonly used way of using a • Complex conditional queries: Possibility
service oriented architecture (SOA) to implement to write a complex order in terms of condi-
these requirements. This solution, however, has tions such as “if the stock value of Cisco
some problems exposed later on that we resolve is higher than X and its volume is lower
exposing a new and innovative way using SOA than Y, ...” that may use different source of
combined with Semantic Web Technologies. information. Logical combinations of the
conditions should be possible.
Functional Requirements • Simple entry point of all services:
Complete integration of the conditional
The functional requirements the brokerage appli- queries and operational functionalities
cation must support in order to fulfil the business within one simple entry point.
case are summarised below:
Non Functional Requirements
• Stock market consultation functional-
ities: The application should be able to re- The non-functional requirements that the broker-
trieve information about the stock market age application must fulfil are:
such as the price of a share, volume of a
share, historical information, etc. Several • Highly maintainable: As the stock market
sources can be used to obtain the informa- is an entity subject to rapid change, it must
tion necessary to the supply the user with be possible to maintain the application in
the information they require to make the an optimal way.
trade. • Usability: The application is aimed at non-
• Customer information consultation expert end users. The application should
functionalities: The application should be as usable as possible in order to present
able to obtain easily the customer informa- to the end user a friendly and easy-to-use
tion such as his portfolio, buy and sell his- interface.
tory and recent searches by the user. • Extensibility of the information source:
• Operational functionalities: Invocation Possibility to easily add and change the pro-
of operations on the stock market using the viders of the banking information services.

974
Using Semantic Web Services in E-Banking Solutions

And to extend and choose the categories of The software design has fundamentally changed
information that the user wants to see. For in system design.
example, if the user is executing a buy or The Organization for the Advancement of
sell transaction the user may want to see Structured Information Standards (OASIS) de-
different sets of information. fines the service oriented architecture as follows
(MacKenzie et al, 2006):
Solution Based on SOA/
Web Services A paradigm for organizing and utilizing distrib-
uted capabilities that may be under the control
Introduction of different ownership domains. It provides a
uniform means to offer, discover, interact with
During the last four decades, software design has and use capabilities to produce desired effects
been prone to many changes. After abstracting consistent with measurable preconditions and
software code from the hardware infrastructure, expectations.
computer scientists thought to write code in so
called black-boxes and invented function oriented In this definition a service is designed as an
software design. The next big revolution was ob- entity with on one hand measurable preconditions
ject oriented software design, in which data was and on the other hand measurable expectations.
intended to be packaged in objects where objects It can be reformulated as the input and output of
are metaphors of real world entities. Objects were a function in a typical programming paradigm,
then abstracted in components in order to manage but in a service the input (precondition) and the
the problem of the increasing number of objects. output (expectation) are no data but state of the
A component can be defined as a set of objects world or effect on the world.
that has a coherent meaning as a standalone entity. We must clearly separate the architecture from
What is the next level of abstraction? A composi- the underlying technology that can be imple-
tion of components will always be a component if mented. The Web Services made the fame of the
we only focus on the data that these components SOA, but there exists other technologies which
contain. The composition must then be thought are totally suitable to be used in a Service-oriented
of a set of components that fulfil a given task. By Architecture such as: RPC, DCOM, CORBA, or
doing so the packaging is no longer data-oriented WCF (Donani, 2006).
but service-oriented, the set of components does no In the following subsection we present a stan-
more contain information and methods to access dard SOA for a brokerage application using Web
to this information but is a black-box that offers Services as the underlying technology.
one specific service. The services can then be
composed in more abstract services and be part of Architecture
the entire system, a service-oriented system.
Choosing service-oriented applications allows An overview of the architecture of a brokerage ap-
the clear separation of the users (commonly called plication based on a SOA/Web Service architecture
‘consumers’) from the service implementation (Booth et al., 2004) is described in Figure 2.
(commonly called ‘producers’). By having this The Web Services are physically located in
distinction, the application can then be distrib- either the service provider such as Bankinter or
uted on several platforms and possibly across Xignite that provides their own Web Services, or
networks. Each platform can have its own tech- in a specific Web Service container in the case that
nology and can be located in any physical place. the Web Service provider does not provide a Web

975
Using Semantic Web Services in E-Banking Solutions

Figure 2. SOA Architecture of a brokerage application

Service Interface and some wrapping mechanisms SOA allows the redundancy inside the sys-
are needed. This is the case with the above yahoo tem to be reduced since the Web-services
based Web Service in which analysis (wrapping) can be used more than once for different
of the Web page is needed to extract the right purposes. This allows also delivering new
information. functionalities in shorter time.
The Web Service Descriptions are stored inside • Loosely-coupled: In a loosely-coupled
a Broker Service Provider which publishes all system, each entity (Web Services) makes
the descriptions and is in charge in managing the its requirements explicit and makes few
publication inside a Service Registry. assumptions about other entities. This per-
Through this registry, Web Services can be mits not being aware where we locate the
found and communication between the application Web Services and thus increases the IT ef-
and the Web Service is done using the SOAP. ficiency (critical Web Services can have an
This quite simple and elegant architecture has adapted hardware framework), improve
a lot of advantages that are explained in the fol- the Quality of Service and reduce the costs.
lowing subsections on a technical and business Another advantage of a loosely-coupled
perspective. system is to have standard interfaces. This
way there is no need for technical people
Benefits of SOA on a to know about the whole details of the sys-
Technical Perspective tem and allows a strategically organisa-
tional separation of skills inside the project
SOA is finding increased adoption across more team.
and more business domains. This evolution can
be explained by a set of technical benefits. The Benefits of SOA on a
two most important ones are the following: Business Perspective

• Reuse: By decentralizing the systems As many times the decision of choosing an archi-
in self contained atomic Web Services, a tecture is done on a business level, it is important

976
Using Semantic Web Services in E-Banking Solutions

for an architecture type to have good benefits on a described at a syntactical level. For exam-
business perspective. The main SOA advantages ple, in WSDL XML Schema technology is
are: used to describe what the interchanged ob-
jects are. However, this kind of technology
• Business effectiveness: If one word could only allows describing the type of the ob-
be chosen to describe a SOA, it would be jects: string, date, integer etc. The seman-
the word “agility”. By using loosely-cou- tic of what the objects are is missing. This
pled services, the responsiveness to market adds inside the project development a lot
is highly increased. Each process of the of potential problems:
system is better controlled and allowed a
deployment of resources based on the busi- Misunderstanding of what is interchanged (if
ness needs. the documentation is badly done, errors can be
• Cost efficiency: Service-oriented architec- quickly done)
tures Enables reduction of the development Integration problems due to different type
costs by separating the skills and efforts in definition (for example, one service could describe
specific development areas. The separation an address with different fields while the service
in services allows putting the resources in consumer use an unique string)
the technical areas that correspond best
to their skills and thus reduce the costs • Poor visibility: As defined by the standard-
of training or hiring new resources. The ization group OASIS inside the OASIS
maintenance costs of such systems are also SOA Reference Model, the visibility refers
highly reduced because of well separated to the capacity for those with needs and
services and technology and location inde- those with capabilities to be able to see
pendence. Last cost reduction is the price/ each other (MacKenzie, 2006. In standard
performance optimization based on the SOA, the visibility is mainly provided by
freedom to select the adequate platform. means of a registry which lists the avail-
• Risk reduction: Dividing problems in able services. By having only a syntactic
smaller parts always has the advantage to description of the Web Service, the vis-
reduce the risk of a project, thus SOA is ibility is highly reduced and more efforts
based on this division, we can say that the in terms of search and analysis are needed
risk of projects based on SOA are inher- by any entity that wants to consume a Web
ently reduced. Another point is the risk Service.
reduction of the system deployment which • Manual work: In standard SOA system,
can be done incrementally. an important effort of Web Services inte-
gration is needed in order to develop the
Problem Statement entire system. The orchestration (composi-
tion) of the necessary Web Services corre-
We have presented the benefits of using a SOA sponds to one the highest effort time spent
based on Web Services to develop brokerage ap- inside the project due to the mostly manual
plications. However, there are a number of points in work that these efforts imply. Middleware
which the standard SOA is unable to respond: is often used in order to solve the orches-
tration. Mediation (data conversion) is on
• Web services heterogeneity: In the Web this topic the major problem. The previ-
Service technologies, the information is ously cited visibility problem also implies

977
Using Semantic Web Services in E-Banking Solutions

a lot of effort time because of the manual the IT components. Each service contains
effort spent during the localization of each the description of what the service propos-
Web Service. es to do. Operation names and the semantic
of the parameters are described in natural
All these points constitute the problems of language.
actual brokerage application based on pure SOA. • Semantic descriptions: This solution
However, as discussed earlier SOA has a lot of envisages a formal description of the ser-
advantages. This encourages us to re-use this solu- vices, called annotation. The services are
tion and improve it. The proposed solution of the not described using natural language but
next section aims at solving these problems of the with the proper mechanism of the chosen
actual solution by adding semantic technologies technology.
to this SOA paradigm.
This last solution, called Semantic Web Service
(SWS), is the one that we propose in this section
Solution: Broker based on because it represents the most advanced and thus
Semantic Web Services suitable of these three solutions. The annotation
is done by using so called semantic technologies,
As seen in the previous section, SOA technology in which the components of the system (here Web
provides a number of powerful concepts that when services) are formally described by using semantic
applied to brokerage applications allow us to resources (usually ontologies). This technology
construct flexible and easily extensible systems. is the base of the vision of Tim Berners-Lee who
However, problems have been identified with this put the base of a Web where the computer will
kind of technical approach; those problems are be able to optimally understand and compute
responsible for an important part of the cost of the information (Berners-Lee, 2001). We will
such applications. The identified problems were explain how the use of ontologies, which is the
vocabulary heterogeneity, poor visibility of the base of most semantic system, adds visibility to
services and manual work needed in the devel- the components by homogenizing the vocabulary
opment and maintenance phases. If we succeed used. We will also point out how this enhancement
developing a homogeneous vocabulary for Web leads to a reduction of manual work and thus a
Services, then we would increase the visibility reduction of cost.
and reduce the manual work. Three solutions were
proposed in (Verma et al., 2007): Powerful Functionalities

• Pre-agree on all terms (operation name, The aim of annotating Web Services is not only to
parameters): This implies a high oral add clarity in the Web Service definitions but also
communication between the development to allow the Web Service to be read by machines.
team and a lot of documentation writing. This machine-readability makes the power of
Pre-agreeing on the terms with no formal- the SWS by adding to the system the following
ly technical structure implies a high risk functionalities:
for any company due to the risk of losing
common knowledge or getting integration • Power to reason: The machine is able
problems. to “understand” what the Web Service is
• Comment all aspects of a service: In this doing. It is able to interpret the messages
solution, the comments are added inside that are interchanged. The messages are no

978
Using Semantic Web Services in E-Banking Solutions

more only pure data structure but are struc- the correct Web Services with the appropri-
tured in such a way that the data can be ate functionalities and Quality of Service
analyzed and transformed (mediated). For parameters are found. This allows a better
example, if a SWS receives information decoupling between: what the user wants
of a “client” but expected a “person”, he and what the system proposes. By separat-
is able to infer that a client is a person and ing these two parts, the system gets more
is able to extract the right information. The flexible.This SWS functionality responds
whole information space is structured and to the functional requirements about the
coherent. Axioms are responsible of main- processing of complex queries. The user
taining this space coherent and reasoners expresses a complex query in Natural
are the medium to do this. As described in Language or through a Web Interface, this
the functional requirements, the brokerage query is translated into the correspond-
system needs a system that is able to handle ing formal language and this formal query
a unique point entry. The reasoning capa- can then be used to retrieve the best Web
bilities given by SWS fit perfectly to these Services. More than one Web Service can
requirements, the queries can be interpreted be accessible for the same functionality;
by the system and the system can identify the system takes care at choosing the more
what the user’s wishes are. Additionally, as adapted one.In terms of maintainability,
the machine is able to interpret what the the separation of the SWS is also impor-
input of the user is, it is able to help the tant in the sense that adding new duplicat-
user at the moment of maintaining the sys- ed SWS of other providers does not require
tem. For example, a company has a broker- modifying the application. If new types
age application that allows the user to buy of SWS are added, some extents must be
stocks. However, for marketing purposes added to the query generation functional-
the number of times that the customer can ity of the brokerage application. But this
buy depends on the profile of the customer. task remains relatively easy because of the
In this case, the maintainer of the broker- use of ontologies which takes care of the
age system could want to add a new type coherence of the system. The automatic
of client. With a SWS-based system, this discovery responds to the non functional
is highly simple, as the only action to do is property of the “extensibility of the infor-
to add the concept of this new type in the mation source”.
ontology and add the information on how • Automatic orchestration: SWS support
many operations he can do. No additional the automatic orchestration or composi-
development efforts are needed and no risk tion of Web services (Medjahed, 2003). By
of adding errors in the application is run. orchestration, we mean the composition
If the maintainer makes an error in adding of the Web Services in order to provide a
the information inside the ontology, the more complex service. As Semantic Web
reasoner will warn him before he put it in Services are semantically annotated, the
production. system has enough information to handle a
• Automatic discovery: Formally describ- user query and respond to it by assembling
ing what each Web Service does adds the the Web Services.Automatic orchestration
functionality of automatically discovering provides an easy way to combine a usable
them. This means that a query written in a interface to the user, with one entry point
formal language can be interpreted so that that provides the three main requirements:

979
Using Semantic Web Services in E-Banking Solutions

stock market consultation, costumer in- actually done: WSDL-S (Akkiraju et al.,
formation consultation and operational 2005) that is at the Member Submission
invocation. stage in W3C and SAWSDL(Farell et al.,
2007) that is a W3C proposed recommen-
SWS Technologies dation. Main advantage of these approach-
es is that the annotation is done directly
The Semantic Web Service technologies have in the WSDL / XML Schema. Thus, the
been in the last few years under intensive research evolution of existing systems is facilitated.
world-wide. In the actual states, two approaches Other advantage is that these specifications
have been developed. Each one of these approach are ontology language independent, thus
were part of research projects and their validity execution engine can be developed for any
was proven by the deployment of concrete use- chosen ontology language. Both languages
cases. The two approaches are: have the necessary development tools.

• Pure Semantic Web Services: These Tools already exist and are operational to
technologies represent a way to write model and run Semantic Web Services. Most of
pure Semantic Web Services. By pure, them were part of a research project and are freely
we mean that they are written directly in available on the Internet.For modelling SWS, the
a formal language and are independent following tools are available:
from any non-Semantic Web Services. Of
course, all SWS technologies need to be • WSMO Studio1: A SWS and semantic
able to be connected with non-Semantic Business Process Modelling Environment.
Web Services (called grounding) in order Also support SAWSDL. As described by
to support any already developed Web ser- the name, this tool supports the WSMO
vice system. But the idea is to be able to Framework. It is Eclipse-based and the last
build new Semantic Web Services that will version is 0.7.2 released on 29/11/2007 (in
not carry on the “old” non semantic tech- the moment that this chapter is written: end
nologies. There are two main technologies of 2007)
based on this approach: OWL-S (Martin • Web Service Modelling Toolkit
et al, 2003) based on the OWL ontology (WSMT) : A lightweighted framework
2

language and WSMO (Fensel et al., 2007) for the rapid creation and deployment of
(de Brujin, Bussler et al., 2005)based on the tools for SWS. It supports WSMO
the WSML (de Brujin, Fensel et al., 2005) Framework. It is Eclipse-based and the last
ontology language. The first is mainly a version is 1.4.1 released on 13/09/2007 (in
North American development effort, while the moment that this chapter is written: end
the second one has been developed within of 2007)
EU-funded projects (Sekt, DIP, Knowledge • Radiant (Gomadam, 2005)3: A WSDL-S
Web, ASG and SUPER projects). They / SAWSDL Annotation Tool developed by
both are submitted to the W3C and have the University of Georgia. The annotation is
the necessary specification, development made using OWL ontologies. It is Eclipse-
tools and execution engines. based and the last version is 0.9.4beta re-
• Semantic Annotation of Web Services: leased on 29/05/2007 (in the moment that
The second approach consists in directly this chapter is written: end of 2007).
annotating the WSDL with semantic infor- • ODE SWS (Corcho, 2003)4: A toolset for
mation. Two main specification efforts are design and composition of SWS. It is based

980
Using Semantic Web Services in E-Banking Solutions

on UMPL and some work has been done to Approach and Architecture
integrate OWL-S.
• OWL-S IDE (Srinivasan, 2006)5: A de- The approach taken for creating a brokerage ap-
velopment environment supporting a SWS plication with SWS is to use the SWS engine as
developer through the whole process from a central component of the architecture. By tak-
the Java generation, to the compilation of ing advantage of the reasoning capacities of the
OWL-S descriptions, to the deployment SWS engine, it is possible to build a simple and
and registration with UDDI. The last ver- extendible Brokerage Application. New Semantic
sion is 1.1 released on 26/07/2005 (in the Web Services are added directly in the engine and
moment that this chapter is written: end of we minimize the development costs of managing
2007). new services. The SWS engine “understands” the
• OWL-S Editor (Elenius, 2005)6: A semantic of the new added SWS and only few
Protégé Ontology Editor plugin for a easy modifications are needed inside the Brokerage
creation of SWS. The last version was re- Application itself. This approach has been proven
leased on 04/11/2004 (in the moment that during the DIP project on a use-case (see the two
this chapter is written: end of 2007). screenshots Figure 3 And 4) that simulated a bro-
kerage application with Bankinter and external
The following SWS Engines are available: Web Services.The user can enter a complex query
composed of several conditions and one action
• WSMX7: The reference implementation of to be taken. The conditions are connected with
WSMO logical operators (AND/OR). The conditions can
• Internet Reasoning Service III (IRS-III)8 be of the types:
(Domingue et al, 2004): A SWS frame-
work, which allows applications to seman- • If the price of a specific stock is higher than
tically describe and execute Web services. a given price.
• OWL-S tools: A series of tools • If the value of an index is lower than a
WSDL2OWL-S, Java2OWL-S, OWL- given value.
S2UDDI, etc. are available at: http://www. • If the expert recommendation is equal to a
daml.ri.cmu.edu/tools/details.html specified one.
• If the variation of the value of a given stock
Through these tools represent good proofs of is higher.
the viability of the technology. That said, further
development would be required to adapt them For each information that needs to be retrieved,
to the needs of real world system. Professional the SWS Engine is responsible for discovering the
benchmarks would be needed to identify efficiency best suitable Semantic Web Service, eventually
and security lacks and allow the development of by composing multiple Semantic Web Services
professional SWS frameworks. (orchestration) and invoke the one (/ones) that
We gave a short overview of the existing correspond to the given Quality of Services pa-
SWS technologies and we explain now how rameters (time to respond, localization, etc.). If a
these technologies can be applied to brokerage Semantic Web Services is grounded on some other
applications. service systems (like normal Web Services), it is
in charge of getting the information and convert-
ing it into the semantic language. The Brokerage

981
Using Semantic Web Services in E-Banking Solutions

Figure 3. First screenshot of the SWS based brokerage prototype

Figure 4. Second screenshot of the SWS based brokerage prototype

982
Using Semantic Web Services in E-Banking Solutions

application then returns the result of the execution the user intention. This goal is the entry point to
to the Web Interface. the SWS Engine.
Figure 5 shows the architecture that implements As SWS Engine, WSMX (Bussler et al., 2005)9
this approach.The three main components are: was chosen over IRIS-III (Domingue et al., 2004)
(Cabral et al., 2006)10 in order to prove the correct
• The Web User Interface that should re- implementation of this SWS Engine inside the
spond to the Usability non-functional research project. From the input goal provided
requirement. by the brokerage application and some optional
• The Brokerage Application that should Quality of Service parameters, the SWS engine
support all the functional requirements and discovers the necessary SWS and invokes the
that is in charge of the communication with retrieved SWS in the right order (orchestration).
the SWS Engine. The brokerage application can then have access
• The SWS Engine that is in charge of man- to all the information it needs to check the con-
aging the semantic resources: discovering, dition provided by the user and if the conditions
invoke and orchestrate the SWS. are validated execute the buy/sell order through
another call to the SWS engine.
The brokerage application prototype developed The Financial ontology, exhaustively described
in DIP has been developed using J2EE technolo- in (López-Cobo, 2008) plays a major role in all
gies. The application makes use of Natural Lan- the tasks of the SWS engine and is the pillar of
guage Processing technologies to offer a simplified the whole brokerage application. It describes the
interface to the user. Receiving one sentence as vocabulary of the application and is used to an-
input, the brokerage application is able to iden- notate the SWS on both levels: the functionality
tify what is the user intention and automatically description (capability) and the interface (message
retrieve the information that it needs to invoke exchange).
the SWS. These parameters are used to generate a By using an architecture based on the Financial
WSMO goal (de Brujin, 2005) formally describing ontology and the SWS engine, we provide a flex-

Figure 5. Brokerage application architecture

983
Using Semantic Web Services in E-Banking Solutions

ible and maintainable application and provide to where virtually any customer, anywhere at anytime
the brokerage system the whole benefits of using could be using the bank transactional. This transi-
SWS technologies. tion meant high investments on scaling the main
In the next section, we describe in a higher frame architecture to a 24*7 architecture adding
level the cost and benefits of adopting such ar- several layers: Web servers, application servers,
chitecture. database servers, security layers...
Fortunately a Web Service scenario is more
natural in the actual client-servers environment,
Cost and Benefits therefore, in terms of cost/benefits the investment
is lower. The same reasoning can be applied to a
From a business point of view, the profits of the SWS scenario, where besides the new Web Ser-
proposed solution must not be focused on new vice layer to be added, semantic pieces appear to
incomes neither on costs, although they both exist. complete the puzzle.
The resulting application is intended to create a Therefore, taking the chance of adding seman-
new product, by giving new options to manage tic layers to an existing bank architecture implies
their portfolio. These options could have been  de- low economical risk since no major implications
veloped using a more traditional approach but, due are needed to expand the current architecture.
to the complex and usually mature architectures As a result, and taking into account the business
used by the financial institutions, the costs would opportunity, the small amount of effort required to
have been significantly higher, both the develop- create and maintain the service, and the technical
ment and the future sub-applications costs. prerequisites, the SWS approach emerges as a
Also, a more traditional approach (i.e. with- smart solution to create the new service at a rea-
out semantic technologies) would have implied sonable level of cost, both for the developer and,
agreements with the information providers (data which is more important, for the final user.
formats, relevant data, how to provide the infor-
mation, how to access to the data, etc), which will
usually implied a one-by-one Trading Partner Risk Assessment
Agreement (TPA). The semantic layer gives us
the ability to smartly read the provided data and From a business point of view, alternatives are
therefore to manage it easily. It is also easy to add almost always more expensive but it could depend
new providers with this approach. Finally, data is on how each Financial Institution manage its own
accessed when required and if required, making Stock Market Services. What is more, in several
use of the Web Service advantages. cases the solution could be so complex that it
The costs and benefits, from the technical could be considered as ‘nearly impossible’ to
point of view, must be also considered. The Cost/ develop without studying in depth what is going
Benefits ratio in terms of adapting actual systems, to be modified, (Stock Market applications are
although not trivial, is not as dramatic change as critical tasks for Financial Institutions due to the
the one that was performed in the transition to volatility of the market and the quality assurance
the Internet era. that is required in this specific market).
Banks were usually based on main frame ar- At the same time, the new proposed service is
chitectures. The scenario in these last 10-20 years intended to give better utilities to the clients. These
has changed from a exclusive main-frame scenario, kinds of services are actually free, although the
where only the bank employees had access to the final user must manage with them. So the price of
IT transactional systems, to a Web Based scenario the service for the final clients could not be high

984
Using Semantic Web Services in E-Banking Solutions

and thus, it has to be developed at a reasonable aggregators are the Internet ‘killer applications’
cost.There are some risks when the use of SWS (i.e.: Google, Yahoo, You Tube, etc) investors are
is considered: expected to make use of Stock Market data ag-
gregators (in fact, the actually use them: Yahoo!
• Security issues: No doubt this is the ma- Finances, Invertia, etc), making their own buying/
jor functional risk to be considered when selling decisions and finally performing them in
deploying SWS technologies. On banking their favourite Stock Market site.
environments security is the mayor column None of these services are designed to perform
on which all the architecture must be built. automatic operations that completely fulfil the
All the security issues must be clarified and investor requirements, nor of them are prepared
solved before any transactional application to perform a personalised strategy when data ag-
using real customer banking data is de- gregation is required. There are several solutions
ployed to the real world. If this milestone for professional investors but they are available
is not achieved all the SWS-based appli- at a high cost, thus they are only interesting when
cations will be forced to handle only with high volumes are regularly performed (high vol-
public data and the real value of semantic ume both in terms of number of transactions and
applications will remain as a proof-of-con- in terms of money invested). But there they are
cept not as a real-world-application. not an option for individuals.
• Evolution issues: SWS techniques are in Our tests11 reveal that people usually make
their first steps of use in business environ- use of at least two Stock Market services just to
ments. As these techniques become more fulfil their information requirements. The SWS is
familiar they will evolve and this evolu- intended to aggregate them and, in the near future,
tion could mean scalability issues that to automatically perform the actions according
should be treated as any other scalability to the investor strategy, combining sources and
issues inside any corporation environment. retrieving specific data form them.
Although the semantic techniques are ma-
ture their wide use could imply changes
that would mean changes on the semantic Conclusion
platform. However these two evolution is-
sues are natural to any IT development or Responding to the need of maintainable and ef-
to the deploy of any new technology. The ficient brokerage applications, we have presented
IT business, no matter if it is the banking in this chapter a novel approach that combines the
business or any other, has got enough ex- SOA architecture and the semantic technologies.
perience to handle these potential risks. By using the proposed solution, we resolve the
three identified problems of a non-semantic SOA
solution: heterogeneity of the vocabulary used in
Future Trends the services which reduces the maintainability and
possibilities of evolution of the application, the
The evolution of the Stock Market and its asso- poor visibility which reduces the possibility for
ciated services must be forecasted as part of the automatic discovery and the lot of manual work
global social tendencies: people (and investors) that is generated by this poor visibility and the
are requiring more and more sophisticated prod- lack of automatic composition possibilities.
ucts and services allowing them to make their We exposed the advantages of a solution based
own decisions. If we consider that information on Semantic Web services: reasoning functional-

985
Using Semantic Web Services in E-Banking Solutions

ities, automatic discovery and automatic orches- Berners-Lee, T., Hendler, J., & Lassila, O. (2001).
tration. Such functionalities allow us to build an The Semantic Web. Scientific American, 284,
architecture centralized on the SWS engine and 34–43.
have a really flexible system.
Booth, D., Haas, H., McCabe, F., Newcomer, E.,
The feasibility to build such brokerage applica-
Champion, M., Ferris, C., & Orchard, D. (2004)
tion has been demonstrated during the European
Web Services architecture.
project called DIP. The output of these projects
is a framework called WSMO associated with Bussler, C., Cimpian, E., Fensel, D., Gomez, J.
an ontology language (WSML) and execution M., Haller, A., Haselwanter, T., et al. Zaremba
engines (WSMX and IRS-III). The European Maciej, Zaremba Michal (2005). Web Service
Commission continues to invest money in the Execution Environment (WSMX).
research in Semantic Web Services for example in
Cabral, L., Domingue, J., Galizia, S., Gugliotta, A.,
the SUPER project which aims to take advantage
Norton, B., Tanasescu, V., & Pedrinaci, C. (2006)
of the Semantic Web Services in order to improve
IRS-III: A Broker for Semantic Web Services based
Business Process Management Systems. A lot of
Applications, The 5th International Semantic Web
research is also done in other technologies such
Conference (ISWC 2006), Athens, GA, USA,
as OWL-S. The high activity of these research
Corcho, O., Fernández-López, M., Gómez-Pérez,
projects reflects the important interest that should
A., and Lama, M. (2003). ODE SWS: A Semantic
have industrial investor in such technology. The
Web Service Development Environment. VLDB-
Semantic Web services are ready and continue
Workshop on Semantic Web and Databases, 203-
to be improved.
216. Berlin, Germany.
Companies should consider the benefits of
SWS on two levels: the strategic level and the de Brujin, J., Bussler, C., Domingue, J., Fensel,
tactical one. On the strategic level, the SWS D., Hepp, M., Keller, U., et al. (2005). Web Service
give the possibility to build highly maintainable Modeling Ontology (WSMO).
applications and profits from a loosely coupled
de Brujin, J., Fensel, D., Keller, U., Kifer, M.,
architecture. On a tactical level, a company
Lausen, H., Krummenacher, R., et al. (2005). Web
should see the benefits on other projects where
Service Modeling Language (WSML).
the ontologies that have been created for the SWS
are reused and form a common base for the ap- Domingue, J., Cabral, L., Hakimpour, F., Sell,
plications. Using such Semantic technologies is D., & Motta, E. (2004). IRS-III: A platform, and
interesting for companies because it promotes the infrastructure for creating WSMO-based Se-
homogeneity of the systems inside the company. mantic Web services. In Proc. Of the Workshop
The intra-company and inter-company applica- on WSMO Implementations. CEUR Workshop
tions are then much easier integrable. Proceedings
Donani M.H. (2006). SOA 2006: State Of The
Art. Journal of Object Technology, 5.
References
Elenius, D., Denker, G., Martin, D., Gilham, F.,
Akkiraju, R., Farrell, J., Miller, J., Nagarajan, M., Khouri, J., Sadaati, S., & Senanayake, R. The
Schmidt, M., Sheth, A., & Verma, K. (2005) Web OWL-S editor- A Development Tool for Semantic
Service Semantics - WSDL-S, Web Services. In The Semantic Web: Research
and Applications. Series: Lecture Notes in Com-
puter Science, Vol. 3532, 78-92. Springer Berlin /
Heidelberg 2005. ISBN: 978-3-540-26124-7

986
Using Semantic Web Services in E-Banking Solutions

Farell, J., & Lausen, H. (2007) Semantic Annota- Srinivasan, N., Paolucci, M., & Sycara, K. (2006).
tions for WSDL and XML Schema. Semantic Web Service Discovery in the OWL-S
IDE. In Proceedings of the 39th Annual Hawaii
Fensel, D., Lausen, H., Polleres, A., Brujin, J.
international Conference on System Sciences -
d., & Stollberg, M. Romand D., Domingue J.
Volume 06 (January 04 - 07, 2006). HICSS. IEEE
(2007). Enabling Semantic Web Services – The
Computer Society, Washington, DC, 109.2.
Web Service Modelin Ontology. Springer.
Verma K., Sheth A. (2007). Semantically An-
Gomadam K., Verma K., Brewer D., Sheth A. P.,
notating a Web Service. IEEE Computer Society,
Miller J. A. (2005) Radiant: A tool for semantic
March/April 2007, 11, 83-85.
annotation of Web Services, ISWC 2005.
López-Cobo, J. M., Cicurel, L., & Losada, S. On-
tology Management in eBanking applications. In
Endnotes
Ontology Management. Semantic Web, Semantic
Web Services, and Business Applications. Series: 1
http://www.wsmostudio.org/download.
Semantic Web and Beyond, Vol. 7. Hepp, M.; De html
Leenheer, P.; de Moor, A.; Sure, Y. (Eds.) 2008, 2
https://sourceforge.net/projects/wsmt/
Approx. ISBN: 978-0-387-69899-1 3
http://lsdis.cs.uga.edu/projects/meteor-s/
MacKenzie, M., Laskey, K., McCabe, F., Brown, downloads/index.php?page=1
P. F., & Metz, R. (2006) Reference Model for
4
http://kw.dia.fi.upm.es/odesws/
Service Oriented Architecture 1.0.
5
http://projects.semwebcentral.org/projects/
owl-s-ide/
Martin, D., Burstein, M., Hobbs, J., Lassila, O., 6
http://owlseditor.semwebcentral.org/
McDermott, D., McIraith, S., et al. (2003). OWL- 7
http://www.wsmx.org/
S: Semantic Markup for Web Services. 8
http://kmi.open.ac.uk/projects/irs/
Medjahed, B., Bouguettaya, A., & Elmagarmid,
9
http://www.wsmx.org
A. K. (2003). Composing Web services on the Se-
10
http://kmi.open.ac.uk/projects/irs/
mantic Web. The VLDB Journal, 12(4), 333–351.
11
Done inside the DIP (http://dip.semanticweb.
doi:10.1007/s00778-003-0101-5 org/) project, Deliverable 10.10.

This work was previously published in Semantic Web for Business: Cases and Applications, edited by R. García, pp. 336-352,
copyright 2009 by Information Science Reference (an imprint of IGI Global).

987
988

Chapter 4.7
Innovating through the Web: 

The Banking Industry Case
Chiara Frigerio
Università Cattolica del Sacro Cuore, Italy

Abstract financial institution. Let us think about Intranet por-


tal, content management tools, and business process
In recent years, the financial services industry has management suites, which are now quite spread in
been witness to considerable consolidation (Berger banks due mainly to their technological ease-of-use.
& Udell, 2006; De Nicolò, Bartholomew, Zaman, Thus, Internet is representing an innovation wave
& Zephirin, 2004; Figueira, Neills, & Schoenberg, extremely relevant for the financial industry as a
2007) and organizational progress in order to sustain whole, and the effects on banks’ performance is
two main objectives: efficiency and commercial emerging. What do we expect in the near future? In
effectiveness (Epsten, 2005; Sherman & Rupert, all probability, the usage of Web-based application
2005). In order to sustain customer-oriented and will be bigger and bigger also in other contexts of
efficiency strategies, banks have started to explore the bank processes, even if some risks could oc-
new ways of conducting their business, introduc- cur when clear strategies and change management
ing areas of innovation in their services, practices, practices do not direct the innovation.
and structures to offer the most complete array of
services possible (Quinn et al., 2000). On the other
hand, new services and products drive retail banks to Introduction: Innovation
explore new ways of producing or delivering these and Web Banking
novelties. This is true especially for Internet banking
services that offer services to customers 24/7, and The banking system is undergoing remarkable stra-
it becomes clear that adding new services, that is, tegic and organizational change processes, which
trading online or bill payments, is easily and quickly will consolidate and spread even in the future. Such
geared towards improving commercial effective- changes are not to be considered as a contingent
ness. The following chapter aims at describing to change in the design and development criteria of the
what extent the Internet has developed new services business, but they point out an innovation character
and businesses, and what are the main figures of the which extends far beyond those boundaries. The ap-
phenomenon in Europe. Moreover, the Internet has plication fields of innovation criteria are numerous
introduced new coordination processes within each (e.g., technological, organizational, commercial)
even though two main critical issues often hinder
DOI: 10.4018/978-1-60566-024-0.ch012

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Innovating through the Web

the creation and spread of innovation: a) the origin mechanisms which have the role of coordinating
of innovation, that is, the characteristics of the and managing skills, relations, and innovation
innovation source; and b) the possibility to know processes as a whole.
beforehand the effects of innovation and to keep With reference to the second critical issue
its sustainability in the long run. source, managerial literature associates to in-
With reference to the subject of the origin of novation the characteristics of “radicality.” An
innovation, the most traditional theoretic version innovation is radical when its adoption process
counters the organizational process innovation over time develops in nonlinear maturity phases,
with the innovation of product and service. “Prod- characterized by times of performance growth on
uct innovation is given by the introduction of a discontinuous effects. The main obstacle in the
new product or service to meet the market request innovation management is given by the difficulty
or a particular external customer, while process of previously predicting moments in which a
innovation is defined as a new element which is performance leap and the relevant extent of the
introduced into the organizational processes in performance growth will take place. The effects
order to produce a product or to distribute a ser- of innovation may be even more uncertain if the
vice” (Damanpour & Gpalakrishanan, 2001, pp. innovation recipient is a little controllable, being
47-48). Organizations develop product or service it external (e.g., the customer). As a matter of
innovation in order to increase their market shares fact, the lack of control over the external party
or to improve their strategic positioning, while prevents knowing the cause-effect relationship of
they innovate in the process in order to achieve the leverages affecting the innovation adoption
economies of scale and increase profitability process. Such leverages may also be generated
(Utterback & Abernathy, 1975). The innovation or limited by solicitations which are outside the
of the first type allows the development of the bank-to-client relationship, difficult to foresee
first phases of the life cycle of a product/service, and to manage.
therefore it involves relevant investments and high All the above mentioned introductory consid-
risks. On the contrary, organizations that operate erations may be useful in evaluating one of the
process innovations generally develop themselves main innovation areas developed by the banking
at the time in which the product/service is in a industry over the last 20 years, that is, the Web.
maturity phase, therefore in the long term. For As a matter of fact, the introduction of this
this reason, process innovation generally requires technology immediately generated a great ap-
less investment (Anderson a& Tushman, 1991; plication interest from financial institutions and
Barras, 1990). was directed mostly towards:
Moreover, by overcoming the dichotomy
between product/service innovation and process a. Development and management of both rela-
innovation it is possible to consolidate an approach tionships with the outside (i.e., stakeholder
which takes into consideration the combination and clients) and the inside (i.e., coopera-
of both drives within an innovation policy (See tors and employees), through the use of the
Figure 1). Therefore, both product/service and Internet and the Web as innovations in the
process innovation may be combined and may publication and circulation of contents or
generate added value within a short period and services.
with minor costs (board 3). Therefore, in some b. The development of “core banking” applica-
cases, the ability to innovate is measured both in tions and information systems in Web-based
the product/service part as well as in the process environments characterized by integrability
one. The convergence of the innovation stra- and pace of important achievements.
tegic approaches is ensured by organizational

989
Innovating through the Web

Figure 1. Product/service/process innovation

This chapter aims at exploring the first ap- Therefore, though the evolution of the Internet,
plicability area, by describing how and in which the banking phenomenon is continuous and steady;
way the Web development in such a context has another application area of the Web use, less visible
been innovative. Specifically, it analyzes first the but equally active, is the one of Intranet banking.
characteristics of Internet banking and then those Intranet is an information support available for
of Intranet banking. By the first term, we refer members of an organization aimed at providing
to the development of distribution channels of easy and immediate access to organizational and
product/services alternative to the physical pres- training information, thus reducing distribution
ence, based on Web technology and distributed costs and the level of information dispersion
through Internet. In this area, product innovations which potentially increases the job complexity
went and go hand-in-hand with process/channel and employee dissatisfaction.
innovations. As a matter of fact, certain products The development of information technology-
specifically created for the Internet channel, like based innovations, like Web banking, raises the
c/a, loans, and financial services, are supported by issue of the importance of an assessment of the
new management processes, aimed at providing threats of technology innovation. It is enough, for
procedure efficiency and standardization. The instance, to look at the problem of trade (products
joint product and process innovation enables to and channels), organization (production processes
support further innovation strategies aimed, on and innovation delivery), and information (data
the one side, at creating increasingly customer and information flows) integration, deriving from
behaviour-oriented products and, on the other, at the implementation of an Internet strategy. The
improving processes even in nonefficient areas. lack of integration represents one of the main
To support the theory concerning the mutual re- defeating threats of the innovation implementation
lationship of product and service innovation, one process (Ciborra, 1996). As a matter of fact, the
may consider the evidence substantiated by the latter is effective only when it is a part of a whole
“multichannel service strategies,” characterized by and is not out of context. Another technology
the presence of cross-channel and cross-customer innovation challenge is represented by the speed
product/services and processes. and the direction of the standard implementation

990
Innovating through the Web

process. Specifically in the Web banking industry, perspective “light” branches with employees able
the interoperability requirements are essential. to support customers. This was, for example, the
Technology innovations which may vary the Italian case of Banca 121. The current situation
widespread standards are particularly critical and is characterized by a higher number of universal
difficult to implement. banks integrating Internet service into their offer.
Only a few operators are solely virtual and they are
generally specialized in online trading. This last
Internet Banking: A service has long been considered by most opera-
Historical Summary tors as the killer-application of Internet banking.
At a time when investments in negotiable secu-
Internet banking started growing at the beginning rities had high returns, online trading developed
of the 90s and saw its maximum development at considerably and, in certain cases, it replaced the
the end of that same decade. figure of the financial promoter or of the financial
Its importance in the definition of bank com- manager. Subsequently, the development of further
mercial and marketing strategies was immediately services, not only linked to brokerage, allowed
clear. The development of applications which also the development of more complete Internet
would allow customers to remotely operate by strategies for retail banking (e.g., c/a transactions,
means of their home PC was minimally complex circulars, requests for consulting support) and also
and fairly cheap. But the decisional factor of the for private banking, characterized by sophisticated
main business plans on Internet banking was the and demanding clients.
evaluation of the transaction marginal efficiency. The e-commerce phenomenon for banks has
As a matter of fact, since the first analysis it was not yet given the expected results. Some banks
clear that the cost of the process of a transaction have activated e-commerce initiatives both in
opened on a virtual channel was considerably the business-to-business (B2B) as well as in the
lower than the physical one (e.g., that of a pro- business-to-consumer (B2C) areas. English and
moter branch agency). This scenario initially led Spanish banks proved to be the most active in this
some banks and then most of them to adopt an situation. Besides payment management, often ac-
Internet strategy. companied by the creation of special credit cards
Especially in the leading banking groups in in the case of B2C, banks also offer partnership
Europe, the potentials made available by Internet formal guarantee services (especially in the case
led to the creation of units fully dedicated to the of B2B portals) and of special financing. We have
management of virtual customers (both retail and progressively witnessed the passage from an activ-
corporate). Such units were often legally autono- ity of incubator and partnership, still present in the
mous companies, that is, subsidiaries of the parent less developed areas, to the activity of facilitating
bank. However, the strategy of direct banks (i.e., or aggregating, though the few existing cases do
those which were exclusively present on the virtual not allow us to consider e-commerce for banks
channel) does not have a long life. As a matter of as a consolidated phenomenon.
fact, on the one hand, besides the totally dedicated
channel, banking groups keep the possibility of
operating their accounts via alternative channels The Presence Online: An
(often by using the same virtual bank technology). Analysis of Web Sites
On the other hand, certain companies born to rep-
resent the mere virtual channel of their parent bank A recent survey carried out in 2003 allowed the
start creating, little by little, within a multichannel comparison of the Web world experience of the

991
Innovating through the Web

European banking key-players, with particular focused on those banking institutions which have
reference to strategies and behaviours (Carignani adopted the most interesting and innovative ini-
& Frigerio, 2003). tiatives from a technological point of view and
The survey targeted seven European coun- which, for this reason, may represent a reference
tries (i.e., France, Germany, Italy, Scandinavian model for banks of other countries like ours, where
Countries, Spain, Switzerland, UK), selected Internet penetration is still low, though growing,
among those having a higher level of Internet and the population is still little accustomed to use
usage penetration and characterized by important technological channels. Considering the central
investments through alternative channels. To role given to the Italian situation within our re-
this purpose, the survey analyzed: first the Ger- search, a wide sample of banking institutions was
man market, considered as the widest and most selected. In particular, attention was given to those
developed market of the European scenario; the banks which form part of the major groups of the
British one, whose financial and banking tradition country, as well as to single institutions having a
is, in many aspects, similar to the US banking considerable volume of customers.
system; the Scandinavian system, important for From the analysis of data provided in the
the extremely widespread Internet use among the field of the strategic choice made, both Italian
population; and the Swiss one, characterized by and European banks show the trend of looking at
a peculiar banking tradition in retail and private electronic channels as a way of diversifying their
banking. Other significant experiences closer to the offer. The search for new services which may be
Italian context are the one of France, characterized offered online, thus contributing to increasing
by a good spread of alternative channels, and the the added value given by the bank, is pursued by
one of Spain which, regardless of the disadvantage 68% of Italian banks and 38% of European banks.
compared to its European competitors, is now Price leadership is a primary value just for 10%
coming up strongly into the direct banking sector of Italian banks (typically those which are merely
with interesting offers. remote) and of 19% of foreign banks. This value
With regards to the individual banks selected makes even more sense if it is compared to the
within each chosen country, in Germany, whose result given by the same survey carried out in 2000.
banking system still represents today the reference As a matter of fact, in that occasion, it appeared
model of all operators of the Old Continent, and in that price reduction represented the main objec-
the UK, the survey investigated traditional banks tive of Italian banks, followed by the advertising
which chose to invest hugely in new channels by channel. Moreover, 24% of European banks are
implementing rather aggressive strategies, and the aiming at product differentiation, by introducing
most significant examples of direct banks born new products and services which have a different
as spin-offs of the major banking groups of their configuration for the Web.
respective countries. This way it is possible to Going more into details with regards to the
monitor and compare the direct banking initia- implemented strategies, the types of services of-
tives of both leading banks and followers. With fered have been analyzed, also comparing the offer
reference to the French market, the survey focused of other transfer services by remote banking.
merely on the main traditional banks in terms of All European and Italian banks which took part
customers, as well as on volumes and on the three in the survey declared they offer online transfer
most successful initiatives in the e-banking field. services, information, and value-added services.
In the Scandinavian banking industry, which, as Call centers basically provide the same offer va-
previously mentioned, is strongly geared towards riety. Actually, the range of operations which may
the use of innovative channels, the attention was be carried out through the use of the most recent

992
Innovating through the Web

Figure 2. Internet banking strategies

e-banking technologies is not so wide. Among available services. Seventy-four percent of Italian
banks which are equipped with Wap technology, Web sites and 53% of foreign Web sites taking
only 70% are able to offer transfer services but part in the sample offer financial simulators which
only 30% more innovative services, which bring a support customers in the most efficient portfolio
value added to the overall offer of the bank. SMS choices. Here again, one can notice the will of
messages are used mainly as an information tool. our virtual banks to aim at clients information. In
Moreover, the few banking institutions which Europe, the percentage is higher in Switzerland
have extended to the Web TV/digital television and Great Britain where banking institutions have
their interaction means with customers are trying a long tradition in managing savings. Discussion
to develop transfer services which are mostly of forums are present in some banks, especially in
transactional type. Central Europe, but they are also rapidly develop-
The online trading expansion has pushed banks ing in southern banking institutions.
to disclose market information. This informa- Financial information is the most widely value-
tion area is mostly explored by foreign banking added service (100%) offered by Italian banks,
institutions; to this effect, 94% of European followed by product information (93%), market
sites, compared to 84% of Italian sites, dispose trend graphics (90%), and online help like FAQ,
of a dedicated area. The majority of banks offer e-mail, and demos, which represent 86%, 86%, and
services for customers while visitors may consult 79% of the surveyed population, respectively. On
only stock exchange quotations. The service is the other end of the rating scale are multilingual
almost exclusively supplied by a specialized supports (10%), trading simulators, and online
information provider. financial consultancy (24% each). The latter, in
Banks also offer many value-added services; particular, shows a certain growth compared to
simulation tools and communication services (e.g., the previous survey, but is still too weak to be
forum, newsletter, mailing) are among the most considered representative.

993
Innovating through the Web

The European sample, on the other hand, shows number of images per page has reduced in order
a somehow different trend. The multilingual sup- to increase browsing speed; legible characters are
port is present in all European countries, with the used, as well as dark tones for the background and
exception of Great Britain, and the same goes for light tones for writing. Well aware that the brows-
financial stimulators. To the possibility of ask- ing difficulty constitutes a discouraging factor for
ing for advice by mail or to finding an answer to the average net surfer, banks are paying particular
frequent doubts, thanks to the FAQs, European attention to this aspect of their site.
financial institutions prefer to offer financial cal-
culators and asset allocation instruments (though
still in embryo). Intranet Banking as
From the institutional analysis of the Web site, an Organizational
it shows that 94% of banks of the sample deem it Integration Tool
indispensable to offer visitors and customers infor-
mation about the bank. A good 92% of Italian sites As we were saying, the study on the effects of the
describe the structure of the banking institution use of the Web technologies in the financial and
and its history, while just 79% of foreign banks banking industry cannot overlook the intercom-
provide this kind of content. Moreover, listed pany application scope. Banks are traditionally
banks offer the possibility to consult and download considered as complex, large-size organizations,
the latest balances and relating reports, as well as characterized by geographical widespread location
the most significant press articles. Italian banking and organizational and product differentiation.
institutions seem to give this information area a Due to the frequent merge and acquisitions trans-
higher value rather than their foreign counterpart, actions (M&As), the organizational complexity
as well as with regards to the array of news avail- has lately grown. As a matter of fact, the search
able on the site. All the analyzed Web sites contain for economies of scale in a minimally differenti-
their product catalogue. In this regard, there is a ated market characterized by end customers be-
difference in price listing: while 83% of Italian ing highly sensitive as far as the “price” variable
Web sites make reference to prices, only 21% of is concerned, pushed the bank top management
European Web sites make this information avail- to increase the volume of products and services
able to visitors. Generally speaking, however, offered and to be present in a wide range of mar-
the information available concerning this issue kets. The trend is therefore to look for economies
is poor and incomplete with regards to contents through merge and integration projects whose
and presentation clarity. main objective is to increase the company size
Access speed for the display of the Internet and to extend the bank geographical boundaries
banking service and of online trading has consider- even in international contexts. The search for
ably increased (8.75 seconds on average). Italian these situations of strong boost towards volumes
sites have a higher uploading speed compared to has an immediate effect on the competition of the
European ones, thanks to the reduced presence industry, which becomes, on the one hand, increas-
of simulators and other value-added services and ingly global and, on the other, increasingly based
to the improvement of pages design. In order to on efficiency and innovation parameters.
optimize the page opening time, HP got equipped Against a rather widespread strategic choice of
with direct service links and improvements were dimensional growth within the banking industry,
made in the site design. Italian banks got higher the distinguishing factor, on the one hand, and
scores in the design of Web pages (on average challenging factor, on the other, is represented by
4.1 against a mean European score of 3.73). The the integration pace issue, that is, the time needed

994
Innovating through the Web

Figure 3. Value-added services offered by online banks

to carry out strategic plans. Therefore, the success referring to the need to exchange information in
factor consists of the ability to create the conditions order to face and solve problems encountered
so that organizational systems and mechanisms during task performance) (Ferraro, 2000).
change and reach a sound configuration in the A common aspect refers to the progressive
shortest time possible. increase of internal communication flows chan-
Within this scenario, a role of paramount neled through groupware applications and, in case
importance is played by technologies and infor- of larger organizations, the presence of dedicated
mation systems in triggering changes through the Intranet and company portals. The networks
specification of common practices and procedures (namely Intranet) increase the capability of col-
in the operations and information management. lecting, processing, and redistributing “codified”
Lorence and Lorsch (1967) had already pointed knowledge. The same infrastructures allow users
out that the need for integration originates from to exchange messages and different texts, forward
organizational differentiation. requests, consult specialized document files, ben-
Increasing abilities aimed at managing or- efit from remote training and coaching courses,
ganizational interdependencies are considered and so forth. The use of such instruments for com-
one of the most evident effects of the spread of munication purposes is led by the growing use of
information technology (Agliati, 1996). In this solutions and services coming from third parties
context the importance of an extensive applica- (i.e., external suppliers or companies referring to
tion of different ICT solutions is stressed, aimed the same banking group).
at facing both operational interdependence (i.e., The introduction of the Internet paradigm for
linked to intermediate and final input and output the communication and exchange management is
flows which are created as an effect of division not so strong; as an example, the use of Extranet
of labour) and information interdependence (i.e., in the relationships with external partners is still

995
Innovating through the Web

marginal. Sometimes the relationship with certain (BPM). By the first term we refer to a set of solu-
outsourcers (with the exclusion only of persons tions that automate, partially or fully, the process
in charge of the management information system) of creation and publication of contents (CMt) and
uses the support offered by similar tools. of documents (document management) for users
Entrusting the operational phases to outside inside the bank (Intranet, internal portals) and out-
bodies, with the consequent need to carry out side the bank (Extranet, Web etc.). By the second
systematic controls on the work performed by the term we refer to the set of workflow technologies
supplier and on the creation of working teams of which manage and monitor the business process,
specialists coming from the extra-banking world, by integrating it with traditional applications.
may be the reason behind the development, in the Both such technologies may be considered in-
future, of computerized applications addressed to novative business integration tools, though their
external counterparts both at the top and at the role may be considered of coordination support
bottom of the company organizational limits. and of automation (or, to a certain extent, of deci-
Should this trend gain ground, we may witness sion support), respectively.
a further expansion of the limits of information However, their choice must be explained in
systems, with clear repercussions on the bodies the light of the organizational change that major
in charge of the operation and management of credit institutions have started in the past few
the said systems. years. Many of the cases considered show that
Numerous banks have realized how Intranets the implementation of CM or BPM solutions was
may increase the visibility of the organization, followed up by information system integrations or
since they make clear the existence of knowledge, migrations (48%), merge or acquisitions (23%),
routine, and interconnections between all parties and, in the case of BPM, after process revision
contributing to the same process. In certain fields projects (25%). Though the implementation of
the analysis perspective is widened and this ends both projects derived from common organiza-
up by making the same organization function tional needs, the data clearly show a different
transparently towards the different players. trend of the first ones compared to the second
ones (Figure 4).
Sixty-fiver percent of interviewed banks
Intranet Applications: declared that they are presently carrying out, or
The Case of Content they will soon carry out, CM projects, while only
Management and of Business 54% of the sample declared that they are pres-
Process Management ently carrying out or they are planning to carry
out in the short term BPM projects. About 10%
Information systems, in their totality, are essential of banks have planned investments in BPM in
tools aimed at achieving coordination. Proof of the long term. Therefore it becomes evident that
this are the investments that financial institutions CM projects are more widespread than BPMs
make in order to adopt innovative technologies among financial institutions, which means that
alongside the most traditional corporate informa- CM support technology is more widespread than
tion systems. BPM support technology.
A survey carried out by CeTIF1 on a sample of When considering the volume of the projects,
9 banking groups and 2 service centers, ended in we realize that if, on the one hand, CM may have
March 2005, showed the existence of significant reached a good circulation because it is applied
innovative projects in the fields of content man- quite frequently to performance information and
agement (CM) and business process management product/market information processes (pic. 3), on

996
Innovating through the Web

Figure 4. CM and BPM projects in financial institutions

the other hand, BPM is spread only in the field of and maintenance to the outside public and, only
credits and information processes (38%), finance in a second phase, inside the financial institution,
and treasury (25%), and auditing and risks (23%). have anticipated the BPM projects. The latter are
In certain cases (i.e., purchase invoice cycle, in- seen as management excellence solutions within
formation processes, collectionsm and payments), processes (68% of interviewed institutions) and
the BPM extends also outside the boundaries of as tools implemented to overcome difficulties
the credit institution, involving also customers due to the need for technical modernization of
or, often, suppliers. the traditional information systems (44%). The
results expected from the two solutions seem to
This Difference May be Due be very different, with consequent differences
to Different Factors in the implementation technical difficulties, in
costs, and in the identification of organizational
First of all, the factors may be the causes and effects. If, on the one hand, CM solutions are
the relating expected results which boosted the applied by the bank alongside the existing ones,
innovation introduction in both technological without the need of organizational changes and
applications. The survey shows that both CM important technological integrations, on the other
and BPM projects are mostly induced by exog- hand, BPM solutions are based on rather more
enous market forces and by the consequent need complex technologies, characterized by organi-
to reduce the so called “time to market” (71%). zational uncertainty.
If, on the one side, the market pressure and the Even the analysis of the obstacles faced by CM
introduction of the customer-oriented culture and BPM projects may be useful to understand
get, in theory, banks to look for ways to improve the origin of the different spread of such projects.
internal processes and communication integration, Aggregated data show a common sensitivity to-
on the other side, reality shows that the equipment wards the cost factor (about 24% of the sample
with tools bringing immediate and recognizable indicated costs as the main obstacle to such
advantages, mostly outside the bank itself, was initiatives) and the relating difficulty in defining
made easier at first. Therefore, CM projects, the return on investment (ROI). Another strong
aiming at facilitating information publication obstacle is due to the cultural and organizational

997
Innovating through the Web

reluctance (78%) to the change brought about play the main role by completing their integration
by these technological innovations. On the other and automation capacity.
hand, difficulties relating to technologies (18%) Moreover, CM and BPM projects differenti-
do not seem to be a fundamental inhibitor factor. ate for their application modality. While CM
The empiric evidence show, however, a differ- is characterized by a big-bang implementation
ence between the obstacles to the CM and BPM logic, which introduces technologies in a sole
projects, with particular reference to the cultural phase, BPM is often characterized by a modular
and organizational aspects. The reasons are to be implementation system.
found, first of all, in the characteristics of managers
and users of these technological innovations and,
secondly, in their implementation modalities. An Innovation Evolution
In the case of CM, 44% of respondents state that Perspective
the people in charge of these projects are active in
the lines of business (LOB), while 32% are active It is estimated that the difference in the pres-
in the human resource management (HR) or in the ent status of CM and BPM will fade more and
organization. At present, only 18% of financial more due to the growing awareness that internal
institutions have “created” a new professional processes are the very same business apparatus;
position (i.e., knowledge officer) which centralizes this is due even in the light of the recent regula-
content management in one organizational area; tions on operational risks and also because the
the new trend seems to confirm the growth of this two technologies may eventually complement
organization solution. BPM projects, instead, are each other.
managed by the LOB in 24% of the cases, and by In particular, the survey shows that the dif-
the IT area in 32% of the cases. When looking ferences between the two technologies may be
at the characteristics of users of the above men- due, on the one hand, to a lesser knowledge of
tioned technologies, important differences arise. the BPM technological tool compared to the CM
CM solutions are more and more spread among one, and on the other, to the lack, limitedly to the
customers rather than being centrally used. This BPM projects, of the domino (or isomorphism)
entail the support of the training structure whose effect which is common during the technological
task is to create professional people able to man- innovation application phases and which leads
age the content publication process, as well as, to a resemblance of technological choices by
in certain cases (about 50% of the sample), the financial institutions.
relevant monitoring process. Even BPM are spread We hereby would like to hypothesize a hidden
amongst users that manage the processes which link between the two technologies and the pres-
make the object of this study. In particular, with ence of an “incremental” effect between them.
regards to processes, a need of coordination and In order to verify this hypothesis, one must look
rationalization of business procedures in the most first of all at the technological and organizational
interdependent areas arose. Bank information innovation which CM on the one side and BPM
systems are made up by parallel and integrated on the other bring about.
applications, often based on different and complex CM technological innovation may be sum-
information structures. However, there is no need marized with the introduction of documental
to create ad hoc professional figures which man- integration tools for archive management and in
age the BPM process and project. As a matter of distributed environments. From a functional point
fact, BPM tools often go hand in hand with or of view, CM allows the automated management
replace transactional information systems which of the lifecycle of a document, as well as the need

998
Innovating through the Web

to trace the documents and information contained We therefore believe that BPM is an evolution
therein. For this reason, such tools are considered and extension of the CM concept (Figure 5).
as tools enabling the spread of the internal knowl- CM projects are therefore destined to evolve
edge of the credit institution. towards BPM projects extended to other processes,
From a technological point of view, BPM whenever the financial institution is able to recog-
introduces innovation integration between the nize the CM as a technological innovation tool.
information systems. Moreover, from a functional
point of view, they lead to the identification of
process interdependencies and to their manage- Some Useful Insights
ment harmonization. Moreover, it is a preparatory
tool aimed at the process mapping, which can The considerations made on Web-based tech-
be used for a range of activities (e.g., risk man- nologies, both Internet and Intranet, enable one
agement, auditing system, definition of internal to draw interesting conclusions already outlined
regulations, etc.). at the beginning of this chapter and concerning
According to this short synthesis, an evolution the innovation characteristics.
perspective for the two technologies seems to Empirical evidence leads us to focus on how
take form. The common characteristics may be the theory may explain the following:
outlined as follows:
1. Which are the technological innovation
• Process concept: Both technologies im- signs? Can they always be measured? Does
ply a strong idea of organization through technological innovation progresses through
processes. In the case of CM, this merely stages or is it an on-going process?
refers to the document creation and publi- 2. Is technological innovation unique? Does
cation processes, while in the case of BPM, it make sense to speak about innovation
this concept is also extended to other busi- of each technology sector (automation,
ness processes. decision-making support, coordination)?
• Culture of change: It shows that the or- 3. What is the relationship between technologi-
ganizational change following the intro- cal innovation and organizational efficiency/
duction of such technologies is difficult to effectiveness?
manage, though it is characterized by dif-
ferent factors. The bank should be able to Innovation is a synonym of “change or im-
change and, for this purpose, the manage- proving evolution of a situation” and surely not
ment sponsorship is important. a novelty. Some writers consider technological
• Organizational and technological moni- innovation as technology flexibility or as an im-
toring: The technology innovation seems provement of performance. Here we would rather
to be linked to the need of monitoring its exclude such definitions, since the definitions of
implementation from both the technologi- flexibility and performance may be ambiguous
cal and organizational point of view. and “contaminated” by other concurrent factors
• Technological flexibility: Technological (Maggi, 2003). For this reason, it is therefore
innovation goes hand in hand with the believed that technological innovation is not mea-
flexibility concept, seen as a capability surable per se. However, the innovation process
to “adapt” to the company structures and may be seen as an on-going process, more or
systems. less fast, characterized by a constant change in
technology, players, and structures. This concept

999
Innovating through the Web

Figure 5. The technological innovation stages

leads us to consider technological innovation as gitudinal analysis over a longer temporal span,
a natural process, only little conscious of the fact for the determination of effects produced by
that it is stimulated by the implementation of information technology. However, a qualitative
certain technologies which may affect the pace consideration can be drawn: The two technolo-
(by increasing it or stopping it) and the process gies showed different efficiency values against
awareness. the same technology use. This proves, as a first
Hence, we conclude that the answer to the estimate, that information technology does not
second question is negative. The division in have a certain and determining role for reaching
classes, given at the beginning of this chapter and organizational efficiency, though it may influence
recognized as valid by the organizational literature, both the information need, as well as the informa-
seems not to be relevant in the case of innovation. tion processing capability. Therefore, its effects
We could at least “defend” such classification by seem mediated by organizational and contextual
construing it as typology, that is, as a set of types variables, leading to different values with regards
(rather than classes) which do not have precise to effectiveness.
demarcation limits and which constitute some Therefore, the objective of this chapter is to
archetypes. The concept of technological innova- define to what extent information technology
tion, however, remains an extensive concept, not may affect the organizational efficiency, that is,
referable to any typology. the capability to reduce the operating costs of
The third questions leads to wider consid- the organizational structure. On this point, the
erations. Though the concept of organizational interviewed persons stressed the high importance
efficiency and technological innovation has been given to the role of information technology. IT
highlighted by many interlocutors, the gathered enables the processing of great volumes of data,
data, however, do not provide empirical evidence. a deeper control, the cost cutting in terms of
The reason shall be found in the lack of a lon- communication and transactions. Hence a new

1000
Innovating through the Web

question arises: Does any information technology ◦◦ The information system is an orga-
have an impact on transaction costs and on oper- nizational planning factor; it is able
ating costs? The answer seems to be yes, though to affect the organizational efficien-
the impact magnitude on efficiency is different. cy and therefore the organizational
While automation and communication technolo- structure.
gies highly influence the economic management ◦◦ Besides information systems, the or-
of the institution, decision-making support tech- ganizational structure is affected by
nologies have an indirect impact, more difficult other variables; this observation leads
to asses beforehand. us to consider the absence of techno-
logic determinism.
◦◦ Information technology has a me-
Conclusion diated role in defining the organi-
zational effectiveness and a direct
This study dealt with technology innovation, with role in the organizational efficiency
particular reference to Web technologies. The dif- management.
ferent empirical evidence shows that technology
may be a tool to improve the service/product of- However, the role played by the said tech-
fered, as well as an efficiency tool. As a matter of nologies affects the organizational structures in
fact, the implementation of Internet technologies a different way.
brought about, on the one hand, the development
of a considerable number of Internet strategies
for products and services of Italian banks. On References
the other hand, Internet technologies supported
the coordination as well as the organizational Agliati, M. (Ed.). (1996). Tecnologia
integration phases of the same banks. dell’informazione e sistema amministrativo.
Process and content innovation mainly implies Milan: Egea.
a change in the cultural attitude and, for the time Anderson, P., & Tushman, M. L. (1991). Managing
being, it is still determined mostly by endogenous through cycles of technological change. Research
variables. Technology Management, 34(3), 26–31.

The analysis of the collected data leads us to Barras, R. (1990). Interactive innovation in finan-
draw some last conclusions: cial and business services: The vanguard of the
◦◦ Information systems are affected by service revolution. Research Policy, 19, 215–237.
information technology; in order to doi:10.1016/0048-7333(90)90037-7
determine the information system,
Berger, A. N., & Udell, G. F. (2006). A more
one cannot do without knowing the
complete conceptual framework for SME fi-
development and implementation sta-
nance. Journal of Banking & Finance, 30(11),
tus of the technology.
2945–2966. doi:10.1016/j.jbankfin.2006.05.008
◦◦ Information systems are affected by
the task uncertainty, by coordination
and control mechanisms, and by com-
munication tools; the organizational
(micro-) variables affect the defini-
tion of the information system.

1001
Innovating through the Web

Buckingham, R., Hirschheim, R. A., Land, F. F., Frigerio, C. (2005). Process IT Innovation in
& Tully, C. J. (1987). Information systems cur- banking. Università Cattolica, Milan, CeTIF
riculum. A basis for course design. In Buckingham Research Centre.
et al. (Ed.), Information systems education: Rec-
Lawrence, P. R., & Lorsch, J. W. (1967). Organiza-
ommendations and implementation (pp. 72-113).
tion and environment, graduate school of business
Cambridge: CUP.
administration. Boston: Harvard University.
Carignani, A., & Frigerio, C. (2003). European
Maggi, B. (2003). De l’Agir Organisationnel.
Web banking strategies. Università Cattolica,
Toulose: Editions Octares.
Milan, CeTIF Research Centre.
Quinn, J. B., Anderson, P., & Finkelstein, S. (2005).
Ciborra, C. U. (1996). Improvisation and informa-
Leveraging intellect. The Academy of Management
tion technology in organizations. In Proceedings of
Executive, 19(4), 78–94.
the 17th International Conference on Information
Systems, Cleveland, OH (pp. 369-380). Sherman, H. D., & Rupert, T. J. (2006). Do
bank mergers have hidden or foregone value?
Damanpour, F., & Gopalakrishnan, S. (2001). The
Realized and unrealized operating synergies in
dynamics of the adoption of product and process
one bank merger. European Journal of Opera-
innovation in organizations. Journal of Manage-
tional Research, 168(1), 253–268. doi:10.1016/j.
ment Studies, 38(1), 45–65. doi:10.1111/1467-
ejor.2004.05.002
6486.00227
Utterback, J., & Abernathy, W. (1975). A dy-
De Nicolò, G., Bartholomew, P., Zaman, J., &
namic model of product and process innovation.
Zephirin, M. (2004). Bank consolidation, inter-
Omega, 3(6), 639–656. doi:10.1016/0305-
nationalization, and conglomeration: Trends and
0483(75)90068-7
implications for financial risk. Financial Markets,
Institutions, & . Instruments, 13(4), 173–217.
Epstein, M. J. (2005). The determinants and
ENDNOTE
evaluation of merger success. Business Hori-
zons, Elsevier, 48, 37–46. doi:10.1016/j.bush- 1
The research mentioned was carried out by
or.2004.10.001 the author between March 2004 and October
Ferraro, F. (2000). L’analisi organizzativa: 2004 through a quali-quantitative analysis,
l’individuo. In R. Mercurio & F. Testa (Eds.), in the form of semi-structured questionnaires
Organizzazione, assetto e relazioni nel sistema di and interviews (Frigerio, 2005).
business (pp. 52-76). Torino: Giappichelli.
Figueira, C., Neills, J., & Schoenberg, R. (2007).
Travel abroad or stay at home? Investigating
the patterns of bank industry M&As in the
EU. European Business Review, 19(1), 23–39.
doi:10.1108/09555340710714135

This work was previously published in Business Web Strategy: Design, Alignment, and Application, edited by L. Al-Hakim; M.
Memmola, pp. 219-234, copyright 2009 by Information Science Reference (an imprint of IGI Global).

1002
1003

Chapter 4.8
Semantic Web for Media
Convergence:
A Newspaper Case
Ferran Perdrix
Universitat de Lleida, Spain & Diari Segre Media Group, Spain

Juan Manuel Gimeno


Universitat de Lleida, Spain

Rosa Gil
Universitat de Lleida, Spain

Marta Oliva
Universitat de Lleida, Spain

Roberto García
Universitat de Lleida, Spain

Abstract and Semantic Web technologies the best choice for


Web-wide information integration. However, news-
Newspapers in the digitalisation and Internet era are papers have made great investments in their current
evolving from mono-channel and static communica- news management systems so a smooth transition
tion mediums to highly dynamic and multi-channel is required in order to reduce implementation costs.
ones, where the different channels converge into a Our proposal is to build an ontological framework
unified news editorial office. Advanced comput- based on existing journalism and multimedia
erised support is needed in order to cope with the standards and to translate existing metadata to the
requirements arising from convergent multimedia Semantic Web. Once in a semantic space, data in-
news management, production and delivery. Such tegration and news management and retrieval are
advanced services require machines to be aware of a facilitated enormously. For instance, Semantic Web
greater part of the underlying semantics. Ontologies tools are being developed in the context of a media
are a clear candidate to put this semantics into play, house that are capable of dealing with the different
kinds of media managed in the media house in an
DOI: 10.4018/978-1-60566-066-0.ch009 integrated and transparent way.

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Semantic Web for Media Convergence

Figure 1. Traditional news information flux (left) and the new trend of convergent news flux (right)

Current Situation transmitters make information in collaboration


with other transmitters and produce messages
Web news publishing is evolving fast, as the major- that include as media as it is necessary (video,
ity of Internet services, and nowadays this service text, audio, images…). Finally, receivers choose
is trying to adapt information to a way that best fits the channel that best fits their needs in order to
users’ interests in order to increase its use. With get access to messages.
that, newspapers are expecting to profit more from The previous situation is the one faced in the
their news sites. In parallel, many of the newspaper context of the Diari Segre Media Group1, which is
companies are changing into news media houses. a journalism holding that in the last years has been
They own radio stations and video production facing this convergence trend. This holding started
companies that produce content unsupported by 25 years ago with a newspaper edition. Today
traditional newspapers, but that is delivered by produces three press editions in two languages,
Web newspapers or new mobile services. Initially, three radio stations, six television regional chan-
Web news was a mere reproduction of those in nels and several Internet Websites. Nowadays, all
the printed edition. Nowadays, they are constantly the editorial staff is applying the convergence of
updated and provide new services for those users information flux approach during news generation
interested on reaching this information as soon as and management. Therefore, they are required
possible and enjoying new ways of interaction with to be versatile journalists because they cannot
them (Eriksen & Ihlström, 2000; Lundberg, 2002; be specialized in any concrete media. They must
Ihlström, Lundberg, & Perdrix, 2003). deal with video, image and text edition. Moreover,
Consequently, news industry communication they must write in different ways, for instance for
model is changing from the traditional one shown press news or for radio or TV voiceover.
on the left of Figure 1 to the one shown in the On the other hand, the Diari Segre archive
right. In the former, each channel is considered system is changing to a new repository build
separately (press, TV, radio, Internet, mobile from the combination of text, images, video and
phones…) and implies his way creating his own audio files. In this sense, archive management
message, transmitting over this channel and using is becoming a big issue and it requires deep im-
his own interface in order to show the message to provements in terms of content search, relations
the receivers. On the other hand, the latter is based among news (e.g. historical relations among news
on an information convergence flux. In this model, items) or information retrieval interfaces. The

1004
Semantic Web for Media Convergence

archive system must be a very productive and ing subsections, they just provide data structures
comprehensive tool in order to assist journalists and schemas that facilitate systems interoperabil-
while they create new content. This business case ity. They do not facilitate knowledge management
details how Semantic Web technologies are being and information retrieval tasks. These tasks are
explored in the context of the Diari Segre Media currently carried out mainly by the documentation
Group in order to face this new challenges. department, who is in charge of the news archival
In general, it has been observed that media process using legacy tools.
houses must adapt to the requirements imposed by
this new model. First of all, there are changes in Journalism Metadata
how they reach consumers. News are build up from
a combination of different content types (video, One of the main standardization frameworks in
audio, the traditional text plus images, etc.) and the journalism domain is the International Press
are delivered to users through different channels Telecommunications Council2, an international
and adapted to many kinds of devices (PC, PDA, consortium of news agencies, editors and news-
smart phones, etc.). Therefore, formats must be papers distributors. IPTC has developed standards
selected and adapted according to the device and like the Information Interchange Model3, News-
connection the user is using. These operations Codes4 (formerly the Subject Reference System),
include transcoding of formats, resizing of im- the News Industry Text Format5 or NewsML6.
ages or recoding for higher levels of compression. Currently, almost all of them have evolved
Moreover, multi-channel distribution must take towards XML-based standards to represent and
into account that for each channel one must define manage news along their whole lifecycle, includ-
its own content, aesthetic and interaction model. ing their creation, exchange and consumption.
These characteristics define what an interactive For instance, NewsML is used to represent news
channel is (McDonald, 2004). as multimedia packages and NITF deals with
However, changes are not just restricted to the document structure, i.e. paragraphs, headlines,
relation with consumers. Digital media eliminates etc. On the other hand, the Subject Reference
many time and space restrictions and changes System (SRS), now part of IPTC NewsCodes, is
editorial team routines. Moreover, all different a subject classification hierarchy with three levels
media converge into a unified news object that and seventeen categories in its first level.
is produced by interdisciplinary teams. Conse- Moreover, a new family of these journalism
quently, more efficient and effective means for standards has been just proposed, expanding the
news management are needed in order to facilitate range of available metadata. The new suite, known
coordination and production of these multimedia as the IPTC G2, is actually a series of specifica-
assets. tions and XML components that can be shared
The news industry is currently using content among all IPTC G2 components for maximum
management solutions for these means, but the efficiency.
additional requirements of a convergent editorial IPTC G2 Standards make possible the integra-
office stress the need for advanced knowledge tion of any news item with text, photos, graphics,
management and information retrieval. Currently, video or other media. The News Architecture
there are specific standardisation efforts in the model (NAR) is used in order to package any
journalism domain, together with more generic combination of these items. Moreover, it makes
ones in the multimedia domain, which carry some stronger use of IPTC’s robust metadata taxonomy
uniformity to the newspaper content management suite, which is based on NewsCodes, and better
systems. However, as it is introduced in the follow- interacts with other IPTC-G2 standards.

1005
Semantic Web for Media Convergence

This standard contains hooks for managing Problem Statement


news items, and its flexibility allows news provid-
ers to choose whether to support all of the IPTC As has been pointed out in the description of the
G2-standards XML tags or a compact subset. It’s current situation in many media houses, archivists
the cost-effective way of managing news, whether classify news using a proprietary hierarchical
for a Web site, news aggregator, newspaper, radio thesaurus while journalists search this information
or television station. when they need to inform themselves on subjects,
histories or events. This search can be performed
Multimedia Metadata in extreme situations, e.g., lack of time, or lack
of knowledge in relation to the archive system.
All the previous initiatives are centred on the jour- This is reflected in the way journalists formulate
nalism specific aspects of a semantic newspaper. their queries. The gap between archivists’ and
However, as has been pointed out, newspapers are journalists’ mental models implies that more flex-
evolving towards the digital multimedia domain. ible content categorization and search systems are
Therefore, they stress more and more their mul- needed. This trend is even bigger when we consider
timedia management requirements. cross-media content production and coordina-
In the multimedia metadata domain, as it tion in order to get multimedia and multichannel
is extensively shown in the literature (Hunter, news.Therefore, in order to take advantage of
2003; Einhoff, Casademont, Perdrix, & Noll, the possibilities offered by the digital medium to
2005), the MPEG-7 (Salembier, & Smith, 2002) exploit a newspaper archive, the aspects that can
standard constitutes the greatest effort for mul- be improved include:
timedia description. It is divided into four main
components: the Description Definition Language • Keyword search falling short in expressive
(DDL, the basic building blocks for the MPEG-7 power
metadata language), Audio (the descriptive ele- • Weak interrelation between archive items:
ments for audio), Visual (those for video) and users may need to combine several indirect
the Multimedia Description Schemes (MDS, the queries manually before they can get an-
descriptors for capturing the semantic aspects of swers to complex queries
multimedia contents, e.g. places, actors, objects, • Lack of a commonly adopted standard rep-
events, etc.). resentation for sharing archive news across
In addition to MPEG-7, which concentrates on newspapers
content description, MPEG-21 defines an open • Lack of internal consensus for content de-
framework for multimedia delivery and con- scription terminology among reporters and
sumption. This standard must be also considered archivists
because it focuses on the content management • Lack of involvement of journalist in the ar-
issues for full delivery and consumption chain, chiving process
from content creators’ applications to end-users’
players. The different MPEG-21 parts deal with These shortcomings are difficult to deal with
diverse aspects like Digital Rights Management if the existing standards are used as provided. The
or Digital Items, the definition of a fundamental main standards that have been presented, both in
content unit for distribution and transaction very the journalism and multimedia domains, are based
useful for convergent media management. on XML and specified by XML Schemas. The
more significant case is the MPEG-7 one. It is
based on a set of XML Schemas that define 1182

1006
Semantic Web for Media Convergence

elements, 417 attributes and 377 complex types. implicit semantics are available from a computer
NewsML and NITF are also very big standards, processing point of view.
they define more than 100 elements, and the
NewsCodes hierarchy of subjects defines more Proposed Solution
than one thousand different subjects.
The complexity of these standards makes it In this chapter, we explore Semantic Web tech-
very difficult to manage them. Moreover, the use of nologies (Berners-Lee, Hendler & Lassila, 2001)
XML technologies implies that a great part of the as a way to overcome many of the challenges of
semantics remains implicit. Therefore, each time digital and convergent media houses. The size and
an application is developed, semantics must be ex- complexity of the stored information, and the time
tracted from the standard and re-implemented. limitations for cataloguing, describing and order-
For instance, if we use XQuery in order to ing the incoming information, make newspaper
retrieve MPEG-7 SegmentType descriptions archives a relatively disorganised and difficult to
from an XML database, we must be aware of the manage corpus. In this sense, they share many of
hierarchy of segment types and implement an the characteristics and problems of the World Wide
XQuery that has to cover any kind of multimedia Web, and therefore the solutions proposed in the
segment, i.e. VideoSegmentType, AnalyticClip- Semantic Web vision are pertinent here.
Type, AudiSegmentType, etc. In order to implement more advanced news-
If the intended interpretation of the segments paper content management applications, they
structure was available for computerised means, should be more informed about the content they
semantic queries would benefit from the cor- are managing. They are not just files with some
responding formal semantics. Consequently, a weak interrelations. There is a lot of knowledge
semantic query for SegmentType will retrieve all embedded in these pieces of content and in their
subclasses without requiring additional develop- interrelationships. In order to make computers
ing efforts. This is not possible with XML tools aware of it, their implicit semantics must be for-
because, although XML Schemas capture some malised, for instance using ontologies. Semantic
semantics of the domain they model, XML tools are Web technologies facilitate the building blocks
based on syntax. The captured semantics remain for Web ontologies, which add the facilities for
implicit from the XML processing tools point of Web-wide ontology sharing and integration. The
view. Therefore, when an XQuery searches for a latter is a key feature for convergent and globalised
SegmentType, the XQuery processor has no way media houses.
to know that there are many other kinds of seg- In order to build an ontological infrastructure
ment types that can appear in its place, i.e. they for the Semantic Newspaper, it is important to
are more concrete kinds of segments. consider the state of the art of the metadata initia-
The previous example only illustrates one kind tives in the journalism domain, which have been
of difficulty derived from the use of just syntax- introduced in the current situation description
aware tools. Another example is that the lack of section. Additionally, digital newspapers have
explicit semantics makes MPEG-7 very difficult stressed the requirements of multimedia manage-
to extend in an independent way, i.e. third party ment. Digital news is managed as multimedia
extensions. The same applies for MPEG-21 or the packages that integrate text, images, video, audio,
journalism standards. Moreover, standards from etc. Therefore, it is also important to consider the
both worlds share many concepts so it would be current situation in the more general multimedia
possible, and easier, to integrate them once their metadata domain.

1007
Semantic Web for Media Convergence

We have undertaken the application of the Overview


Semantic Web proposals to the newspapers world
by following a smooth transition strategy (Haust- The proposed solution is detailed in Section 3.
ein, & Pleumann, 2002). This strategy advises First of all, Section 3.1 presents the methodology
about keeping compatibility, at least initially, that produces an ontological framework based on
with current newspaper content management existing standards. This methodology is based on
systems and journalism and multimedia standards. two mappings. The first one from XML Schema,
Consequently, we have rooted our proposed ap- the language used in most of the considered
proach on existing journalism and multimedia standards, to ontologies based on the Semantic
standards and provide a methodology to move Web language Web Ontology Language (OWL)
them, together with existing data, to the Semantic (McGuinness & Harmelen, 2004). The second
Web domain. one is based on the previous one and makes it
possible to map from XML metadata, based on
Objectives XML Schemas previously mapped to OWL, to
Semantic Web metadata, based on the Resource
The objective is then to design a Semantic Web- Description Framework (RDF) (Becket, 2004).
based platform that is an extension of previously The ontologies produced using this methodol-
working systems in mass media companies, par- ogy constitute the foundation on top of which an
ticularly in the context of the Diari Segre Media architecture based on Semantic Web technologies
Group. The manual creation of semantic instances is built. This architecture, described in Section
for news items, at a regular daily pace, is indeed a 3.2, takes profit from the semantics formalised
feasible goal as long as this process is integrated by these ontologies and loads Semantic Web
into existing systems and it just causes a slightly metadata based on them in order to offer services
greater work load while producing observable like semantic integration, semantics queries or
benefits. Consequently, the introduction of new logic reasoning. These services are used in order
semantic documentation tools requires a careful to build applications that facilitate managing het-
work of analysis, design, testing and balancing of erogeneous media repositories and the underlying
the additional burden that such tools may impose knowledge. One example of such an application
on archivists, journalists or readers. is given in Section 3.3.
In order to produce a semantic platform that The described application builds on top of a
seamlessly integrates into newspapers content text-to-speech and a semantic annotation tool.
management systems, the first objective is to The generated annotations are based on existing
develop an ontological framework based on exist- standards ontologies and loaded into the proposed
ing standards. Once this ontological infrastructure semantic architecture, which makes it possible to
based on existing journalism and multimedia stan- manage audio, audiovisual and text content in an
dards is developed, the objective is then to put it integrated way. However, the key point here is to
into practice in the context of an architecture based offer all the semantic services to users in a usable
on Semantic Web tools for semantic integration, and accessible way. To this end, the application
querying and reasoning. However, all this effort is based on a user interface that provides an
must end up reaching users through applications object-action interaction paradigm best suited for
that offer to them the extra benefits of semantic heterogeneous information spaces. The interface
metadata while avoiding them the burden of deal- does not solely facilitate content management,
ing with the underlying extra complexity. it also allows browsing the underlying domain
knowledge, formalised using specialised ontolo-

1008
Semantic Web for Media Convergence

gies, and constitutes a useful tool in media houses has been already produced using XML, which
in order to facilitate news tracking and producing is extensively used by many newspaper content
new content. management systems.
There are many attempts to move metadata
from the XML domain to the Semantic Web.
Solution Details Some of them just model the XML tree using
the RDF primitives (Klein, 2002). Others con-
This section provides a detailed description of centrate on modelling the knowledge implicit
the proposed solution. The different modules are in XML languages definitions, i.e. DTDs or the
described in the following subsections starting XML Schemas, using Web ontology languages
from the methodology used in order to benefit (Amann, Beer, Fundulak, & Scholl, 2002; Cruz,
from existing standards and produce ontologies Xiao, & Hsu, 2004). Finally, there are attempts
that formalise them. These ontologies make pos- to encode XML semantics integrating RDF into
sible to develop an architecture that takes profit XML documents (Lakshmanan, & Sadri, 2003;
from their semantics in order to offer advanced Patel-Schneider, & Simeon, 2002).
services like semantic integration, querying and However, none of them facilitates an extensive
reasoning. Finally, these services are used in order transfer of XML metadata to the Semantic Web in
to build an application that makes the benefits a general and transparent way. Their main problem
emerging from semantic metadata and ontologies is that the XML Schema implicit semantics are not
available for end users. made explicit when XML metadata instantiating
this schemas is mapped. Therefore, they do not
benefit from the XML semantics and produce
XML Semantics Reuse RDF metadata almost as semantics-blind as the
Methodology original XML. Or, on the other hand, they capture
these semantics but they use additional ad-hoc
In order to put into practice the smooth transition semantic constructs that produce less transparent
strategy, the first step has been to reuse exist- metadata.
ing standards in the journalism and multimedia Therefore, we propose the XML Semantics
fields, which have been for long very active in Reuse methodology, which is implemented by the
standardization. ReDeFer project7 as an XML Schema to OWL plus
However, as has been highlighted in current and XML to RDF mapping tool. This methodol-
situation analysis, all the more recent standards ogy combines an XML Schema to Web ontology
are based on XML and lack formal semantics mapping, called XSD2OWL, with a transparent
that facilitate applying a Semantic Web approach. mapping from XML to RDF, XML2RDF. The
Therefore, in order to facilitate the transition from ontologies generated by XSD2OWL are used
current standards and applications to the semantic during the XML to RDF step in order to gener-
world, we have applied the XML Semantics Reuse ate semantic metadata that makes XML Schema
methodology (García, 2006). semantics explicit. Both steps are detailed next.
The main caveat of semantic multimedia
metadata is that it is sparse and expensive to XSD2OWL Mapping
produce. If we want to increase the availability
of semantic multimedia metadata and, in general, The XML Schema to OWL mapping is responsible
of semantic metadata, the more direct solution is for capturing the schema implicit semantics. This
to benefit from the great amount of metadata that semantics are determined by the combination of

1009
Semantic Web for Media Convergence

Table 1. XSD2OWL mappings from XML Schema building blocks to OWL ones plus an explanation of
why they are interpreted as equivalent modelling constructs

XML Schema OWL Explanation


rdf:Property
element | attribute owl:DatatypeProperty Named relation between nodes or nodes and values
owl:ObjectProperty
element@substitutionGroup rdfs:subPropertyOf Relation can appear in place of a more general one
element@type rdfs:range The relation range kind
complexType|group |attributeGroup owl:Class Relations and contextual restrictions package
complexType//element owl:Restriction Contextualised restriction of a relation
extension@base restriction@base rdfs:subClassOf Package concretises the base package
owl:maxCardinality
@maxOccurs @minOccurs Restrict the number of occurrences of a relation
owl:minCardinality
owl:intersectionOf
Sequence choice Combination of relations in a context
owl:unionOf

XML Schema constructs. The mapping is based To conclude, one important aspect is that the
on translating these constructs to the OWL ones resulting OWL ontology may be OWL-Full de-
that best capture their semantics. These transla- pending on the input XML Schema. This is due
tions are detailed in Table 1. to the fact that, in some cases, the XSD2OWL
The XSD2OWL mapping is quite transparent translator must employ rdf:Property for those
and captures a great part XML Schema semantics. xsd:elements that have both data type and object
The same names used for XML constructs are used type ranges.
for OWL ones, although in the new namespace
defined for the ontology. Therefore, XSD2OWL XML2RDF Mapping
produces OWL ontologies that make explicit the
semantics of the corresponding XML Schemas. Once all the metadata XML Schemas are available
The only caveats are the implicit order conveyed by as mapped OWL ontologies, it is time to map the
xsd:sequence and the exclusivity of xsd:choice. XML metadata that instantiates them. The inten-
For the first problem, owl:intersectionOf does tion is to produce RDF metadata as transparently
not retain its operands order, there is no clear as possible. Therefore, a structure-mapping ap-
solution that retains the great level of transpar- proach has been selected (Klein, 2002). It is also
ency that has been achieved. The use of RDF possible to take a model-mapping approach (Tous,
Lists might impose order but introduces ad-hoc García, Rodríguez, & Delgado, 2005).
constructs not present in the original metadata. XML model-mapping is based on representing
Moreover, as has been demonstrated in practise, the XML information set using semantic tools.
the element ordering does not contribute much This approach is better when XML metadata is
from a semantic point of view. For the second semantically exploited for concrete purposes.
problem, owl:unionOf is an inclusive union, the However, when the objective is semantic metadata
solution is to use the disjointness OWL construct, that can be easily integrated, it is better to take a
owl:disjointWith, between all union operands in more transparent approach.
order to make it exclusive. Transparency is achieved in structure-mapping
models because they only try to represent the XML

1010
Semantic Web for Media Convergence

metadata structure, i.e. a tree, using RDF. The different ontologies or semantic rules, are auto-
RDF model is based on the graph so it is easy to matically propagated to instance metadata due
model a tree using it. Moreover, we do not need to inference.
to worry about the semantics loose produced by However, before continuing to the next section,
structure-mapping. We have formalised the under- it is important to point out that these mappings
lying semantics into the corresponding ontologies have been validated in different ways. First, we
and we will attach them to RDF metadata using have used OWL validators in order to check the
the instantiation relation rdf:type. resulting ontologies, not just the MPEG-7 Ontol-
The structure-mapping is based on trans- ogy but also many others (García, Gil, & Delgado,
lating XML metadata instances to RDF ones 2007; García, Gil, Gallego, & Delgado, 2005).
that instantiate the corresponding constructs in Second, our MPEG-7 ontology has been compared
OWL. The more basic translation is between to Hunter’s (2001) and Tsinaraki’s ones (2004).
relation instances, from xsd:elements and Both ontologies, Hunter’s and Tsinaraki’s,
xsd:attributes to rdf:Properties. Concretely, provide a partial mapping of MPEG-7 to Web
owl:ObjectProperties for node to node relations ontologies. The former concentrates on the kinds
and owl:DatatypeProperties for node to values of content defined by MPEG-7 and the latter on
relations. two parts of MPEG-7, the Multimedia Descrip-
However, in some cases, it would be necessary tion Schemes (MDS) and the Visual metadata
to use rdf:Properties for xsd:elements that have structures. It has been tested that they constitute
both data type and object type values. Values subsets of the ontology that we propose.
are kept during the translation as simple types Finally, the XSD2OWL and XML2RDF map-
and RDF blank nodes are introduced in the RDF pings have been tested in conjunction. Testing
model in order to serve as source and destination XML instances have been mapped to RDF, guided
for properties. They will remain blank for the by the corresponding OWL ontologies from the
moment until they are enriched with semantic used XML Schemas, and then back to XML. Then,
information. the original and derived XML instances have been
The resulting RDF graph model contains all compared using their canonical version in order
that we can obtain from the XML tree. It is al- to correct mapping problems.
ready semantically enriched due to the rdf:type
relation that connects each RDF properties to the Ontological Infrastructure
owl:ObjectProperty or owl:DatatypeProperty
it instantiates. It can be enriched further if the As a result of applying the XML Semantics Reuse
blank nodes are related to the owl:Class that methodology, we have obtained a set of ontolo-
defines the package of properties and associated gies that reuse the semantics of the underlying
restrictions they contain, i.e. the corresponding standards, as they are formalised through the
xsd:complexType. This semantic decoration of the corresponding XML Schemas. All the ontologies
graph is formalised using rdf:type relations from related to journalism standards, i.e. NewsCodes
blank nodes to the corresponding OWL classes. NITF and NewsML, are available from the Se-
At this point we have obtained a semantics- mantic Newspaper site8. This site also contains
enabled representation of the input metadata. some of the ontologies for the MPEG-21 useful
The instantiation relations can now be used to for news modelling as convergent multimedia
apply OWL semantics to metadata. Therefore, units. The MPEG-7 Ontology is available from
the semantics derived from further enrichments the MPEG-7 Ontology site9. These are the ontolo-
of the ontologies, e.g. integration links between

1011
Semantic Web for Media Convergence

gies that are going to be used as the basis for the OWL ontology, generated by XSD2OWL, consti-
semantic newspaper info-structure: tutes the basic ontological framework for semantic
multimedia metadata integration and appears at
• NewsCodes subjects ontology: An OWL the centre of the architecture. In parallel, there
ontology for the subjects’ part of the IPTC are the journalism ontologies. The multimedia
NewsCodes. It is a simple taxonomy of related concepts from the journalism ontologies
subjects but it is implemented with OWL are connected to the MPEG-7 ontology, which
in order to facilitate the integration of the acts as an upper ontology for multimedia. Other
subjects’ taxonomy in the global ontologi- ontologies and XML Schemas can also be easily
cal framework. incorporated using the XSD2OWL module.
• NITF 3.3 ontology: An OWL ontology Semantic metadata can be directly fed into
that captures the semantics of the XML the system together with XML metadata, which
Schema specification of the NITF standard. is made semantic using the XML2RDF module.
It contains some classes and many proper- For instance, XML MPEG-7 metadata has a great
ties dealing with document structure, i.e. importance because it is commonly used for low-
paragraphs, subheadlines, etc., but also level visual and audio content descriptors auto-
some metadata properties about copyright, matically extracted from its underlying signals.
authorship, issue dates, etc. This kind of metadata can be used as the basis for
• NewsML 1.2 ontology: The OWL ontol- audio and video description and retrieval.
ogy resulting from mapping the NewsML In addition to content-based metadata, there
1.2 XML Schema. Basically, it includes a is context-based metadata. This kind of metadata
set of properties useful to define the news higher level and it usually, in this context, related to
structure as a multimedia package, i.e. journalism metadata. It is generated by the system
news envelope, components, items, etc. users (journalist, photographers, cameramen, etc.).
• MPEG-7 ontology: The XSD2OWL map- For instance, there are issue dates, news subjects,
ping has been applied to the MPEG-7 titles, authors, etc.
XML Schemas producing an ontology that This kind of metadata can come directly from
has 2372 classes and 975 properties, which semantic sources but, usually, it is going to come
are targeted towards describing multime- from legacy XML sources based on the standards’
dia at all detail levels, from content based XML Schemas. Therefore, in order to integrate
descriptors to semantic ones. them, they will pass through the XML2RDF
• MPEG-21 digital item ontologies: A digi- component. This component, in conjunction
tal item (DI) is defined as the fundamen- with the ontologies previously mapped from
tal unit for distribution and transaction in the corresponding XML Schemas, generates the
MPEG-21. RDF metadata that can be then integrated in the
common RDF framework.
System Architecture This framework has the persistence support
of a RDF store, where metadata and ontologies
Based on the previous XML world to Semantic reside. Once all metadata has been put together,
Web domain mappings, we have built up a system the semantic integration can take place, as shown
architecture that facilitates journalism and mul- in the next section.
timedia metadata integration and retrieval. The
architecture is sketched in Figure 2. The MPEG-7

1012
Semantic Web for Media Convergence

Figure 2. News metadata integration and retrieval architecture

Semantic Integration Outline These relationships capture the semantics of the


data integration. Then, once metadata is incorpo-
As mentioned in the introduction, one of the main rated into the system and semantically-decorated,
problems in nowadays media houses is that of the integration is automatically performed by
heterogeneous data integration. Even within a applying inference. Table 2 shows some of these
single organization, data from disparate sources mappings, performed once all metadata has been
must be integrated. Our approach to solve this moved to the semantic space.
problem is based on Web ontologies and, as the First, there are four examples of semantic map-
focus is on multimedia and journalism metadata pings among the NITF Ontology, the NewsML On-
integration, our integration base are the MPEG-7, tology and the IPTC Subjects Ontology. The first
MPEG-21 and the journalism ontologies. mapping tells that all values for the nitf:tobject.
In order to benefit from the system architecture subject property are from class subj:Subject.
presented before, when semantic metadata based The second one that the property nitf:tobject.
on different schemes has to be integrated, the subject.detail is equivalent to subj:explanation.
XML Schemas are first mapped to OWL. Once The third one that all nitf:body instances are also
this first step has been done, these schemas can be newsml:DataContent instances and the fourth
integrated into the ontological framework using one that all newsml:Subject are subj:Subject.
OWL semantic relations for equivalence and in- Finally, there is also a mapping that is performed
clusion: subClassOf, subPropertyOf, equivalent- during the XML to RDF translation. It is neces-
Class, equivalentProperty, sameIndividualAs, etc. sary in order to recognise an implicit identifier,
These relations allows simple integration relations, nitf:tobject.subject.refnum is mapped to rdf:ID
for more complex integration steps that require in order to make this recognise this identifier in
changes in data structures it is possible to use the context of NITF and make it explicit in the
Semantic Web rules (Horrocks, Patel-Schneider, context of RDF.
Boley, Tabet, Grosof, & Dean, 2004).

1013
Semantic Web for Media Convergence

Table 2. Journalism and multimedia metadata integration mapping examples

Semantic Mappings
∀ nitf:tobject.subject . subj:Subject
nitf:tobject.subject.detail ≡ subj:explanation
nitf:body ⊆ newsml:DataContent
newsml:Subject ≡ subj:Subject
XML2RDF Mappings
nitf:tobject.subject.refnum → rdf:ID

Semantic Media Integration On the other end, IR is addressed in the


from Human Speech Semantic Web field from a much more formal
perspective (Castells, Fernández, & Vallet, 2007).
This section introduces a tool, build on top of the In the Semantic Web vision, the search space
ontological infrastructure described in the previ- consists of a totally formalized corpus, where all
ous sections, geared towards a convergent and the information units are unambiguously typed,
integrated news management in the context of a interrelated, and described by logic axioms in
media house. As has been previously introduced, domain ontologies. Such tools enabled the de-
the diversification of content in media houses, who velopment of semantic-based retrieval technolo-
must deal in an integrated way with different mo- gies that support search by meanings rather than
dalities (text, image, graphics, video, audio, etc.), keywords, providing users with more powerful
carries new management challenges. Semantic retrieval capabilities to find their way through in
metadata and ontologies are a key facilitator in increasingly massive search spaces.
order to enable convergent and integrated media Semantic Web based news annotation and
management. retrieval has already been applied in the Diari
In the news domain, news companies like the Segre Media Group in the context of the Neptuno
Diari Segre Media Group are turning into news research project (Castells, Perdrix, Pulido, Rico,
media houses, owning radio stations and video Benjamins, Contreras, & Lorés, 2004). However,
production companies that produce content not this is a partial solution as it just deals with textual
supported by the print medium, but which can content. The objective of the tool described in
be delivered through Internet newspapers. Such this section is to show how these techniques can
new perspectives in the area of digital content call also be applied to content with embedded human-
for a revision of mainstream search and retrieval speech tracks. The final result is a tool based on
technologies currently oriented to text and based Semantic Web technologies and methodologies
on keywords. The main limitation of mainstream that allows managing text and audiovisual content
text IR systems is that their ability to represent in an integrated and efficient way. Consequently,
meanings is based on counting word occurrences, the integration of human speech processing tech-
regardless of the relation between words (Salton, nologies in the semantic-based approach extends
& McGill, 1983). Most research beyond this the semantic retrieval capabilities to audio content.
limitation has remained in the scope of linguistic The research is being undertaken in the context
(Salton, & McGill, 1983) or statistic (Vorhees, of the S5T research project10.
1994) information. As shown in Figure 3, this tool is based on
a human speech recognition process inspired

1014
Semantic Web for Media Convergence

Figure 3. Architecture for the Semantic Media Integration from Human Speech Tool

by (Kim, Jung, & Chung, 2004) that generates journalists and archivist, are shown on the left of
the corresponding transcripts for the radio and Figure 3. They exploit the semantic richness of
television contents. From this preliminary pro- the underlying ontologies upon which the search
cess, it is possible benefit from the same semi- system is built. Semantic queries are resolved,
automatic annotation process in order to generate using semantic annotations as has been previously
the semantic annotations for audio, audiovisual described, and retrieve content items and pieces
and textual content. Keywords detected during of these contents. News contents are packaged
speech recognition are mapped to concepts in together using annotations based on the MPEG-21
the ontologies describing the domain covered and MPEG-7 ontologies, as it is described in Sec-
by audiovisual and textual content, for instance tion 3.3.1. Content items are presented to the user
the politics domain for news talking about this through the Media Browser, detailed in Section
subject. Specifically, when the keyword forms 3.3.2, and the underlying semantic annotations and
of a concept are uttered in a piece of speech, the the ontologies used to generate these annotations
content is annotated with that concept. Polysemic can be browsed using the Knowledge Browser,
words and other ambiguities are treated by a set of described in Section 3.3.3.
heuristics. More details about the annotation and
semantic query resolution processes are available Semantic News Packaging
from (Cuayahuitl, & Serridge, 2002). Using MPEG Ontologies
Once audio and textual contents have been
semantically annotated (Tejedor, García, Fernán- Actually, in an editorial office there are a lot of
dez, López, Perdrix, Macías, et al., 2007), it is applications producing media in several formats.
possible to provide a unified set of interfaces, This is an issue that requires a common structure
rooted on the semantic capabilities provided by to facilitate management. The first step is to treat
the annotations. These interfaces, intended for each unit of information, in this case each new,

1015
Semantic Web for Media Convergence

Figure 4. Content DI structure

as a single object. Consequently, when searching VideoSegmentType…) because all of them are
something upon this structure, all related content formalised as subclasses of SegmentType and
is retrieved together. the implicit semantics can be directly used by a
Another interesting issue is that news can be semantic query engine.
linked to other news. This link between news al- Table 3 shows a piece of metadata that describes
lows the creation of information threads. A news an audio segment of a Diari Segre Media Group
composition metadata system has been developed news item used in the S5T project. This semantic
using concepts from the MPEG-21 and MPEG-7 metadata is generated from the corresponding
ontologies. It comprises three hierarchical levels XML MPEG-7 metadata using the XML to RDF
as shown in Figure 4. mapping and takes profit from the MPEG-7 OWL
The lower level comprises content files, in ontology in order to make the MPEG-7 semantics
whatever format they are. The mid level is formed explicit. Therefore, this kind of metadata can be
by metadata descriptors (what, when, where, how, processed using semantic queries independently
who is involved, author, etc.) for each file, mainly from the concrete type of segment. Consequently,
based on concepts from the MPEG-7 ontology it is possible to develop applications that process
generated using the methodology described in in an integrated and convergent way the different
Section 3.1. They are called the Media Digital kinds of contents that build up a new.
Items (Media DI). The top level in the hierarchy is based on
These semantic descriptors are based on the descriptors that model news and put together all
MPEG-7 Ontology and facilitate automated the different pieces of content that conform them.
management of the different kinds of content that These objects are called News Digital Items (News
build up a news item in a convergent media house. DI). There is one News DI for each news item
For instance, it is possible to generate semantic and all of them are based on MPEG-21 meta-
queries that benefit from the content hierarchy data. The part of the standard that defines digital
defined in MPEG-7 and formalised in the ontology. items (DI) is used for that. DI is the fundamental
This way, it is possible to pose generic queries unit defined in MPEG-21 for content distribu-
for any kind of segment (e.g. AudioSegmentType, tion and transaction, very useful for convergent

1016
Semantic Web for Media Convergence

Table 3. MPEG-7 Ontology description for a audio segment generated from XML MPEG-7 metadata
fragment

<?xml version=”1.0”?>
<rdf:RDF
xmlns:mpeg7=”http://rhizomik.net/ontologies/2006/03/Mpeg7-2001.owl#”>
<mpeg7:AudioType rdf:about=”http://rhizomik.net/audio/2007-01-13.mp3”>
<mpeg7:Audio>
<mpeg7:AudioSegmentType>
<mpeg7:MediaTime>
<mpeg7:MediaTimeType>
<mpeg7:MediaTimePoint
rdf:datatype=”&xsd;time”>01:27.0</mpeg7:MediaTimePoint>
<mpeg7:MediaDuration
rdf:datatype=”&xsd;time”>P5S</mpeg7:MediaDuration>
</mpeg7:MediaTimeType>
</mpeg7:MediaTime>
</mpeg7:AudioSegmentType>
</mpeg7:Audio>
</mpeg7:AudioType>
</rdf:RDF>

media management. As in the case of MPEG-7 descriptions are based on a generic rendering of
metadata, RDF semantic metadata is generated RDF data as interactive HTML for increased us-
from XML using the semantics made explicit by ability (García, & Gil, 2006).
the MPEG-21 ontologies. This way, it is possible The multimedia metadata is based on the
to implement generic processes also at the news Dublin Core schema for editorial metadata and
level using semantic queries. IPTC News Codes for subjects. For content-based
On top of the previous semantic descriptors at metadata, especially the content decomposition de-
the media and news level, it is possible to develop pending on the audio transcript, MPEG-7 metadata
an application for integrated and convergent news is used for media segmentation, as it was shown
management in the media house. The application in Table 3. In addition to the editorial metadata
is based on two specialised interfaces described and the segments decomposition, a specialized
in the next subsections. They benefit from the audiovisual view is presented. This view allows
ontological infrastructure detailed in this chapter, rendering the content, i.e. audio and video, and
which is complemented with ontologies for the interacting with audiovisual content through a
concrete news domain. However, the application click-able version of the audio transcript.
remains independent from the concrete domain. Two kinds of interactions are possible from the
transcript. First, it is possible to click any word
Media Browser in the transcript that has been indexed in order to
perform a keyword-based query for all content in
The Media Browser, shown in Figure 5, takes the database where that keyword appears. Second,
profit from the MPEG-21 metadata for news and the transcript is enriched with links to the ontol-
MPEG-7 metadata for media in order to imple- ogy used for semantic annotation. Each word in
ment a generic browser for the different kinds of the transcript whose meaning is represented by
media that constitute a news item in a convergent an ontology concept is linked to a description of
newspaper. This interface allows navigating them that concept, which is shown by the Knowledge
and presents the retrieved pieces of content and the Browser detailed in the next section. The whole
available RDF metadata describing them. These interaction is performed through the user Web

1017
Semantic Web for Media Convergence

Figure 5. Media Browser interface presenting content metadata (left) and the annotated transcript
(right)

browser using AJAX in order to improve the retrieved, for instance a list of all the members
interactive capabilities of the interface. of the parliament. In addition to this recursive
For instance, the transcript includes the name navigation of all the domain knowledge, at any
of a politician that has been indexed and modelled browsing step, it is also possible to get all the
in the ontology. Consequently, it can be clicked multimedia content annotated using the concept
in order to get all the multimedia content where currently being browsed. This step would carry
the name appears or, alternatively, to browse all the user back to the Media Browser.
the knowledge about that politician encoded in Thanks to this dual browsing experience, the
the corresponding domain ontology. user can navigate through audiovisual content us-
ing the Media Browser and through the underlying
Knowledge Browser semantic models using the Knowledge Browser in
a complementary an inter-weaved way. Finally, as
This interface is used to allow the user browsing for the Media Browser, the Knowledge Browser
the knowledge structures employed to annotate is also implemented using AJAX so the whole
content, i.e. the underlying ontologies. The same interactive experience can be enjoyed using a
RDF data to interactive HTML rendering used in Web browser.
the Media Browser is used here. Consequently,
following the politician example in the previous
section, when the user looks for the available Alternatives
knowledge about that person and interactive view
of the RDF data modelling him is shown. This way, There are other existing initiatives that try to
the user can benefit from the modelling effort and, move journalism and multimedia metadata to
for instance, be aware of the politician party, that the Semantic Web world. In the journalism field,
he is a member of the parliament, etc. the Neptuno (Castells, Perdrix, Pulido, Rico,
This interface constitutes a knowledge browser Benjamins, Contreras, et al., 2004) and NEWS
so the link to the politician party or the parliament (Fernández, Blázquez, Fisteus, Sánchez, Sintek,
can be followed and additional knowledge can be Bernardi, et al., 2006) projects can be highlighted.

1018
Semantic Web for Media Convergence

Both projects have developed ontologies based on ing XML metadata based on the corresponding
existing standards (IPTC SRS, NITF or NewsML) standards to semantic metadata. Consequently, it
but from an ad-hoc and limited point of view. is difficult to put them into practice as there is a
Therefore, in order to smooth the transition from lack of metadata to play with. On the other hand,
the previous legacy systems, more complex and there is a great amount of existing XML metadata
complete mappings should be developed and and a lot of tools based on XML technologies. For
maintained. example, the new Milenium Quay11 cross-media
The same can be said for the existing at- archive system from PROTEC, the worldwide
tempts to produce semantic multimedia meta- leadership in cross-media software platforms, is
data. Chronologically, the first attempts to make XML-based. This software is focused on flex-
MPEG-7 metadata semantics explicit where ibility using several XML tags and mappings,
carried out, during the MPEG-7 standardisation increasing interoperability with other archiving
process, by Jane Hunter (2001). The proposal used systems. The XML-based products are clearly a
RDF to formalise a small part of MPEG-7, and trend in this scope. Every day, new products from
later incorporated some DAML+OIL construct the main software companies are appearing, which
to further detail their semantics (Hunter, 2001). deal with different steps in all the news life-cycle,
More recent approaches (Hausenblas, 2007) are from production to consumption.
based on the Web Ontology Language (McGuin- Nowadays, commercial tools based on XML
ness & Harmelen, 2004), but are also constrained technologies constitute the clear option in news-
to a part of the whole MPEG-7 standard, the paper media houses. Current initiatives based on
Multimedia Description Scheme (MDS) for the Semantic Web tools are constrained due to the
ontology proposed at (Tsinaraki, Polydoros, & lack of “real” data to work with; they constitute
Christodoulakis, 2004). a too abrupt breaking from legacy systems. More-
An alternative to standards-based metadata are over, they are prototypes with little functionality.
folksonomies (Vanderwal, 2007). Mainly used Consequently, we do not see the semantic tools
in social bookmarking software (e.g. del.icio.us, as an alternative to legacy systems, at least in
Flickr, YouTube), they allow the easy creation the short term. On the contrary, we think that
of user driven vocabularies in order to annotate they constitute additional modules that can help
resources. The main advantage of folksonomies dealing with the extra requirements derived from
is the low entry barrier: all terms are acceptable media heterogeneity, multichannel distribution
as metada, so no knowledge of the established and knowledge management issues.
standards is needed. Its main drawback is the lack The proposed methodology facilitates the
of control over the vocabulary used to annotate production of semantic metadata from existing
resources, so resource combination and reason- legacy systems, although it is simple metadata as
ing becomes almost impossible. Some systems the source is XML metadata that is not intended
combine social and semantic metadata and try to for carrying complex semantics. In any case, it
infer a formal ontology from the tags used in the constitutes a first and smooth step toward adding
folksonomy (Herzog, Luger & Herzog, 2007). In semantic-enabled tools to existing newspaper
our case we believe that it is better to use standard content management systems. From this point,
ontologies both from multimedia and journalism more complex semantics and processing can
fields than open and uncontrolled vocabularies. be added without breaking continuity with the
Moreover, none of the proposed ontologies, for investments that media houses have done in their
journalism of multimedia metadata, is accompa- current systems.
nied by a methodology that allows mapping exist-

1019
Semantic Web for Media Convergence

Cost and Benefits which is reflected in the way archivists categorise


content and journalist perform queries.
One of the biggest challenges in media houses This gap is a clear threat to productivity, al-
is to attach metadata to all the generated content though the flexibility of semantic structures makes
in order to facilitate management. However, this it possible to relate concepts from different mental
is easier in this context as in many media houses models in order to attain a more integrated and
there is a department specialized in this work, shared view (Abelló, García, Gil, Oliva, & Perdrix,
which is carried out by archivists. Consequently, 2006), which improves the content retrieval results
the additional costs arising from the application and consequently improves productivity.
of Semantic Web technologies are mitigated due Moreover, the combination of semantic meta-
to the existence of this department. It is already in data and ontologies, together with tools like the
charge of indexation, categorization and content ones presented for project S5T, make it possible
semantic enrichment. for journalists to navigate between content meta-
Consequently, though there are many organi- data and ontology concepts and benefit from an
zational and philosophy changes that modify how integrated and shared knowledge management
this task is currently carried out, it is not necessary effort. This feature mitigates current gaps among
to add new resources to perform this effort. The editorial staff that seriously reduce the possibilities
volume of information is another important aspect of media production.
to consider. All Semantic Web approaches in this Another point of interest is the possibility that
field propose an automatic or semi-automatic an- journalists produce some metadata during the
notation processes. content generation process. Nowadays, journal-
The degree of automation attained using Se- ists do not consider this activity part of their job.
mantic Web tools allows archivists spending less Consequently, this task might introduce additional
time in the more time consuming and mechanical costs that have not been faced at the current stage
tasks, e.g. the annotation of audio contents which of development. This remains a future issue that
can be performed with the help of speech-to-text requires deep organisational changes, which are
tools as in the S5T project example presented in not present yet in most editorial staffs, even if
Section 3.3. Consequently, archivists can spend they are trying to follow the media convergence
their time refining more concrete and specific philosophy.
metadata details and leave other aspects like To conclude, there are also the development
categorization or annotation to partially or totally costs necessary in order to integrate the Semantic
automatic tools. The overall outcome is that, with Web tools into current media houses. As has been
this computer and human complementary work, already noted, the choice of a smooth transition
it is possible to archive big amounts of content approach reduces the development costs. This ap-
without introducing extra costs. proach is based on the XSD2OWL and XML2RDF
Semantic metadata also provides improve- mappings detailed in Section 3.1.
ments in content navigability and searching, maybe Consequently, it is not necessary to develop a
in all information retrieval tasks. This fact implies a full newspaper content management system based
better level of productivity in the media house, e.g. on Semantic Web tools. On the contrary, existing
while performing event tracking through a set of systems based on XML technologies, as it is the
news in order to produce a new content. However, common case, are used as the development plat-
it is also important to take into account the gap form on top of which semantic tools are deployed.
between journalists’and archivists’mental models, This approach also improves interoperability with
other media houses that also use XML technolo-

1020
Semantic Web for Media Convergence

gies, though the interoperation is performed at research project (Castells et al., 2004) also in the
the semantic level once source metadata has been journalism domain.
mapped to semantic metadata. There is a lack of implementations supporting
massive content storage and management. In other
words, experimental solutions cannot be applied
Risk Assessment to real system considering, as our experience has
shown, more than 1 million of items, i.e. news,
In one hand we can consider some relevant posi- photos or videos. This amount can be generated
tive aspects from the proposed solution. In fact, in 2 or 3 months in a small news media company.
we are introducing knowledge management A part from the lack of implementations, there is
into the newspaper content archive system. The also the lack of technical staff with Semantic Web
proposal implies a more flexible archive system development skills.
with significant improvements in search and Despite all these inconveniences, there is
navigation. Compatibility with current standards the opportunity to create a platform for media
is kept while the archive system allows search- convergence and editorial staff tasks integration.
ing across media and the underlying terms and It can become an open platform that can manage
domain knowledge. Finally, the integrated view future challenges in media houses and that is
on content provides seamless access to any kind adaptable to different models based on specific
of archived resources, which could be text, audio, organizational structures. Moreover, this platform
video streaming, photographs, etc. Consequently, may make it possible to offer new content inter-
separate search engines for each kind of media action paradigms, especially through the World
are no longer necessary and global queries make Wide Web channel.
it possible to retrieve any kind of resources. One of these potential paradigms has already
This feature represents an important improve- started to be explored in the S5T project. Currently,
ment in the retrieval process but also in the ar- it offers integrated and complementary browsing
chiving one. The introduction of a semi-automatic among content and the terms of the underlying
annotation process produces changes in the archi- domain of knowledge, e.g. politics. However,
vist work. They could expend more time refining this tool is currently intended just for the editorial
semantic annotation and including new metadata. staff. We anticipate a future tool that makes this
Existing human resources in the archive depart- kind of interaction available from the Diari Segre
ment should spend the same amount of time than Web site to all of its Web users. This tool would
they currently do. However, they should obtain provide an integrated access point to different
better quality results while they populate the ar- kinds of contents, like text or news podcasts, but
chive with all the semantically annotated content. also to the underlying knowledge that models
The overall result is that the archive becomes a events, histories, personalities, etc.
knowledge management system. There are some threats too. First of all, any
On other hand, we need to take into account organizational change, like changing the way
some weaknesses in this approach. Nowadays, the archive department works or giving unprec-
Semantic Web technologies are mainly prototypes edented annotation responsibilities to journalists,
under development. This implies problems when constitutes an important risk. Changes inside an
you try to build a complete industrial platform organization never be easy and must be well done
based on them. Scalability appears as the main and follow very closely if you want to make them
problem as it was experienced during the Neptuno successful. Sometimes, the effort-satisfaction ratio
may be perceived as not justified by for some

1021
Semantic Web for Media Convergence

journalist or archivists. Consequently, they may Personalisation ranges from interfaces, to pro-
react against the organisational changes required cesses or query construction approaches applying
in order to implement rich semantic metadata. static or dynamic profiles. Static profiles could
be completed by users in when they first register.
Dynamic profiles must be collected by the system
Future Trends based on the user system usage (Castells et al.,
2007). Per user profiles introduce a great amount
The more relevant future trend is that the Semantic of complexity, which can be mitigated building
Web is starting to be recognised as a consolidated groups of similar profiles, for instance groups
discipline and a set of technologies and methodolo- based on the user role.
gies that are going to have a great impact in the Moreover, to collect system usage informa-
future of enterprise information systems (King, tion while users navigate through the underlying
2007). The more important consequence of this conceptual structures makes it possible to discover
consolidation is that many commercial tools are new implicit relations among concepts with some
appearing. They are solid tools that can be used semantic significance, at least from the user, or
in order to build enterprise semantic information group to which the user belongs, point of view.
systems with a high degree of scalability. If there are a lot of users following the same
As has been shown, the benefits of semantic navigation path between items, maybe it would
metadata are being put into practice in the Diari be better to add a new conceptual link between
Segre Media Group, a newspaper that is becom- the initial and final items. Currently, this kind of
ing a convergent media house with press, radio, relations can only be added manually. In the near
television and a World Wide Web portal. As has future, we could use the power of Semantic Web
been detailed, a set of semantics-aware tools have technologies in order to do this automatically.
been developed. They are intended for journalist This would improve user experience while they
and archivists in the media house, but they can be search or navigate as the underlying conceptual
also adapted to the general public at the portal. framework would accommodate the particular
Making the Diari Segre semantic tools publicly user view on the domain.
available is one of the greatest opportunities and To conclude this section, it is also important
in the future, with the help of solid enterprise se- to take into account the evolution of the standards
mantic platforms, is the issue where the greatest upon which the ontological framework has been
effort is going to be placed. In general, a bigger build. On the short range, the most import nov-
users base puts extra requirements about the elty is the imminent release of the NewsML G2
particular needs that each user might have. This standard (Le Meur, 2007). This standard is also
is due to the fact that each user may have a dif- based on XML Schemas for language formalisa-
ferent vision about the domain of knowledge or tion. Therefore, it should be trivial to generate the
about searching and browsing strategies. In this corresponding OWL ontologies and to start map-
sense, we need some degree of personalisation ping metadata based on this standard to semantic
beyond the much more closed approach that has metadata. More effort will be needed in order to
been taken in order to deploy these tools for the produce the integration rules that will allow inte-
editorial staff. grating this standard into existing legacy systems
augmented by Semantic Web tools.

1022
Semantic Web for Media Convergence

Conclusion References

This research work has been guided by the need Abelló, A., García, R., Gil, R., Oliva, M., &
for a semantic journalism and multimedia metadata Perdrix, F. (2006). Semantic Data Integration in
framework that facilitates semantic newspaper ap- a Newspaper Content Management System. In
plications development in the context of a convergent R. Meersman, Z. Tari, & P. Herrero (Eds.), OTM
media house. It has been detected, as it is widely Workshops 2006. LNCS Vol. 4277 (pp. 40-41).
documented in the bibliography and professional Berlin/Heidelberg, DE: Springer.
activity, that IPTC and MPEG standards are the best Amann, B., Beer, C., Fundulak, I., & Scholl, M.
sources for an ontological framework that facilitates (2002). Ontology-Based Integration of XML Web
a smooth transition from legacy to semantic informa- Resources. Proceedings of the 1st International
tion systems. MPEG-7, MPEG-21 and most of the Semantic Web Conference, ISWC 2002. LNCS
IPTC standards are based on XML Schemas and Vol. 2342 (pp. 117-131). Berlin/Heidelberg, DE:
thus they do not have formal semantics. Springer.
Our approach contributes a complete and auto-
matic mapping of the whole MPEG-7 standard to Becket, D. (2004). RDF/XML Syntax Specifi-
OWL, of the media packaging part of MPEG-21 cation. World Wide Web Consortium Recom-
and of the main IPTC standard schemas (NITF, mendation.
NewsML and NewsCodes) to the corresponding
Berners-Lee, T., Hendler, J., & Lassila, O. (2001).
OWL ontologies. Instance metadata is automati-
The Semantic Web. Scientific American, 284(5),
cally imported from legacy systems through a XML-
34–43.
2RDF mapping, based on the ontologies previously
mapped from the standards XML schemas. Once Castells, P., Fernández, M., & Vallet, D. (2007). An
in a semantic space, data integration, which is a Adaptation of the Vector-Space Model for Ontol-
crucial factor when several sources of information ogy-Based Information Retrieval. IEEE Transac-
are available, is facilitated enormously. tions on Knowledge and Data Engineering, 19(2),
Moreover, semantic metadata facilitates the de- 261–272. doi:10.1109/TKDE.2007.22
velopment of applications in the context of media
Castells, P., Perdrix, F., Pulido, E., Rico, M., Ben-
houses that traditional newspapers are becoming.
jamins, R., Contreras, J., et al. (2004). Neptuno:
The convergence of different kinds of media,
Semantic Web Technologies for a Digital News-
that now constitute multimedia news, poses new
paper Archive. In C. Bussler, J. Davies, D. Fensel,
management requirements that are easier to cope
& R. Studer, (Eds.), The Semantic Web: Research
with if applications are more informed, i.e. aware
and Applications: First European Semantic Web
of the semantics that are implicit in news and the
Symposium, ESWS 2004, LNCS Vol. 3053 (pp.
media that constitute them. This is the case for
445-458). Berlin/Heidelberg, DE: Springer.
the tools we propose for archivists and journalists,
the Media Browser and the Knowledge Browser. Castells, P., Perdrix, F., Pulido, E., Rico, M.,
These tools reduce the misunderstandings among Benjamins, R., Contreras, J., & Lorés, J. (2004).
them and facilitate keeping track of existing news Neptuno: Semantic Web Technologies for a Digital
stories and the generation of new content. Newspaper Archive. LNCS Vol. 3053 (pp. 445-
458).Berlin/Heidelberg, DE: Springer.

1023
Semantic Web for Media Convergence

Cruz, I., Xiao, H., & Hsu, F. (2004). An Ontology- García, R., Gil, R., Gallego, I., & Delgado, J.
based Framework for XML Semantic Integration. (2005). Formalising ODRL Semantics using Web
Proceedings of the Eighth International Data- Ontologies. In R. Iannella, S. Guth, & C. Serrao,
base Engineering and Applications Symposium, Eds., Open Digital Rights Language Workshop,
IDEAS’04, (pp. 217-226). Washington, DC: IEEE ODRL’2005 (pp. 33-42). Lisbon, Portugal: AD-
Computer Society. ETTI.
Cuayahuitl, H., & Serridge, B. (2002). Out-of- Hausenblas, M., Troncy, R., Halaschek-Wiener,
vocabulary Word Modelling and Rejection for C., Bürger, T., & Celma, O. Boll, et al. (2007).
Spanish Keyword Spotting Systems. Proceedings Multimedia Vocabularies on the Semantic Web.
of the 2nd Mexican International Conference on W3C Incubator Group Report, World Wide Web
Artificial Intelligence. Consortium.
Einhoff, M., Casademont, J., Perdrix, F., & Haustein, S., & Pleumann, J. (2002). Is Participa-
Noll, S. (2005) ELIN: A MPEG Related News tion in the Semantic Web Too Difficult? In Pro-
Framework. In M. Grgic (Ed.), 47th International ceedings of the First International Semantic Web
Symposium ELMAR: Focused on Multimedia Conference on The Semantic Web, LNCS Vol. 2342
Systems and Applications (pp.139-142). Zadar, (pp. 448-453). Berlin/Heidelberg: Springer.
Croatia: ELMAR.
Herzog, C., Luger, M., & Herzog, M. (2007).
Eriksen, L. B., & Ihlström, C. (2000). Evolution Combining Social and Semantic Metadata for
of the Web News Genre - The Slow Move Beyond Search in Document Repository.Bridging the Gap
the Print Metaphor. In Proceedings of the 33rd Between Semantic Web and Web 2.0. International
Hawaii international Conference on System Sci- Workshop at the 4th European Semantic Web Con-
ences. IEEE Computer Society Press. ference in Insbruck, Austria, June 7, 2007.
Fernández, N., Blázquez, J. M., Fisteus, J. A., Horrocks, I., Patel-Schneider, P. F., Boley, H.,
Sánchez, L., Sintek, M., Bernardi, A., et al. (2006). Tabet, S., Grosof, B., & Dean, M. (2004). SWRL:
NEWS: Bringing Semantic Web Technologies A Semantic Web Rule Language Combining OWL
into News Agencies. The Semantic Web - ISWC and RuleML. W3C Member Submission, World
2006, LNCS Vol. 4273 (pp. 778-791). Berlin/ Wide Web Consortium.
Heidelberg, DE: Springer.
Hunter, J. (2001). Adding Multimedia to the
García, R. (2006). XML Semantics Reuse. In A Semantic Web - Building an MPEG-7 Ontology.
Semantic Web Approach to Digital Rights Man- Proceedings of the International Semantic Web
agement, PhD Thesis (pp. 116-120). TDX. Working Symposium (pp. 260-272). Standford,
CA.
García, R., & Gil, R. (2006). Improving Human-
Semantic Web Interaction: The Rhizomer Expe- Hunter, J. (2003). Enhacing the Semantic Interop-
rience. Proceedings of the 3rd Italian Semantic erability of Multimedia through a Core Ontology.
Web Workshop, SWAP’06, Vol. 201 (pp. 57-64). IEEE Transactions on Circuits and Systems for
CEUR Workshop Proceedings. Video Technology, 13(1), 49–58. doi:10.1109/
TCSVT.2002.808088
García, R., Gil, R., & Delgado, J. (2007). A Web
ontologies framework for digital rights manage-
ment. Artificial Intelligence and Law, 15(2),
137–154. doi:10.1007/s10506-007-9032-6

1024
Semantic Web for Media Convergence

Ihlström, C., Lundberg, J., & Perdrix, F. (2003) Patel-Schneider, P., & Simeon, J. (2002). The
Audience of Local Online Newspapers in Swe- Yin/Yang Web: XML syntax and RDF semantics.
den, Slovakia and Spain - A Comparative Study. Proceedings of the 11th International World Wide
In Proceedings of HCI International Vol. 3 (pp. Web Conference, WWW’02 (pp. 443-453). ACM
749-753). Florence, Kentucky: Lawrence Erlbaum Press.
Associates.
Salembier, P., & Smith, J. (2002). Overview of
Kim, J., Jung, H., & Chung, H. (2004). A Key- MPEG-7 multimedia description schemes and
word Spotting Approach based on Pseudo N-gram schema tools. In B.S. Manjunath, P. Salembier,
Language Model. Proceedings of the 9th Conf. & T. Sikora (Ed.), Introduction to MPEG-7:
on Speech and Computer, SPECOM 2004 (pp. Multimedia Content Description Interface. John
256-259). Patras, Greece. Wiley & Sons.
King, R. (2007, April 29). Taming the World Wide Salton, G., & McGill, M. (1983). Introduction
Web. Special Report, Business Week. to Modern Information Retrieval. New York:
McGraw-Hill.
Klein, M. C. A. (2002). Interpreting XML
Documents via an RDF Schema Ontology. In Sawyer, S., & Tapia, A. (2005). The sociotechni-
Proceedings of the 13th International Workshop cal nature of mobile computing work: Evidence
on Database and Expert Systems Applications, from a study of policing in the United States.
DEXA 2002 (pp. 889-894). Washington, DC: International Journal of Technology and Human
IEEE Computer Society. Interaction, 1(3), 1–14.
Lakshmanan, L., & Sadri, F. (2003). Interoper- Tejedor, J., García, R., Fernández, M., López, F.,
ability on XML Data. Proceedings of the 2nd In- Perdrix, F., Macías, J. A., et al. (2007). Ontology-
ternational Semantic Web Conference, ICSW’03, Based Retrieval of Human Speech. Proceed-
LNCS Vol. 2870 (pp. 146-163). Berlin/Heidelberg: ings of the 6th International Workshop on Web
Springer. Semantics, WebS’07 (in press). IEEE Computer
Society Press.
Le Meur, L. (2007). How NewsML-G2 simplifies
and fuels news management. Presented at XTech Tous, R., García, R., Rodríguez, E., & Delgado,
2007: The Ubiquitous Web, Paris, France. J. (2005). Arquitecture of a Semantic XPath Pro-
cessor. In K. Bauknecht, B. Pröll, & H. Werthner,
Lundberg, J. (2002). The online news genre: Vi-
Eds., E-Commerce and Web Technologies: 6th
sions and state of the art. Paper presented at the
International Conference, EC-Web’05, LNCS
34th Annual Congress of the Nordic Ergonomics
Vol. 3590 (pp. 1-10). Berlin/Heidelberg, DE:
Society, Sweden.
Springer.
McDonald, N. (2004). Can HCI shape the future of
Tsinaraki, C., Polydoros, P., & Christodoulakis, S.
mass communications. Interaction, 11(2), 44–47.
(2004). Integration of OWL ontologies in MPEG-7
doi:10.1145/971258.971272
and TVAnytime compliant Semantic Indexing. In
McGuinness, D. L., & Harmelen, F. V. (2004). A. Persson, & J. Stirna, Eds., 16th International
OWL Web Ontology Language Overview. World Conference on Advanced Information Systems
Wide Web Consortium Recommendation. Engineering, LNCS Vol. 3084 (pp. 398-413).
Berlin/Heidelberg, DE: Springer.

1025
Semantic Web for Media Convergence

Tsinaraki, C., Polydoros, P., & Christodoulakis, Endnotes


S. (2004). Interoperability support for Ontology-
based Video Retrieval Applications. Proceedings
1
http://www.diarisegre.com
of 3rd International Conference on Image and
2
IPTC, http://www.iptc.org
Video Retrieval, CIVR 2004. Dublin, Ireland.
3
IIM, http://www.iptc.org/IIM
4
http://www.iptc.org/NewsCodes
Vanderwal, T. (2007) Folksonomy Coinage and 5
NITF, http://www.nitf.org
Definition. 6
http://www.newsml.org
Vorhees, E. (1994). Query expansion using lexical
7
http://rhizomik.net/redefer
semantic relations. Proceedings of the 17th ACM
8
http://rhizomik.net/semanticnewspaper
Conf. on Research and Development in Informa-
9
http://rhizomik.net/ontologies/mpeg7ontos
tion Retrieval, ACM Press.
10
http://nets.ii.uam.es/s5t
11
Milenium Quay, http://www.mileniumcross-
media.com

Additional Reading

Kompatsiaris, Y., & Hobson, P. (Eds.). (2008).


Semantic Multimedia and Ontologies: Theory and
Applications. Berlin/Heidelberg, DE: Springer.

This work was previously published in Semantic Web for Business: Cases and Applications, edited by R. García, pp. 170-193,
copyright 2009 by Information Science Reference (an imprint of IGI Global).

1026
1027

Chapter 4.9
Applying Semantic Web
to E-Tourism
Danica Damljanović
University of Sheffield, UK

Vladan Devedžić
University of Belgrade, Serbia

Abstract – they enable machine readable and machine


understandable representation of the data and
Traditional E-Tourism applications store data more importantly reasoning.
internally in a form that is not interoperable with
similar systems. Hence, tourist agents spend plenty
of time updating data about vacation packages in INTRODUCTION
order to provide good service to their clients. On
the other hand, their clients spend plenty of time A mandatory step on the way to the desired va-
searching for the ‘perfect’ vacation package as cation destination is usually contacting tourist
the data about tourist offers are not integrated and agencies. Presentations of tourist destinations
are available from different spots on the Web. We on the Web make a huge amount of data. These
developed Travel Guides - a prototype system for data are accessible to individuals through the of-
tourism management to illustrate how semantic ficial presentations of the tourist agencies, cities,
web technologies combined with traditional municipalities, sport alliances, etc. These sites
E-Tourism applications: a.) help integration of are available to everyone, but still, the problem is
tourism sources dispersed on the Web b) enable to find useful information without wasting time.
creating sophisticated user profiles. Maintaining On the other hand, plenty of systems on the Web
quality user profiles enables system personaliza- are maintained regularly to provide tourists with
tion and adaptivity of the content shown to the up-to-date information. These systems require a
user. The core of this system is in ontologies lot of efforts from humans - especially in travel

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Applying Semantic Web to E-Tourism

agencies where they want to offer tourists a good web technologies. In the central section we first
service. discuss problems that are present in existing E-
We present Travel Guides – a prototype system Tourism systems, and then describe how we solve
that is combining Semantic Web technologies some of these problems with Travel Guides: we
with those used in mainstream applications (cp. give details of the design of the domain ontology,
Djuric, Devedzic & Gasevic, 2007) in order to the creation of the knowledge base and finally
enable data exchange between different E-Tourism system architecture. To illustrate Travel Guides
systems and thus: environment we give an example of using this
system by providing some screenshots. Finally,
• Ease the process of maintaining the systems we conclude and give the ideas of future work and
for tourist agencies also future research directions in the field.
• Ease the process of searching for perfect
vacation packages for tourists
BACKGROUND
The core of Travel Guides system is in ontolo-
gies. We have developed domain ontology for E-Tourism comprises electronic services which
tourism and described the most important design include (Aichholzer, Spitzenberger & Winkler,
principles in this chapter. 2003):
As ontologies enable presenting data in a ma-
chine-readable form thus offering easy exchange • Information Services (IS), e.g. destination,
of data between different applications, this would hotel information.
lead to increased interoperability and decreased • Communication Services (CS), e.g. discus-
efforts tourist agents make to update the data in sion forums, blogs.
their systems. To illustrate increased interoper- • Transaction Services (TS), e.g. booking,
ability we initialized our knowledge base using payment.
data imported from some other system. We built
an environment to enable transferring segments Among these three services Information
of any knowledge base to the other by selecting Services are the most present on the Web. Hotels
some criteria - this transfer is possible even if the usually have their Web sites with details about
knowledge bases rely on different ontologies. the type of accommodation, location, and contact
Ontology-aware systems provide the possi- information. Some of these Web sites even offer
bility to perform semantic search – the user can Transaction Services so that it is possible to access
search the destinations covered by Travel Guides the prices and availability of the accommodation
using several criteria related to travelling (e.g., for the requested period and perform booking
accommodation rating, budget, activities and and payment.
interests: concerts, clubbing, art, sports, shopping, Transaction Services are usually concentrated
etc.). For even more sophisticated search results on sites of Web tourist agencies such as Expedia,
we introduce user profiles created based on data Travelocity, Lastminute, etc. These Websites
that system possesses about the user. These data sometimes include Communication Services in the
are analysed by a reasoner, and the heuristics is form of forums where people who visited hotels
residing inside the ontology. give their opinion and reviews. With emerging
The chapter is organized as follows: in next popularity of social web applications many sites
section we describe different systems that are de- specialize in CS only (e.g., www.43places.com).
veloped in the area of tourism which use semantic

1028
Applying Semantic Web to E-Tourism

However, for complete details about a certain language. GATE (General Architecture for Text
destination (e.g., activities, climate, monuments, Engineering) is an infrastructure for developing
and events) one often must search for several sourc- and deploying software components that process
es. Apparently all of these sources are dispersed human language (Cunningham, 2002). It can an-
on different places on the Internet and there is an notate documents by recognizing concepts such
“information gap” between them. The best way to as: locations, persons, organizations and dates. It
bridge this gap would be to enable communication can be extended to annotate some domain-related
between different tourist applications. concepts, such as hotels and beaches.
For Transaction Services this is already The most common approaches for applying
partly achieved by using Web portals that serve Semantic Web in E-Tourism are:
as mediators between tourists and tourist agen-
cies. These portals (e.g., Bookings.com) gather 1. Making applications from scratch using
vacation packages from different vendors and recommended standards
use Web services to perform booking and some- 2. Using ontologies as mediators to merge
times payment. Communication Services are already existing systems
tightly coupled with Information Services, in a 3. Performing annotations in respect to the
way that the integration of the first implies the ontology of already existing Web content
integration of the latter. Henriksson (2005) dis-
cusses that the one of the main reasons for lack One of the first developed E-Tourism systems
of interoperability in the area of tourism is the was onTour (http://ontour.deri.org/) developed by
tourism product itself: immaterial, heterogeneous DERI (Siorpaes & Bachlechner, 2006; Prantner,
and non-persistent. Travel Guides demonstrates 2004) where they built a prototype system from
how Semantic Web technologies can be used scratch and stored their data in the knowledge base
to enable communication between Information created based on the ontology. They developed
Services dispersed on the Web. This would lead domain ontology following the World Tourism
to easier exchange of communication services, Organization standards, although they considered
thus resulting in better quality of E-Tourism and a very limited amount of concepts and relations.
increased interoperability. Later on, they took over the ontology developed
Hepp, Siorpaes and Bachlechner (2006) claim as a part of Harmonize project and now planning
that “Everything is there, but we only have insuf- to develop an advanced E-Tourism Semantic Web
ficient methods of finding and processing what’s portal to connect the customers and virtual travel
already on the Web” (p. 2). This statement re- agents (Jentzsch, 2005).
veals some of the reasons why Semantic Web is The idea of Harmonize project was to in-
not frequently applied in real-time applications: tegrate Semantic Web technologies and merge
Web today contains content understandable to tourist electronic markets yet avoiding forcing
humans hence only humans can analyse it. To tourist agencies to change their already existing
retrieve information from applications using information systems, but to merge them using
computer programs (e.g., intelligent agents) two ontology as a mediator (Dell’erba, Fodor, Hopken,
conditions must be satisfied: 1) data must be in a & Werthner, 2005).
machine-readable form 2) applications must use The third approach is very challenging for
technologies that provide information retrieval researchers as with the current state of the Web
from this kind of data. it is not easy to add semantics to the data with-
Many academic institutions are making efforts out changing the technologies used to develop
to find methods for computer processing of human the Web applications. Cardoso (2006) presents

1029
Applying Semantic Web to E-Tourism

a system that creates vacation packages dy- providing current data about vacation packages
namically using previously annotated data in and destinations. Data about different destinations
respect to the ontology. This is performed with are not static – they change over time and thus
a service that constructs itinerary by combining require E-Tourism systems to be updated. With
user preferences with flights, car rentals, hotel, the current state of the development of E-Tourism
and activities on-fly. In 2005, Cardoso founded applications, each travel agency performs data
a lab for research in the area of Semantic Web update individually.
appliance in E-Tourism. The main project called In Travel Guides we employ Semantic Web
SEED (Semantic E-Tourism Dynamic packaging) technologies by combining the first and the third
aims to illustrate the appliance of Web services approach. We use the first approach to build the
and Semantic Web in the area of tourism. One of core of the system, and to initialize the repository,
the main objectives of this project is the develop- whereas in later phase we propose using annotation
ment of OTIS ontology (Ontology for Tourism tools such as GATE to perform semi-automatic
Information Systems). Although they discuss the annotation of documents and update of knowledge
comprised concepts of this ontology, its develop- base accordingly. Some of the existing Knowledge
ment is not yet finished, and could not be further Management platforms such as KIM (Popov,
discussed in this chapter. Kiryakov, Ognyanoff, Manov & Kirilov, 2004)
On the other side, Hepp et al. (2006) claim use GATE for performing automatic annotation
that there are not enough data in the domain of of documents and knowledge base enrichment.
tourism available on the Web - at least for Tyrol, Due to the very old and well-known problem of
Austria. Their experiment revealed that existing syntactic ambiguity (Church & Patil, 1982) of
data on the Web are incomplete: the availability human language widely present inside the Web
of the accommodation and the prices are very content that is used in the process of annotation, we
often inaccessible. argue that the role of human is irreplaceable.
Additionally, most of E-Tourism portals store The core of the Travel Guides system is in
their data internally, which means that they are ontologies. Many ontologies have been already
not accessible by search engines on the Web. developed in the area of tourism. Bachlechner
Using Semantic Web services, e.g. Web Service (2004) has made a long list of the areas that need
Modelling Ontology - WSMO (Roman et al., 2005) to be covered by E-Tourism relevant ontologies
or OWL-based Web service ontology - OWL-S and made a brief analysis of the developed domain
(Smith & Alesso, 2005) it would be possible to and upper level ontologies. Another good sum-
access data from data-intensive applications. mary of E-Tourism related ontologies is given in
SATINE project is about deploying semantic (Jentzsch, 2005).
travel Web services. In (Dogac et al., 2004) they However, no ontology includes all concepts
present how to exploit semantics through Web and relations between them in such a way that it
service registries. can be used without any modifications, although
Semantic Web services might be a good some of them such as Mondeca’s (http://www.
solution for performing E-Tourism Transaction mondeca.com) or OnTour’s ontology (Prantner,
Services, and also for performing E-Tourism 2004) are developed following World Tourism
Information Services, as they enable integrating Organization standards. While developing Travel
homogenous data and applications. However, us- Guides ontology we tried to comprise all possible
ing Semantic Web services, as they are applied concepts that are related to the area of tourism
nowadays, will not reduce every-day efforts and also - tourists. Concepts and relations that
made by tourist agents who are responsible for describe user’s activities and interests coupled

1030
Applying Semantic Web to E-Tourism

with built-in reasoner enable identifying the user which means that if, for instance, a new hotel
as a particular type: some tourists enjoy comfort is built on a certain destination and one tourist
during vacation, whereas others don’t care about agency updates the repository, all others can use
the type of the accommodation but more about the it immediately.
outdoor activities or the scenery that is nearby. We suggest this approach as Semantic Web
Most of the developed ontology-aware systems technologies nowadays are still weak to handle a
nowadays propose using a RDF repository instead huge amount of data, and could not be compared by
of using conventional databases (Stollberg, Zh- performance with relational databases in the terms
danova & Fensel, 2004). RDF repositories are not of transaction handling, security, optimization and
built to replace conventional databases, but to add a scalability (cf. Guo, Pan & Heflin, 2004).
refinement which is not supported by conventional
databases, specifically – to enable representing
machine-readable data and reasoning. In Travel Applying semantic web to
Guides system we distinguish between data that E-Tourism
are stored in RDF repositories and those that
are stored in conventional databases. In RDF E-Tourism Today
repositories we store machine-understandable
data used in the process of reasoning, and rela- Searching for information on a desired spot for
tional databases are used to store and retrieve all vacation is usually a very time-consuming. Fig-
other data – those that are not important in this ure 1 depicts the most frequent scenario which
process and also being specific for each travel starts with the vague ideas of the user interested
agency which means they are not sharable. We in travelling, and ends with the list of tourist
propose sharable data to be those that could be destinations. In most of cases the user is aware
easily exchangeable between applications. This of a few criteria that should be fulfilled (the dis-
way, applications can share a unique repository tance from the shopping centre, sandy beach, a

Figure 1. The usual scenario of searching the Internet for a ‘perfect’ vacation package

1031
Applying Semantic Web to E-Tourism

possibility to rent a car, etc.), as well as of some avoiding spamming the user with an unattractive
individual constraints (prices, departure times, content.
etc.). After processing the user’s query (using Creating user profiles is widely used in many
these criteria as input data), the search engine of applications nowadays, not only in the area of tour-
a tourist agency will most likely return a list of ism. In order to create his/her profile, the user is
vacation packages. It is up to the user to choose usually prompted to register and fill in few forms
the most appropriate one. If the user is not satisfied with some personal info such as location, year of
with the result, the procedure is repeated, with birth, interests, etc. Filling these forms sometimes
another tourist agency. This scenario is restarted can take a lot of time and thus carries the risk of
in N iterations until the user gets the desired re- ‘refusing’ the user. The best way is to request a
sult. The essential disadvantage of this system is minimum data from the user on his first log in,
a lack of the integrated and ordered collection of and then update his data later step by step.
the tourist deals. Tourist deals are dispersed on
the Web and being offered from different tourist Travel Guides
agencies each of which maintain their system
independently. Disadvantages of traditional E-Tourism systems
Additional problem with existing E-Tourism imply requirements for a new system that is
applications is the lack of interactivity. It is always focusing on the integration of Information Ser-
the user who provides the criteria for the search/ vices present on the Web and also on introduc-
query and who analyzes the results returned. ing sophisticated user profiles. Travel Guides
The problem of dispersed information about prototype system has been developed to satisfy
tourist deals would be reduced totally if all vaca- these requirements by combining semantic web
tion packages would be gathered at one place - the technologies with those used in traditional E-
Web portal. This assumption could not be taken Tourism systems.
as realistic, but apparently the distribution of the Using semantic web technologies enables
tourist offers would be decreased by adding the representing the data in machine-readable form.
tourist offers of each tourist agency into the por- Such a representation enable easier integration
tal. Although the portals are more sophisticated of tourist resources as data exchange between
than simple Web applications, they usually do not applications is feasible. Integration of tourist
compensate for the lack of interactivity. resources would decrease efforts tourist agents
Some of the popular Web tourist agencies, such make in tourist agencies to maintain these data.
as Expedia, expand their communication with The final result would affect the tourist who will
the user by offering various services based on be able to search for details about destinations
user selection during visiting their site. Namely, from the single point on the Web.
they track user actions (mouse clicks) so that In Travel Guides we introduce more sophisti-
when user browse throughout the site the list of cated user profiles – these are to enable person-
user recently visited places is always available. alization of the Web content and to act as agents
In case the user provides his personal e-mail who work for users, while not spamming them
they send some special offers or advertisements with commercial content and advertisements. For
occasionally. Although their intentions are to example, if during registration the user enters that
improve communication with their clients, this he is interested in extreme sports, and later moves
kind of service can be irritating sometimes. De- on to the search form where he does not specify
veloping more sophisticated user profiles would any sport requirement, the return results could be
help developing more personalized systems thus

1032
Applying Semantic Web to E-Tourism

filtered in a manner that the first listed are those Travel Guides Ontologies
that are flagged as “adventurer destinations”.
Developing sophisticated user profiles requires The Travel Guides Ontologies are written in OWL
analyse of the user behaviour while visiting the (Antoniou & Harmelen, 2004) and developed
portal. This behaviour is determined by the data using Protégé (Horridge, Knublauch, Rector, Ste-
the system collects about the user: his personal vens & Wroe, 2004). To develop a well-designed
data, interests, activities, and also the data that ontology, it was important to:
system tracks while ‘observing’ the user: user
selection, mouse clicks, and the like. To be able 1. Include all important terms in the area of
to constantly analyze the user’s profile the portal tourism to represent destinations in gen-
requires intelligent reasoning. To make a tour- eral, excluding data specific for any tourist
ism portal capable of intelligent reasoning, it is agency. For instance, information about a
necessary to build some initial and appropriate city name, its latitude and longitude, and
knowledge in the system, as well as to maintain the country it belongs to is to be included
the knowledge automatically from time to time here.
and during the user’s interaction with the system. 2. Classify user interests and activities so that
Simply saving every single click of the user could they can be expressed in the manner of a
not be enough to make a good-quality user profile. collection of user profiles, and identify the
It is much more suitable to use a built-in reasoner concepts to represent them.
to infer the user’s preferences and intentions from 3. Identify concepts to represent the facts
the observations. about destinations that are specific for each
Any practical implementation of the afore- tourist agency. This information is extracted
mentioned requirements leads to representing from expert knowledge, where an expert is
essential knowledge about the domain (tourism) a tourist agent who would be able to clas-
and the portal users (user profiles) in a machine- sify destinations according to the different
readable and machine-understandable form. In criteria; for instance, if the destination is a
other words, it is necessary to develop and use a family destination, a romantic destination,
set of ontologies to represent all important con- etc. After identifying these concepts they
cepts and their relations. need to be connected with other relevant
When ontologies are developed, it is necessary concepts, e.g. create relations between
to populate the knowledge base with instances of destination types and relevant user profile
concepts from the ontology and with relevant rela- types.
tions. After some knowledge is created, it needs
to be coupled with a built-in inference engine to Representing aforementioned three steps in a
support reasoning. Finally, it is essential to enable manner of a formal representation of concepts and
input in the system from the user as reasoning relations results in the creation of the following:
requires some input data to be processed.
In the next sections we describe the ontology 1. The World ontology, with concepts and
we developed to satisfy the requirements, fol- relations from the real world: geographi-
lowed by the knowledge base creation and the cal terms, locations with coordinates, land
architecture of the system that enables processing types, time and date, time zone, currency,
the input from end users. languages, and all other terms that are ex-
pressing concepts that are in a way related
to tourism or tourists, but not to vacation

1033
Applying Semantic Web to E-Tourism

packages that could be offered by some The World Ontology


tourist agencies. This ontology should also
contain the general concepts necessary for The PROTON upper level ontology contains all
expression of semantic annotation, indexing, concepts required by Travel Guides World ontol-
and retrieval (Kiryakov et al., 2003). ogy. In addition, it contains concepts and relations
2. The User ontology containing concepts necessary for information extraction, retrieval and
related to the users – the travellers who semantic annotation. PROTON class we used the
visit the Travel Guides portal. This ontol- most frequently in our World ontology is the class
ogy describes user interests and activities, Location. Figure 2 depicts the hierarchy of the
age groups, favourite travel companies, and class Location and its subclasses in the PROTON
other data about different user profiles. Upper Level ontology.
3. The Travel (Tourism) ontology contains con- The classes and properties from PROTON
cepts related to vacation packages, types of used in Travel Guides are shown in Figure 3.
vacations, and traveller types w.r.t. various Following aliases have been used instead of
tourist destinations. It includes all terms full namespaces: pkm for PROTON Knowledge
being specific to vacation packages offered Management, psys for PROTON System Module,
in tourist agencies and being important for ptou for PROTON Upper Module, and ptop for
travellers, like the type of accommoda- Proton Top module.
tion, food service type, transport service, For more information about PROTON ontology
room types in a hotel, and the like. It is this we refer reader to (Terziev et al., 2005).
ontology that makes a connection between
users and destinations. This is accomplished The User Ontology
by creating user profiles for the users, and
determining the type of destination for each PROTON Knowledge Management (KM) ontol-
vacation package. Finally, user profiles are ogy has been extended to suit the User ontology
linked to relevant destination types (and needs. The most frequently used classes are:
vice versa). User, UserProfile, and Topic. According to the
PROTON documentation, Protont:Topic (the
After the evaluation of existing domain and PROTON top module class) is “any sort of a
upper-level ontologies, we have found that the one topic or a theme, explicitly defined for classifica-
that suits the Travel Guides the best is the PRO- tion purposes”. For the needs of Travel Guides,
TON ontology (Terziev, Kiryakov and Manov, protont:Topic class has been extended to represent
2005). PROTON upper-level ontology includes user interests and activities. Its important relations
four modules, each of which is a separate ontol- and concepts are depicted in Figure 4.
ogy. For the purpose of Travel Guides develop- For determining user profile types, the age
ment, the Upper module of PROTON (Terziev et and the user preferred travel company is of a
al., 2005) was used as the World ontology. This great importance hence relevant concepts have
module was extended to fit the Tourism (Travel) been created inside the ontology: AgeGroup is
ontology. The PROTON Knowledge management a representation of the first and the TravelCom-
module (Terziev et al., 2005) was extended to pany is a representation of the latter (Figure 5).
serve as the User ontology. For example, if the user selects that he/she travels
with family very often, he/she could be considered
as a FamilyType.

1034
Applying Semantic Web to E-Tourism

Figure 2. The Location class and its subclasses in the PROTON Upper Level ontology

Figure 3. The classes and properties from the PROTON ontology frequently used in Travel Guides

The UserProfile class is extended to represent could be assigned for each type of user profile.
various types of tourists. These profiles are made It is this property that reflects the importance of
based on user interests and activities. Figure 6 certain profiles. For example:
depicts various types of user profiles.
In practice, many tourists would be determined User hasUserProfile Adventurer (weight = 2),
to belong to more than one type of user profiles. User hasUserProfile ClubbingType (weight = 1).
For this purpose, there is a property weight that

1035
Applying Semantic Web to E-Tourism

Figure 4. The most important concepts and their relation in the User ontology

Figure 5. Extension of PROTON Group class offered in a tourist agency. Figure 7 depicts the
TouristOffer class and types of destinations as-
signed to tourist offers. These types are used as
indicators of types of tourist offers which are later
being assigned to relevant user profile types.
Figure 8 depicts classes and relations between
them in the Travel ontology. Since the Travel
ontology is an extension of the PROTON Upper
module ontology, there are some concepts and
relations from PROTON that are frequently used.
In this case the user profile is a mixture of They all have appropriate prefixes.
the Adventurer and the Clubbing type, but due As shown on Figure 8, a vacation package being
to the weight values adventure destinations have an instance of TouristOffer class isAttractiveFor
a priority over those that are “flagged” as great- certain type of UserProfile, where this type is
night-life destinations. determined by user’s interests and activities.
Travel Guides User Ontology is available on- Travel Guides Travel Ontology is available on-
line at http:// goodoldai.org.yu/ns/upproton.owl. line at http://goodoldai.org.yu/ns/tgproton.owl.

Tourism (Travel) Ontology Travel Guides Knowledge Base


Due to a huge amount of data that is stored inside
In order to design the domain ontology for the area the knowledge base (KB), it is essential that its
of tourism as well as to “link” tourist destination structure allows easy maintenance. To meet this
types to the user profile types, we extended the requirement, we represent the KB as a collection of
PROTON Upper module ontology. The class Of- .owl files (Figure 9).The circle on the top represents
fer is extended with the subclass of TouristOffer the core and contains concepts such as continents
representing a synonym term for vacation package and countries used by all other parts of the KB.

1036
Applying Semantic Web to E-Tourism

Figure 6. Subclasses of PROTON UserProfile class

Figure 7. The extended ptou:Offer class in Travel KIM KB contains more data that are of interest
Guides Tourism ontology to Travel Guides system and also is built based
on the ontology whose core is PROTON ontology
(Popov et al, 2004), we successfully exploited it
to build our core (continents and countries). This
core is available online at http:// goodoldai.org.yu/
ns/travel_wkb.owl, and is used to initialize other
elements of the KB.
This way we avoided entering permanent data
about various destinations manually, and also
showed that it is possible to share the knowledge
between different platforms when it is represented
The other parts are independent .owl files that
using RDF structure and achieve interoperabil-
are country specific and contain all destinations
ity - the content of one application can be of
inside the country, all hotels on the destinations
use inside the other application, even if they are
and finally all vacation packages related to the
based on different ontologies. Our environment
hotels. For the clarity of the presentation Figure
for knowledge base exploitation is applicable for
9 depicts only 3 elements of the KB apart from
any knowledge base and ontology; the only pre-
the core. Ideally, the number of these elements is
condition is selection of criteria that will define
equal to the number of existing countries.
the statements to be extracted.
To alleviate the creation, extensions, and
Apart from many concepts (e.g., organizations
maintenance of the KB, and also to address the
and persons), KIM Platform KB includes data
interoperability issue, we explored some other
about continents, countries and many cities. The
ontology-based systems that include instances
environment created inside Travel Guides enables
of concepts that are of interest to Travel Guides
extracting of concepts by selecting some of the
system. We have built an environment that en-
criteria, e.g., name of the property. We selected
ables exploiting instances of classes (concepts)
hasCapital, as this property has class Country
and relations of the arbitrary KB in accordance
as a domain and class City as a range. Our envi-
to the predefined criteria. We considered using
ronment extracts not only the concepts that are
KIM KB and also WordNet (Fellbaum, 1998). As
directly related to the predefined property, but also

1037
Applying Semantic Web to E-Tourism

Figure 8. Concepts and relations in the Travel Guides Travel ontology

Figure 9. Organization of the knowledge base inside the Travel Guides system

all other statements that are the result of transitive Ideally, the knowledge base should contain
relations of this property. For example, if defined descriptions of all destinations that could (but
relation Country isLocatedIn Continent exists, need not necessarily) be included in the offers of
statements that represent this relation will also the tourist agencies connected to the portal.
be extracted.
Figure 10 depicts some of the classes and The Portal Architecture
relations whose instances are imported during
the KB extraction. This section gives details about the architecture of
Travel Guides system (Figure 11) and its design.
The system comprises following four modules:

1038
Applying Semantic Web to E-Tourism

Figure 10. Classes and relations whose concepts are imported during KB extraction

Figure 11. Travel Guides Architecture  Travel Module when updating data
about vacation packages and destina-
tions.
4. Knowledge base enrichment module:
For knowledge base enrichment based on
annotations in respect to the ontology. It
communicates with Travel Module to update
knowledge base with new instances and
relations between them.

Following are details about key modules.


1. User Module: For generating user profiles
and maintaining user data. User Module
2. Travel Module: For generating and main-
taining vacation packages and all other data User Controller (Figure 12) accepts requests from
related to vacation packages and destina- the user (via User registration form) and fires
tions. appropriate actions. Actions (at the presentation
3. System Scheduler Module: For update layer) are directly connected to the business layer
of the knowledge base. It communicates of the system represented by User Manager (UM).
with: The UM has the following roles:
 User Module when updating user pro-
files

1039
Applying Semantic Web to E-Tourism

• Store and retrieve data about the user. 1. Using User interface: the user is prompted
• Observe and track the activities of the user to fill the forms to input data about him/
during his visit to the portal. herself. These data are: gender, birth date,
social data (single, couple, family with kids,
For manipulation with data stored in the data- friends), the user’s location, profession,
base UM uses the User DAO (User Data Access education, languages, interests and activities
Object). These data are user details that are not (art, museums, sightseeing, sports, exploring
subject to frequent changes and are not important new places during vacation, animals, eating
for determining the user profile: the username, out, nightlife, shopping, trying local food/
password, first name, last name, address, birth experiencing local customs/habits, natural
date, phone and email. beauties, books), budget, visited destina-
For logging user activities during visiting the tions.
portal UM uses User Log DAO. 2. The system collects data about the user’s
When reasoning over the available data about interests and preferences while the user
the user and determining user profile types UM is reading about or searching for vacation
use the User Profile Expert. The User Profile packages using the portal. Each time the
Expert is aware of the User ontology and also of user clicks on some of the vacation package
the User profile knowledge base (User kb) that details, the system stores his/her action in
contains instances of classes and relations from the database, and analyse it later on.
the User ontology.
The data about users are collected in two Travel Module
ways:
Travel Module generates and maintains data about
vacation packages, destinations and related con-
cepts. The User interface of Travel Module com-
ponent comprises following forms (Figure 13):
Figure 12. User module components
1. Recommended Vacation Packages form:
This form shows the list of vacation pack-
ages that the user has not explicitly searched
for - system generates this list automatically
based on the user profile.
2. Vacation Packages Form: This form is
important for travel agents when updating
vacation packages data.
3. Vacation Package Semantic Search Form:
This form enables semantic search of vaca-
tion packages.

Each of the available forms communicates


with the Controller who dispatches the requested
actions to the Travel Manager (TM). The Travel
Manager is responsible for fetching, storing and
updating the data related to vacation packages. It

1040
Applying Semantic Web to E-Tourism

includes a mechanism for storing and retrieving this TM uses the Travel Offer Expert (TOE)
data from the database using Vacation Package and the World Expert (WE) components.
DAO (Data Access Object). The data stored in 2. The list of destinations retrieved in the first
the database are those that are subject to frequent step is filtered using the constraints the user
changes and are not important in the process of provided (for example, the start/end dates
reasoning: start date, end date, prices (accom- of the vacation). TM filters retrieved result
modation price, food service price, and transport using the Vacation Package DAO.
price), benefits, discounts and documents that
contain textual descriptions with details about the TOE and WE components include inference
vacation packages. Some of these data are used in engines. These inference engines are aware of the
the second phase of retrieving a ‘perfect’ vacation ontologies and knowledge bases: TOE works with
package, when the role of the inference engine Travel ontology and a knowledge base (Travel
is not important. Retrieving a ‘perfect’ vacation kb) created based on this ontology. WE uses the
package is performed in two steps: World ontology and the knowledge base (World
kb) created based on it.
1. Matching the user’s wishes with certain des- After the initial knowledge base is deployed
tinations – the user profile is matched with into Travel Guides application, its further up-
certain types of destinations. To perform date could be performed semi-automatically by
Knowledge base enrichment module (KBEM)
deployed inside Travel Guides. For example, when
a new hotel is built, the knowledge base should
be enriched with this information. This can be
Figure 13. Travel Module Components performed either by:

• Using the Travel Guides environment, where


a tourist agent or administrator manually
enters the name and other data about the
new hotel (Figure 13).
• Performing annotation of the relevant content
with regards to the Travel Guides ontology,
semi-automatically (Figure 14).

Knowledge Base Enrichment Module


(KBEM)

Semi-automatic annotation process starts with


Crawler actions. Crawler searches the Internet
and finds potentially interesting sites with details
about destinations, hotels, beaches, new activities
in a hotel, news about some destinations, popular
events, etc. The result (HTML pages) is trans-
formed into .txt format and redirected to JMS (Java
Message Service) to wait in a queue for annota-
tion process (aQueue). JMS API is a messaging

1041
Applying Semantic Web to E-Tourism

standard that allows application components based Figure 14. Knowledge base enrichment module
on the Java 2 Platform, Enterprise Edition (J2EE) inside the Travel Guides
to create, send, receive, and read messages. It
enables distributed communication that is loosely
coupled, reliable, and asynchronous.
Annotation Manager consumes these plain
documents and connects to the Annotation Server
to perform process of annotation with regards to
Travel Guides ontologies. After the annotation
process is completed, the annotated documents
are sent to JMS to wait in a queue for verification
(vQueue). The Notification Manager consumes
these massages and sends an e-mail to the admin-
istrator with the details about annotated documents
(e.g., location of the annotated documents). The
administrator starts Annotation Interface and
performs the process of verification. The output
of the annotation process is correctly annotated
documents.
Retrieved annotations that refer to the new
concepts/instances could be further used to would annotate them in wrong ways, if it does it
enrich the KB and also for semantic search over automatically without any verification.
the knowledge store that includes processed
documents. Similar approach uses KIM Platform: An Example of Using Travel Guides
they provide querying of the knowledge store that
includes not only the knowledge base created Travel Guides users are divided in 3 groups, each
w.r.t. ontologies, but also annotated documents of which contributing to the knowledge base in
(Popov et al., 2004). its own way.
Annotation of documents performed by KBEM End users (i.e., tourists) visit this portal to
would be simplified in case that verification step search for useful information. They can feed the
is skipped. The implementation of the system system with their personal data, locations, and
would also be simpler. In addition, there would interests, which then get analyzed by the system
not be a human influence, but the machine would in order to create/update user profiles. Note that
do everything by itself. This would lead to many the system also uses logged data about each user’s
missed annotations, though. A machine cannot activities (mouse clicks) when updating the user’s
always notice some “minor” refinements as hu- profile. User profile form for feeding the system
mans can. For example, if in the title “Maria’s with user personal information, activities and
sand” the machine notices “Maria” and finds it interests is depicted on Figure 15.
in the list of female first names, it will annotate it On the left hand side there is a section with
as an instance of a class Woman. “Maria” can be results of system personalization. This section
an instance of a woman, but in this context it is a provides a list of potentially interested destinations
part of the name of a beach. These kinds of mis- for the tourist. The section is created based on the
takes would happen frequently, and the machine user profile analyse, which means that offered

1042
Applying Semantic Web to E-Tourism

Figure 15. The User profile form in Travel Guides

Figure 16. The Search form in Travel Guides

destinations should be matching user wishes, Figure 18 depicts a form for entering a new hotel
interests and activities. To explicitly search for into the KB.
a ‘perfect’ vacation package the user uses form Portal administrators mediate the knowledge
shown on Figure 16. base updates with destinations not covered by
Tourist agents create vacation packages and the tourist agencies connected to the portal. This
similar offers in tourist agencies. They feed and process is very similar to the process conducted
update the database with new vacation packages by tourist agents. The major difference is that
and also knowledge base with new information this part of the knowledge base contains mostly
about destinations. To do this, they fill appro- static and permanent information about some
priate forms and save the filled-in information geographical locations, such as countries, their
(Figure 17). capitals, mountains, rivers, seas, etc. all over the
To successfully fill in this form and save the world. The idea is that tourist agents can use this
vacation package, the hotel has to be selected. If part of the knowledge base as the basis for creating
the hotel does not exist in the system, it has to be new vacation packages and other tourist offers.
entered before creating the new vacation package.

1043
Applying Semantic Web to E-Tourism

Figure 17. The Vacation package form in Travel Guides

Figure 18. Entering a new hotel using the Travel Guides environment

CONCLUSION Built-in heuristics inside ontologies and use


of a reasoner enable implying the user profile
Representing tourism-related data in a machine- types for different tourists w.r.t. their activities
readable form can help the integration of E- and interests. Coupled with the destination types
Tourism Information Services. If tourism sources which are derived from the specific vacation
would be centralized in a unique repository, package descriptions, user profiles can improve
the maintenance efforts would be significantly the process of searching for the perfect vacation
decreased. Integration of all E-Tourism sources package. Additionally, building a good quality user
would result in the possibility to search for tour- profiles provides personalization of dynamically
ist deals from one place – this would drastically created content.
reduce the time tourists spend while searching The system’s prototype described here includes
various tourism-related Web sites. a limited collection of vacation packages. The
main precondition for its evaluation and usability

1044
Applying Semantic Web to E-Tourism

would be feeding it with vacation packages from in large scale applications. At the moment, only
real tourist agencies. few ontology-based systems exist in the area of
As Travel Guides focus on integration of tourism, among which Mondeca (www.mondeca.
Information Services, such as information about com) is applying the most of them to tourism in
destinations, hotels and the like, it would be worth different regions in France. Their ontologies define
exploring the possibility to integrate such a system the structure of data they are working with but the
with existing applications that offer Transactional use of a reasoner is on the minimum level.
Services, so that it can be possible to book and Emerging popularity of social web applications
pay for recommended vacation packages after raises another interesting field of research, spe-
searching repository with available tourist offers cifically information retrieval from user created
covered by Travel Guides. In addition, there are content. Existing Natural Language Processing
opportunities to extend Travel Guides or to develop Tools are still weak to extract and retrieve mean-
an independent module for integration of Commu- ingful answers based on the understanding of the
nication Services, so that tourists can contribute query given in a form of natural language. For
to the system knowledge about the destinations example, searching a social web application (e.g.,
and express their experience as well. a forum with reviews of different hotels), it would
Finally, as the current version of Travel Guides be hard to find ‘the hotel in the posh area’ using
ontology supports only representing hotel accom- mainstream search engines as some of the posts
modation, there is a space for future improvements might talk about luxury hotels, but not using ‘posh’
that include extending types of accommodations to describe them. Developing Natural Language
with hostels, private apartments for rent, and Processing tools that could analyse text so that
campgrounds. machines can understand it is a field with lots
of research opportunities that would contribute
not only to the E-Tourism applications, but to all
FUTURE research directions applications on the Web.
Improving the process of automatic annota-
Integrating semantic web technologies in tradi- tion and developing algorithms for training such
tional existing Web applications has a lot of space a process would be another important contri-
for improvement. The most popular way to per- bution. Up to date, only Named Entities (e.g.,
form this integration is by employing ontologies organisations, persons, locations) are known to
as they enable presenting data in machine readable be automatically retrieved to the reasonable level
form, reasoning and running intelligent agents, of accuracy. Additionally, as current systems for
semantic Web services and semantic search. Each performing annotation process usually require the
of these is partly applied in E-Tourism applications knowledge and understanding of the underlying
nowadays. However, current state of the art in this software such as GATE, research in this field can
field is not mature enough to be used in industry, lead to developing more user-friendly interfaces
meaning that there is lots of space for different to allow handling annotations and verifications
research topics, some of which could be implied without any special knowledge of the underlying
from reading this chapter. software. The most natural way would be that
Reasoning over ontologies is very expensive similar to using tags in Web 2.0 applications, or
due to the state of development of current infer- any other simple way that requires no training
ence engines. Development of better and faster for the user.
reasoner is a precondition for using ontologies

1045
Applying Semantic Web to E-Tourism

References Dogac, A., Kabak ,Y., Laleci, G., Sinir, S., Yildiz,
A. Tumer, A. (2004). SATINE Project : Exploiting
Aichholzer, G., Spitzenberger, M., Winkler, R. Web Services in the Travel Industry . eChallenges
(2003, April). Prisma Strategic Guideline 6: 2004 (e-2004), 27 - 29 October 2004, Vienna,
eTourism. Retrieved January 13, 2007, from: Austria.
http://www.prisma-eu.net/deliverables/sg6tour-
Fellbaum, C. (1998). WordNet - An Electronic
ism.pdf
Lexical Database. The MIT Press.
Antoniou, G., Harmelen, F. V. (2004). Web On-
Guo, Y; Pan, Z; and Heflin, J. (2004). An Evalua-
tology Language: OWL. In Staab, S., Studer, R.
tion of Knowledge Base Systems for Large OWL
(Eds.): Handbook on Ontologies. International
Datasets. The Semantic Web – ISWC 2004: The
Handbooks on Information Systems, Springer,
Proceedings of the Third International Semantic
pp. 67-92.
Web Conference, Hiroshima, Japan, November
Bachlechner, D. (October, 2004), D10 v0.2 7-11, 2004. Springer Berlin/Heidelberg. 274-
Ontology Collection in view of an E-Tourism 288.
Portal, E-Tourism Working Draft. Retrieved
Henriksson, R., (November, 2005), Seman-
January 15, 2007 from: http://138.232.65.141/
tic Web and E-Tourism, Helsinki University,
deri_at/research/projects/E-Tourism/2004/d10/
Department of Computer Science. [Online].
v0.2/20041005/#Domain
Available: http://www.cs.helsinki.fi/u/glinskih/
Cardoso, J. (2006). Developing Dynamic Packag- semanticweb/Semantic_Web_and_E-Tourism.
ing Systems using Semantic Web Technologies. pdf
Transactions on Information Science and Ap-
Hepp, M., Siorpaes, K., Bachlechner, D. (2006).
plications. Vol. 3(4). 729-736.
Towards the Semantic Web in E-Tourism: Can An-
Cunningham, H. (2002). GATE, a General Ar- notation Do the Trick? In Proc. of 14th European
chitecture for Text Engineering. Computers and Conf. on Information System (ECIS 2006), June
the Humanities. 36 (2). 223–254. 12–14, 2006, Gothenburg, Sweden.

Church, K., Patil, R. (1982). Coping with Syntactic Horridge, M., Knublauch, H., Rector, A., Stevens,
Ambiguity or How to Put the Block in the Box. R., Wroe, C. (2004). A Practical Guide To Building
American Journal of Computational Linguistics, OWL Ontologies Using The Protege-OWL Plugin
8(3-4). and CO-ODE Tools Edition 1.0. The University
of Manchester, August 2004. [Online]. Available:
Dell’erba, M., Fodor, O. Hopken, W., Werthner, H.
http://protege.stanford.edu/publications/ontol-
(2005). Exploiting Semantic Web technologies for
ogy_development/ontology101.html
harmonizing E-Markets. Information Technology
& Tourism. 7(3-4). 201-219(19). Jentzsch, A. (April, 2005) XML Clearing House
Report 12: Tourism Standards. Retrieved Septem-
Djuric, D., Devedzic, V. & Gasevic, D. (2007).
ber 6, 2007, from http://www.xml-clearinghouse.
Adopting Software Engineering Trends in AI.
de/reports/Tourism%20Standards.pdf
IEEE Intelligent Systems. 22(1). 59-66.

1046
Applying Semantic Web to E-Tourism

Kiryakov, A., Popov, B., Ognyanoff, D., Manov, ADDITIONAL READING


D., Kirilov, A., Goranov, M., (2003), Semantic
Annotation, Indexing, and Retrieval, Lecture Bennett, J. (2006, May 25). The Semantic Web
Notes in Computer Science, Springer-Verlag. is upon us, says Berners-Lee. Silicon.com re-
Pages 484-499. search panel: WebWatch. Retrieved January 3,
2007, from: http://networks.silicon.com/Web-
Popov, B., Kiryakov,A., Ognyanoff, D.,Manov, D.,
watch/0,39024667,39159122,00.htm
Kirilov, A. (2004). KIM - A Semantic Platform
For Information Extraction and Retrieval. Journal Bussler, C. (2003). The Role of Semantic Web
of Natural Language Engineering, Cambridge Technology in Enterprise Application Integra-
University Press. 10 (3-4). 375-392. tion. IEEE Data Engineering Bulletin. Vol. 26,
No. 4, pp. 62-68.
Prantner, K. (2004). OnTour: The Ontology
[Online]. Retrieved June 2, 2005, from http://E- Cardoso, J. (2004). Semantic Web Processes and
Tourism.deri.at/ont/docu2004/OnTour%20-%20 Ontologies for the Travel Industry. AIS SIGSEMIS
The%20Ontology.pdf/ Bulletin. Vol. 1, No. 3, pp. 25-28.
Roman D., Keller, U., Lausen, H., Bruijn J. Cardoso, J. (2006). Developing An Owl Ontol-
D., Lara, R., Stollberg, M., Polleres, A., Feier, ogy For e-Tourism. In Cardoso, J. & Sheth, P. A.
C.,Bussler, C., Fensel, D. (2005). Web Service (Eds.). Semantic Web Services, Processes and
Modeling Ontology. Applied Ontology. 1(1): Applications (pp. 247-282), Springer.
77 - 106.
Davidson, C., Voss, P. (2002). Knowledge Man-
Siorpaes, K., Bachlechner, D. (2006). OnTour: agement. Auckland: Tandem.
Tourism Information Retrieval based on YARS.
Davies, J., Weeks, R., Krohn. U. (2003a). Quiz-
Demos and Posters of the 3rd European Semantic
RDF: Search Technology for the Semantic Web.
Web Conference (ESWC 2006), Budva, Montene-
Towards the Semantic Web: Ontology-Driven
gro, 11th – 14th June, 2006.
Knowledge Management, pp. 133-43.
Smith, C. F., Alesso, H. P. (2005). Developing
Davies, J., Duke, A., Stonkus, A. (2003b). On-
Semantic Web Services. A K Peters, Ltd.
toShare: Evolving Ontologies in a Knowledge
Stollberg, M., Zhdanova, A.V., Fensel, D. (2004). Sharing System. Towards the Semantic Web:
“h-TechSight - A Next Generation Knowledge Ontology-Driven Knowledge Management, pp.
Management Platform”, Journal of Information 161-177.
and Knowledge Management, 3 (1), World Sci-
Djuric, D., Devedžić, V.,Gašević, D. (2007).
entific Publishing, 45-66.
Adopting Software Engineering Trends in AI.
Terziev, I., Kiryakov, A., Manov, D. (2005). D1.8.1 IEEE Intelligent Systems. 22(1). 59-66.
Base upper-level ontology (BULO) Guidance,
Dzbor, M., Domingue, J., Motta, E. (2003). Magpie
SEKT. Retrieved January, 15th, 2007 from: http://
- Towards a Semantic Web Browser, In Proc. of
www.deri.at/fileadmin/documents/deliverables/
the 2nd International Conference (ISWC 2003),
Sekt/sekt-d-1-8-1-Base_upper-level_ontology__
pp. 690-705. Florida, USA.
BULO__Guidance.pdf

1047
Applying Semantic Web to E-Tourism

Edwards, S. J., Blythe, P. T., Scott, S., Weihong- Kanellopoulos, D., Panagopoulos, A. (2005).
Guo, A. (2006). Tourist Information Delivered Exploiting tourism destinations’ knowledge in
Through Mobile Devices: Findings from the a RDF-based P2P network, Hypertext 2005, 1st
Image. Information Technology & Tourism. 8 International Workshop WS4 – Peer to Peer and
(1). 31-46(16). Service Oriented Hypermedia: Techniques and
Systems, ACM Press.
Engels R., Lech, T. (2003). Generating Ontolo-
gies for the Semantic Web: OntoBuilder. Towards Kanellopoulos, D. (2006). The advent of Semantic
the Semantic Web: Ontology-Driven Knowledge web in Tourism Information Systems. Tourismos:
Management, pp. 91-115. an international multidisciplinary journal of
tourism. 1(2), pp. 75-91.
E-Tourism Working Group (2004). Ontology
Collection in view of an E-Tourism Portal. Oc- Kanellopoulos, D. & Kotsiantis, S. (2006). To-
tober, 2004. Retrieved January 13, 2007, from: wards Intelligent Wireless Web Services for Tour-
http://138.232.65.141/deri_at/research/projects/e- ism. IJCSNS International Journal of Computer
tourism/2004/d10/v0.2/20041005/ Science and Network Security. 6 (7). 83-90.
Fensel, D., Angele, J., Erdmann, M., Schnurr, H., Kanellopoulos, D.,Kotsiantis, S., Pintelas, P.
Staab, S., Studer, R., Witt, A. (1999). On2broker: (2006), Intelligent Knowledge Management for
Semantic-based access to information sources at the Travel Domain ,GESTS International Trans-
the WWW. In Proc. of WebNet, pp. 366-371. actions on Computer Science and Engineering.
30(1). 95-106.
Fluit, C., Horst, H., van der Meer, J., Sabou, M.,
Mika, P. (2003). Spectacle. Towards the Semantic Kiryakov, A. (2006). OWLIM: balancing between
Web: Ontology-Driven Knowledge Management, scalable repository and light-weight reasoner.
pp. 145-159. Presented at the Developer’s Track of WWW2006,
Edinburgh, Scotland, UK, 23-26 May, 2006.
Hepp, M. (2006). Semantic Web and semantic
Web services: father and son or indivisible twins? Maedche, A., Staab S., Stojanovic, N., Studer, R.,
Internet Computing, IEEE. 10 (2). 85- 88. Sure, Y. (2001). SEmantic PortAL -
Heung, V.C.S. (2003). Internet usage by inter- The SEAL approach. In D. Fensel, J. Hendler,
national travellers: reasons and barriers. Inter- H. Lieberman, W. Wahlster (Eds.) In Creating
national Journal of Contemporary Hospitality the Semantic Web. Boston: MIT Press, MA,
Management, 15 (7), 370-378. Cambridge.
Hi-Touch Working Group (2003). Semantic Web McIlraith, S.A., Son, T.C., & Zeng, H. (2001).
methodologies and tools for intraEuropean sus- Semantic Web Services. IEEE Intelligent Systems
tainable tourism [Online]. Retrieved April 6, 2004, 16(2), 46-53.
from http://www.mondeca.com/articleJITT-
Missikoff, M., Werthner, H. Höpken, W.,
hitouch-legrand.pdf/
Dell’Ebra, M., Fodor, O. Formica, A., Francesco,
Kanellopoulos, D., Panagopoulos, A., Psillakis, T. (2003) HARMONISE: Towards Interoperability
Z. (2004). Multimedia applications in Tourism: in the Tourism Domain. In Proc. ENTER 2003,
The case of travel plans. Tourism Today. No. 4, pp. 58-66, Helsinki: Springer.
pp. 146-156.

1048
Applying Semantic Web to E-Tourism

Passin, B., T.(2004). Explorer’s Guide to the Stojanovic, LJ., Stojanovic N.,Volz, R. (2002). Mi-
Semantic Web. Manning Publications Co., grating data-intensive Web sites into the Semantic
Greenwich. Web. Proceedings of the 2002 ACM symposium
on Applied computing, Madrid, Spain, ACM
Sakkopoulos, E., Kanellopoulos, D., Tsakalidis, A.
Press. 1100-1107.
(2006). Semantic mining and web service discov-
ery techniques for media resources management. Sycara, K., Klusch, M., Widoff, S., Lu, J. (1999)
International Journal of Metadata, Semantics and Dynamic service matchmaking among agents in
Ontologies. Vol. 1, No. 1, pp. 66-75. open information environments. ACM SIGMOD
Record. Vol. 28(1), pp. 47-53.
Singh, I., Stearns, B., Johnson, M. and the
Enterprise Team (2002): Designing Enterprise World Tourism Organization, 2001, The-
Applications with the J2EE Platform, Second saurus on Tourism & Leisure Activities:
Edition. Prentice Hall. pp. 348. Online: http://java. http://pub.world-tourism.org:81/epages/Store.
sun.com/blueprints/guidelines/designing_enter- sf/?ObjectPath=/Shops/Infoshop/Products/1218/
prise_applications_2e/app-arch/app-arch2.html SubProducts/1218-1
Shadbolt, N., Berners-Lee T., Hall, W. (2006). WTO (2002). Thesaurus on Tourism & Leisure
The Semantic Web Revisited. IEEE Intelligent Activities of the World Tourism Organization
Systems. 21(3). 96-101. [Online]. Retrieved May 12, 2004, from http://
www.world-tourism.org/aciduis ciduisi bla facil-
Stamboulis, Y. Skayannis P. (2003). Innovation
lum nulla feuguer adignit amet
Strategies and Technology for Experience-Based
Tourism. Tourism Management. Vol. 24, pp.
35-43.

This work was previously published in The Semantic Web for Knowledge and Data Management: Technologies and Practices,
edited by Z. Ma and H. Wang, pp. 243-265, copyright 2009 by Information Science Reference (an imprint of IGI Global).

1049
1050

Chapter 4.10
E-Tourism Image:
The Relevance of Networking for
Web Sites Destination Marketing
Lluís Prats-Planagumà
Universitat de Girona, Spain

Raquel Camprubí
Universitat de Girona, Spain

ABSTRACT real tourist perception. The authors specifically


discuss the importance of a common agreement of
The competitiveness of tourism destinations is a tourism agents on virtual tourism images projected
relevant issue for tourism studies, moreso, is a key through official Web sites, considering that the
element on the daily basis of tourism destinations. In literature focused mainly in how to promote and
this sense, the management of tourism destinations sell destinations trough Internet but not in terms of
is essential to maintain competitive advantages. exploiting a destination joint image. Finally, in order
In this chapter tourism destination is considered to analyze the integration of a tourism product and
as a relational network, where interaction and determine their consequences in tourism promotion
cooperation is needed among tourism agents, to an empirical research has been done, using the case
achieve major levels of competitive advantage and of Girona’s province. The main findings determine
a more effective destination management system. that, although interactions among tourism agents can
In addition, the perceptions of tourists are obtained improve destination competitiveness, little coopera-
from two main sources. The first one is the social tion in tourism promotion on Web sites is achieved,
construction of a tourism destination previous to as well as a few uses of technological resources in
the visit and the second one is obtained from the the Web sites to facilitate to tourists a better under-
interaction between tourists and tourism destination standing of tourism resources in the area.
agents during the visit. In this sense, the manage-
ment of tourism destination to emit a homogenous INTRODUCTION
and collective image is a factor that can reduce
the gap if dissatisfaction from the previous and Each tourism destination can be considered a market
in itself. At these destinations tourism suppliers (i.e.,
DOI: 10.4018/978-1-60566-134-6.ch009 accommodations, restaurants, museums, and tour-

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
E-Tourism Image

ism offices, among others) interact simultaneously agery formation to consumers (Hashim, Murphy
with the tourists who consume these products or & Muhamd Hashim, 2007) and giving to tourism
services. For that reason a market approach is more destination an opportunity to improve destination
appropriate than a supply or a demand one. marketing through the use of “Internet’s unique
A tourism destination is the geographical area features, such as geographical interactivity with
where a set of tourism agents interact and intervene audience, low-cost accessibility, world-wide, hy-
in tourism activities. These interactions, from a perlinks with other travel suppliers and design flex-
supply point of view, help develop a relational ibility, to attract more tourists and better position
network at the destination. A relational network their state in the intense competition for visitors”
is the set of economic and personal relationships (Lee, Cai & O’Leary, 2006, p. 816). Third, Internet
established among a number of agents who share and destination websites, in particular, act as an
goals, cooperation systems, knowledge, reputa- information tool for tourists, being an influencing
tion, and image, among other elements, in com- element in their decision-making. This article will
mon. These elements help the destination network discuss the attainment of these two benefits and
generate collective learning and knowledge, and the relevance of Internet in them using the tour-
consequently, achieve greater levels of competi- ism image and social network theories to clarify
tiveness than individual agents would obtain. how supply and demand interact in a tourism
In addition, from the demand point of view, destination. A conceptual model will be proposed
these interactions within the destination help mini- as part of a theoretical market approach to tourism
mize the existing gap between perceived and real destinations, which integrates supply and demand,
images. All tourists have a socially constructed explains interactions between them and highlights
image of a destination (Urry, 1990; Galí & Donaire, the relevance of this scope of analysis to better
2005; Larsen & George, 2006), which conditions understand the dynamics of a tourism destination
their decision-making, and it is important for the and the possibility of improving its competitive
tourism agents involved in the network to control advantage. In addition, the article demonstrates
the image of a destination. the necessity of using this integrated approach
This control has two simultaneous benefits. for planning and managing a tourism destination
The first one, related to the tourism demand, is to improve its competitiveness and highlight this
the potential to influence tourist decision-making. theoretical view.
The second one is related to the tourism supply One of the elements that can be planned with
chain: the competitive advantage brought to tour- an integrated approach, mainly because it helps to
ism destinations by projecting a real image. establish scale economies in terms of promotion,
Internet is a very important channel that helps it is the promotional website content of a desti-
tourism agents to achieve these two benefits de- nation. Usually different agents take part in this
rived from this control in three aspects. First, a promotion, Destination Marketing Organization’s
number of authors assume the relevance of Internet (DMO) local governments, private companies or
as a tourism destination image generator (Baloglu associations creating different sources of informa-
& Pekan, 2006; Choi, Lehto & Morrison, 2007; tion. Apart from these possible scale economies
Hashim, Murphy & Muhammad Hashim, 2007); this will also help in terms of unification of the
although “research on the Internet as an image destination image. If this image is homogeneous
formation agent is still at an infancy stage” (Choi, gives also an extra value of competitive advantage
Lehto & Morrison, 2007, p. 118). Second, Internet preventing incoordination.
brings a great number of opportunities to tourism Finally, a case study is conducted in order to
image formation, contributing to destination im- analyze network configuration trough promo-

1051
E-Tourism Image

tional websites in Internet and determine if the has of a destination” (Crompton, 1979).
tourism product of a destination is integrated Gunn (1988), in her main academic study
and promoted globally or, on the contrary, each mentions “all of us have images of destinations,
tourism agent acts independently. Results show whether or not we have traveled to them. These
the existence of a reduced network, that means images may be sharp or vague, factual or whimsi-
a little integration of the tourism products in the cal, but in all cases they are indicative of likes and
province of Girona. dislikes.” (p. 23) This means that all places have
This article is organized in five main sections. an image, which has not appeared out of nowhere,
The first explains the process of tourism image but they have consciously or unconsciously been
formation and how the tourism agents that inter- created by “somebody”. In this sense, one needs
vene in this process affect the image that tourists to think about how a tourism destination image
have of tourism destinations, from the perspec- is constructed.
tive of the social construction of tourism images. From the realization of a task and its reiterated
The second focuses on the network configuration repetition by people in a society, this task ends
of the destination, taking into account the tour- up being institutionalized by this society. The ac-
ism agents who take part in the tourism system ceptance of this task as habitual makes it “settle”
and how networking can generate competitive in this society and form part of its traditions, so
advantages. The third part presents a theoretical in that sense, the reality of this society has been
model of an integrated market approach to tour- constructed collectively (Berger & Luckmann,
ism destinations. The forth section presents a case 1968). Using this approach to tourism image, it
study, which analyze the network configuration could be contemplated that a tourism image is
of tourism products in the province of Girona. constructed socially in the same way as a task
Finally, the conclusions based on the theoretical is accepted as a normal way to do something in
model and the case study is drawn, the model’s a society.
limitations are considered and proposals for future Tourism images are full of visual elements and
research are made. signs that evoke socially constructed images (Urry,
1990), for example a couple of lovers in Paris
suggest romantic Paris. In this sense, the image
HOW TOURISTS PERCEIVE construct implies some overriding impression or
DESTINATIONS stereotype (Mazanec & Schweiger, 1981). How-
ever, this tourism image does not always reflect
Social Construction of Tourism the reality, because “the tourism image is, at the
Destination Images same time, a subjective construction (that varies
from person to person) and a social construction,
Images have been used in a number of contexts based on the idea of collective imagination” (Galí
and disciplines: psychology perceives the im- & Donaire, 2005, p. 778).
age as a visual representation; thought behavior Variations in tourism image are complex if one
geography emphasizes the association of impres- considers how these images are formed, a little bit
sions, knowledge, emotions, values and beliefs; at a time. As Gallarza, Gil and Calderón (2002)
and marketing focus on the relationship between have exposed “image is not static, but changes
image and behavior of consumers (Jenkins, 1999). depending essentially in two variables: time and
The majority of academics from the 1970s to the space”. (p. 72) The influence of time on image is
present day agree that tourism image is “the sum demonstrated in a number of studies on tourism
of beliefs, ideas, and impressions that a person image (Gartner, 1986; Gartner & Hunt, 1987;

1052
E-Tourism Image

Figure 1. Seven-stage model of a tourism experience (Source: Gunn, 1972)

Chon, 1991; Selby & Morgan, 1996), especially to the existence of factors or components that
if one considers its formation as a process (Gunn, form part of every tourism image and influence
1972). At the same time, the space variable also its formation process (Baloglu & McCleary, 1999;
influences the image of a tourism destination. Gallarza, Gil & Calderón, 2002; Beerli & Martín,
Some studies in this field show that the distance 2004). In this sense, Gartner (1993) mentions that
between potential tourists and the tourism desti- some authors have systematized the elements that
nation affects the perceived image of the place influence the process of tourism image formation
(Miossec, 1977; Talisman-Kosuta, 1989; Gallarza, in different conceptual models. At the same time,
Gil & Calderón, 2002). Considering the dynamic it is possible to find a number of authors who
nature of tourism, image is useful if the effect of focus on the existence of a formation process of
marketing actions on time and space variables the tourism image, which is made up of different
(Gallarza Gil & Calderón, 2002; Talisman-Kosuta, stages that contribute to how a tourism image is
1989) is taken into account. In this manner the formed (i.e., Gunn, 1972; Govers & Go, 2004).
periodic evaluation of tourism image is relevant One of the most important models that show
(Talisman-Kosuta, 1989). how a tourism image is formed is the seven-stage
process of tourism experience, which has been
How is the Image of a Tourism developed by Gunn (1972). This model shows
Destination Formed? that images held by potential visitors, nonvisi-
tors, and returned visitors differ (Gunn, 1972).
Accepting as valid the fact that tourism image is (Figure 1)
socially constructed (Urry, 1990; Larsen & George, At the first stage, potential tourists assimilate
2004; Galí & Donaire, 2005), some studies point general information, such as, newspapers, televi-

1053
E-Tourism Image

sion documentaries, books, and school lessons. ena through the contact of visitors with tourism
This process generates an organic image of the destinations.
destination; this is because the mere mention of
these places evokes images, which are not neces- Image Management as a
sarily tourism images. Competitive Advantage
The second stage implies a modification of
perceived images based on consulting tourism Academic literature recognizes the need to manage
information (i.e., tourism posters, guides, articles tourism image, as it is one of the most impor-
in specialized reviews, etc.). These changes in tant factors that influences the decision-making
perceived images are influenced by induced im- process of tourists that choose a destination to
ages, which are the result of a conscious effort to spend their holidays (Gartner, 1993; Govers &
develop, promote, and advertise a destination. Go, 2004). Gunn (1972), in her model explains
When the potential visitor has a perceived that tourism images are conditioned by the actions
image based on the organic and induced images of a number of agents that influence the creation
of the place, then they are prepared to make a of tourism images. Although it is agreed that the
decision. Other factors such as previous experi- tourism image is socially constructed, agents
ence or the money available are also taken into intervene in this process emitting images, which
consideration. end up being consolidated and accepted as valid
Travel to the destination may condition the in a specific society.
image that a visitor has, but the key factor of a According to Gartner’s (1993) agents’ classi-
new change in a visitor’s perceived image is their fication, there are four types of agents. The first,
personal experience at the destination, as well as Overt Induced is a kind of agent who promotes
their participation in different activities, such as, the creation of a specific tourism image of the
visiting museums or the use of tourism services tourism destination in a conscientious way, to
such as accommodation. At this stage, visitors have influence a tourist’s process of decision-making.
a modified-induced image, which is the result of Gartner (1993), makes a distinction between these
the balance between the perceived image before agents, who are of two types. On one hand, Overt
visiting the destination and the perceived image Induced I are “the promoters of the destination
after the visit. [that] construct an image of the salient attributes
Returning home after traveling, visitors evalu- of the destination in the minds of the targeted
ate and make reflections about their experience audience” (Gartner, 1993, p. 197) with the tradi-
and discuss it with other travelers. At the final tional forms of advertising (i.e., television, radio,
stage, tourists accumulate new information if brochures, etc.). In this case, one could also include
one considers that this is a circular process. In tourism businesses of the destinations, such as,
this sense, it is widely recognized in academic accommodation, restaurants, activities, and so on.
literature that experienced tourists will become On the other hand, Overt Induced II are usually
a “source of information” for other potential “tour-operators, wholesalers or organizations
visitors, which will be based on their experience who have a vested interest in the travel decision
at the destination (Balogru & McCleary, 1999; process, but which are not directly associated with
Beerli & Martin, 2004). a particular destination area” (Gartner, 1993, p.
As is noted in this model, the creation and 199). As Gartner (1993), mentions “destination
modification of tourism images are constant area promoters do have some control over the
and demonstrate the dynamism of the tourism images projected through tour operator” (p. 199)
image. The space variable shows these phenom- because if the tourism image does not conform to

1054
E-Tourism Image

the reality of the destination it could create dis- agents, in this context, is undeniable, especially if
satisfactions to both locals and visitors (Govers one considers the sustainable competitive advan-
& Go, 2004). tage of the destination. Sustainable competitive
The second, Covert Induced are agents that ap- advantage is generally based on either core com-
parently emit a tourism image that is not induced. petences or unique resources that are superior to
In this case the author also defines two types of those possessed by competitors and are difficult to
Covert Induced agents. The first is called Covert imitate (Johnson & Scholes, 1999; Aaker, 2001).
Induced I, who is related to a recognizable spokes- Govers and Go (2004, p. 169), established that
person who recommends a destination to support superior resources for a tourism destination “are
a higher level of credibility of tourism destination generally to be found in either its unique and natu-
advertising. The second is Covert Induced II, this ral environment (climate, wildlife or landscape)
category corresponds to people or organizations or its cultural heritage” and also mention that
who write articles, reports or stories about a par- “competitive advantage might be created through
ticular place. Often this published information is core competences, such as, the host community’s
a result of a familiarization tour for travel writers existing unique capabilities in attracting visitors
or special interest media groups. These actions from outside.” (i.e., destination’s ability to stage
increase credibility and allow destination promot- world class events, festivals or exploit its folklore
ers to project a specific image. and prevailing traditions).
The third kind of agents is called Autonomous. Following these considerations, the manage-
These agents are people or organizations who ment of tourism image is viewed as a manage-
produce reports, documentaries, movies, and news ment tool (Ritchie, 1993). Govers and Go (2004),
articles independently without the specific aim of propose that it is necessary “to formulate a plan
creating a tourism image of a place. for projecting the ‘right’ image” (p. 170) as one of
The last group of agents that Gartner (1993), the essential parts of tourism development strat-
identifies is called Organic, and is related to infor- egy. Gartner (1993), mentions the importance of
mation and opinions about a place that a person considering the “image mix”, as a continuum of
receives from other people, from their previous factors that have to be taken into account to decide
experience in this place, Unsolicited Organic cor- which agents will intervene in the formation of
responds to people who give information about a tourism image, as well as, the amount of money
destination where they have been, without having budgeted for image development, characteristics
been specifically asked by the other interlocutor, of target market, and demographic characteristics
for example, when this is a topic of conversation or timing. This task is obviously attributed to
with friends in colleges. The existence of Solicited promoters of the destination who can select the
Organic agents implies that individuals actively right mix of image formation agents to maximize
search for information about a destination and their scarce resources (Gartner, 1993).
somebody informs them using their own experi-
ence. Friends or relatives usually constitute these
kind of agents, who have a high level of cred- THE NETWORK CONFIGURATION
ibility and are an extremely important part of the AND A TOURISM DESTINATION
destination selection process.
When people visit a destination they become an Tourism Destination Agents
Organic (pure) agent, having the capability to give
information in a solicited or unsolicited way. Tourism agents are an essential part of the system
This model shows that the task of Overt Induced and of any destination, therefore they have been

1055
E-Tourism Image

Figure 2. Tourism destination agents

identified. Some definitions that are more appli- generating specialized training and/or research
cable to industrial destinations, like innovation in the scope of tourism, such as, universities,
systems, clusters, milieux innovateurs, or indus- research institutes, or consultants.
trial districts, always consider three main types These three main agents appear in the academic
of agents. Moreover, it is considered that tourism literature on innovation systems (Lundvall, 1992;
specificity needs another main agent who helps Nelson, 1993), as well as, in clusters and indus-
to define the situation of the system and therefore trial districts. In addition, the tourism scheme
needs to complete it. proposed by Gunn (1997), and later adapted by
Two other types of agents support these main the OMT (1999) is taken, and the tourism industry
agents at all times and are also necessary to as a functional tourism system (Prats & Balagué,
maintain the main set of agents stable. Figure 2 2005) is conceptualized. They demonstrate that the
illustrates these agents in detail. local community also has an essential role in the
Following this scheme, those organizations development of tourism activity, and consequently
that take part directly in generating the tourism of the system. The local community is defined
experience are private companies. These include as the inhabitants of a territory. These people are
basic tourism companies, and also others whose individuals or organizations without economic
main activity is related to tourism. aims, such as, NGO’s, civic organizations, or
Public administrations are those organisms of others. The relevance of local community in tour-
governmental function that take part in the tourism ism is emphasized, seeing that civic movements
processes and whose intervention can generate have been able to modify important decisions
new legislations, give incentives for research, in city-planning, ecological subjects, or others,
planning, and others. restraining or impelling tourism.
Research, development, and training centers After describing the basic elements, the tourism
(R+D+T) are the essential elements capable of auxiliary agents can be defined as those agents

1056
E-Tourism Image

who do not have activities directly related to the edge (Sundbo, 1998; Roberts, 2001) and learning.
tourism industry, but who support the main agents. It is also totally accepted as a key element in the
Looking at the economic theories, the auxiliary innovation process. Thus, innovation also arises
agents are some of the receivers of the multiplying and takes place through the interactions between
effect (McIntosh, Goeldner & Ritchie, 2000). And companies (Sorensen, 2004), and between these
the external agents are those tourism agents who companies and other relevant actors who are
are part of other destinations, but who interact important for their activity (Prats & Guia, 2005).
with one or more internal agents. These ties must be understood as intense flows of
The set of agents in a tourism destination is ba- knowledge and, therefore, essential for innovation,
sically located in the same geographical territory. and also for competitiveness.
However, a territory by itself does not have enough However, Sorensen (2004) presents a definition
conditions for their collective coordination, and that considers networks as the set of conscious and
also the proximity does not generate synergies by accepted business relationships, whether formal
itself, but it can contribute to their effectiveness or informal, with transmission of resources, im-
with other dimensions shared between the agents material or material, within the company’s scope.
(Zimmermann, 2001). A good example of an agent In any case, it is useful to adopt the perspective
who belongs to a distant destination could be a of social network analysis, which studies specific
specialized tour operator who commercializes relationships between a defined series of elements,
destinations, which are geographically distant like people, groups, organizations, countries or
but relationally close. events, among others (Molina, 2001). It is nec-
essary to consider that social network analysis is
Relational Networks based on relationships and not on the attributes of
elements. Then, a social network can be defined as
The use of relational networks in the analysis of a the group of people, organizations or other social
company’s competitive advantage can be related entities connected by a set of significant relations
to several approaches in the fields of economics (Wellman, 1997).
or sociology, among others (Sorensen, 2004). Granovetter (1985) and Hite (2003), affirm that
Therefore, in the most static frame, this analysis the existing relationships within social networks
has appeared within the network of individual influence economic actions, and Hite (2003),
companies who have useful and important connec- distinguishes seven different types of ties that can
tions with other companies, becoming more than take place inside a social network: the main three
just a unit inside an atomized market (Håkansson are business ties, personal ties, and hollow ties,
& Snehota, 1995). In this sense, these companies and the other four types are formed as a result of
must be analyzed considering their relationships the relationship between the main types.
with other companies outside the network and Porter (1990), with his five forces model and
also the existing relationships among other com- his later approach to clusters, universalized the
panies within the network (Holmen, Pedersen & necessity to maintain the business or commercial
Torvatn, 2005). ties that had been previously valued by Becattini
In relation to the most dynamic frame, it is (1979), and other authors. Other theories such as
observed that it was contemplated not to see the the industrial districts theory show that personal
innovation process as a linear and consecutive relationships have to be considered as a value that
process, meaning that the result of the initial stage contributes to empower the agents’ ties making
brings up the following one and so on. Innovation them more efficient and trustworthy (Becattini,
is considered an intensive activity in both knowl- 1979). Hollow ties appeared only recently in

1057
E-Tourism Image

network theories and have become very common, websites. “Because destination images influence
because they represent all those ties that you accept tourist behavior, a destination must be favorably
with the mediation of a third person, so your trust differentiated from its competition and positively
in the relationship is not with the agent to whom positioned in the minds of consumers” (Hudson
you are related, but with the agent who did the & Ritchie, 2006, p. 388), and this can be achieved
mediation (Prats, Camprubí & Comas, 2005). also by the common internet promotion.
It seems evident then, that a tourism destination Belonging to a destination or relational network
can be defined as a relational network. In every involves interacting with other members, which is
single destination exist relationships among its usually transformed into routines of the organiza-
agents (Prats, Guia & Molina, 2007), consider- tion. This is what Rallet and Torre (2004) call the
ing the specificities of the tourism product, and belonging logic. This logic and interaction will be
the existence of the different types of ties able to easier a priori if there is a common knowledge;
generate an active and beneficial set of agents and this is called logic of similarity.
relationships. As an example, a tourism package The interaction of these agents generates a
is integrated by different items that are provided number of factors that determine if a destination
by various companies, which are linked though or local innovation system is successful or not in
some kind of relationship. However, tourists do all scopes. A first and fundamental factor is the
not distinguish that these items belong to different internal and external relationships that take place
companies, although they need a perfect integra- in the system. These relationships can be very
tion of them, in order for the tourism products to different and they have been summarized into
be successful. Another example can be the usual two characteristic groups.
sectorial associations, where hotels, travel agen- On the one hand, depending on the relational
cies or other tourism agents are associated in order structure that is adopted in a system, the degree
to gain power in front of the suppliers or in front of success will vary. In this factor the key element
of the administration. is the degree of connectivity that is obtained,
understanding that the better the connectivity
Networking as a Generator between the agents is, the closer it will be to “the
of Competitive Advantage ideal” system. It is understood that good internal
connectivity will contribute to a more fluid cir-
If different agents interact among themselves, it culation of knowledge between the agents, and
can be argued that these interactions often allow this will increase the trust among them. But at
the agents to have joint benefits from infrastruc- the same time, an excess of internal connectivity
tures, common engineering, and transfer of tacit can make the trust on external agents decrease to
knowledge. It also makes productive combina- such an extent that they are considered intruders
tions and interactions more difficult to carry out (Zimmerman, 2001). The lack of trust between
in atomization or individual isolation. external agents could have serious consequences
Even with the continuous growth of the on-line in the new knowledge generation, because the
travel expenditure, the academic literature focused closure of relational networks in itself could limit
mainly in how to effectively build and evaluate information flows that come from outside, block-
hospitality and tourism websites (Han & Mills, ing the possibility of generating new knowledge
2006). Moreover the analysis is focused on indi- and collective learning (Lazerson & Lorenzoni,
vidual perspectives comparing different websites 1999).
as example, but there are not published results On the other hand, however, it must also be
related to collective image construction trough observed that the quality of relationships within

1058
E-Tourism Image

Figure 3. Tourism local innovation system model (Source: Prats & Guia, 2005; Prats, Guia and Molina,
2007)

a system such as this, affects its success. The key be able to generate essential collective knowledge
element in this factor is trust, as a greater trust and learning for the evolution of the system. The
between the elements of a system will transmit main purpose of this collective knowledge and
more relevant information, and greater benefits learning is being able to generate a constant in-
for the whole destination will increase. novation capacity that will bring dynamism to the
Another determining factor is the macro-envi- system as shown in Figure 3.
ronment, which is divided into five elements: (1) This innovation capacity allows the system
political, such as, decisions or political elements to obtain four successive outputs, which can be
that affect the system; (2) economic, for instance observed in Figure 4. Each stage must be achieved
economic situations that affect the system; (3) to obtain the desired results. If an “ideal” configu-
technological, which has two levels: (a) the hard ration of the system is obtained, the four outputs
level such as the automation level, and (b) the will also be obtained, and this will revert again
soft level such as the training level of the popu- to the tourism destination.
lation; (4) social, this contributes to the system The first unquestionable output of the innova-
culture, for example the degree of associationism tion capacity is innovations in any of their modali-
or the cultural level; and finally, (5) historical ties. In the opinions of Prats and Guia (2005) the
macro-environment, which gives perspective and innovation must allow the system to generate a
historical experience, such as, political periods or competitive advantage, allowing the destination
natural disasters. to satisfy the needs of the tourists better than the
Using the agents’ interactions and macro-en- competitor’s destinations.
vironment variables, tourism destinations should The competitive advantage, consequently,

1059
E-Tourism Image

Figure 4. Tourism local innovation system outputs (Source: Prats and Guia, 2005)

must contribute to the system’s collective wealth, among internal and external agents of the tourism
which in turn increases the wellbeing of all the destination are also necessary, so that tourists can
agents who join it. Wellbeing is understood to perceive a real tourism image of the destination
be an improvement of the quality of life of all both before and after traveling to the destination.
the elements, which is not based solely on the If this situation occurs, it will be easier for tour-
economic, environmental, or social benefits at an ism destinations to maintain their competitive
individual level, but is a perfect balance between advantage in a sustainable way.
all of them at a collective level. Therefore, tourism destinations, which are
This balance allows the system to become established as a network and based on trust among
sustainable and generates a new and better situ- its members, can better guarantee a unique tourism
ation that is a territory improvement, and which image, which is more coherent with the reality of
also feeds the agents and the macro-environment, the tourism destination. This configuration has a
varying the behavior of the system constantly, close relationship with the structure of network,
forcing it to reframe itself, and be constantly the quality of ties, and its macro-environment.
dynamic. Moreover, the innovation capacity generated has
to be useful to adapt the induced image to the
tourism product.
DESTINATION TOURIST Figure 5 shows a market approach model,
PERCEPTION & NETWORK which focuses on demand and supply simultane-
CONFIGURATION: A ously and explains the relationship between them,
THEORETICAL PROPOSAL taking into account the multiple factors that affect
the behavior of tourism agents. As a consequence
The tourism image perceived by tourists and the competitive advantage of tourism destinations
represented in Gunn’s model (1972) has a close is explained.
connection with tourism destinations, and in par- In this model the interaction between tour-
ticular, with agents that interact in the promotion ism agents and tourists is highlighted, giving an
of the destination. overall picture of what happens in a destination.
The seven-stages of tourism experience (Gunn, In this context, those tourism agents who in-
1972), show interactions between tourists and teract frequently with tourists tend to be public
tourism agents. This materializes, initially, in the administrations, private companies, and the local
process of searching for information, which is community.
done voluntarily by tourists; and later, if tourists First, public administrations have a direct re-
travel to the destination the tourism image is again lationship with tourists through the promotional
modified by direct contact with tourism agents. actions of tourism destinations. These agents act
In this context, direct contact and coordination as induced agents of tourism images by acting as

1060
E-Tourism Image

Figure 5. Market approach of tourism destination: Conceptual model

promoters and developers of destinations. In this relationship with the tourist and usually the tour-
case, following Gartner’s classification (1993), ist’s real image is strongly modified by this kind
public administrations act as Overt induced agents of contact. In the author’s opinion a key factor at
I, who emit an induced tourism image, which this level is the perception the local community
influences both the tourists’ perceived image of has of the tourism activity and the benefits that
a destination and the decision-making process at the inhabitants receive from it, because if a local
the moment of choosing a destination (Gartner, community thinks that the benefits and damages
1993). that tourism causes are in perfect balance or in
Second, tour operators, as they form part of a more beneficial situation for the local com-
the private companies of a destination, also act munity, these inhabitants will contribute to the
as Overt induced agents II, because they have a tourists well-being, otherwise they will behave
clear interest in influencing the decision-making to the contrary.
process of tourists at the time of selecting a des- Tourism research and training centers play a
tination. Public administrations, as well as tour secondary, but fundamental role, especially as
operators, influence the “perceived pre-travel they might condition the induced tourism image
image” of tourists. through research projects, and simultaneously,
When the potential tourist travels to a destina- they could also influence perceived tourism im-
tion and becomes a real tourist a direct interaction ages indirectly by training the tourism workers
between tourists and tourism companies takes who help tourists during their stay.
place, and this influences the perceived image of Therefore, the influence on “perceived post-
the place that tourists had before going there, creat- travel image” comes from tourism companies,
ing a new image of the place (Gunn, 1972). as well as the local community and research and
Finally, the local community has a strong training centers.

1061
E-Tourism Image

This model shows that, the relationships be- a tourism destination in a local and regional level
tween tourists and tourism agents are systematic in order to know if tourism product is configured
and necessary throughout the whole process. globally through collaboration between destina-
This means that this interaction is essential both tion’s tourism agents; or in a contrary way the
before a tourist travels to a destination as well as tourism product is fragmented. This can help us to
during his/her stay. understand the level of tourism image coherence
If tourists do not go to a destination, this might that is transmitted to tourist through Internet.
mean that tourism agents cannot control the fac-
tors that generate the appropriate knowledge and Methodology
transmit the right image to convince tourists.
However, other uncontrollable factors exist, such In order to analyze how a tourism product is
as, the travel time needed, the distance to the des- promoted in a local and regional level a website
tination, the money available or to what extent a analysis of public institutions from the Girona’s
tourism product fulfils tourist needs. province, which has been conducted during the
When tourists are dissatisfied with their first trimester of the year.
visit because the “perceived post-travel image” The Girona’s province is situated in the north-
is extremely different from the “perceived pre- eastern of Catalonia and their capital is the city of
travel image”, there is another scenario where Girona. It is a very rich region in terms of natural
the agents have not transmitted the reality of the resources and heritage; and it has a very privi-
destination. In this situation tourism agents had leged placement between the sea and the oriental
the innovation capacity, but they had not used it Pyrenees. Tourism activity is mainly developed
in the correct way to obtain the desired image in the coastal area, although in Pyrenees tourism
outputs. This context shows the importance of is also a relevant activity for their economy. In
communication and coordination among all the general terms, Girona’s province is integrated by
tourism agents of a destination, to induce a real two tourism brands: (a) Costa Brava for the coastal
and homogenous image. area, and (b) Girona’s Pyrenees for the mountain
area. These two brands divide their territory and
facilitate their promotion.
CASE STUDY Websites of 41 public institutions have been
analyzed, taking into consideration three labels of
As we mentioned before, tourism products and public institutions: (a) DMO of the province, (b)
even more evident tourists doesn’t understand regional institutions, and (c) local institutions.
political boundaries, but regional and local The existence of a section in their website re-
governments use it to divide the territory. This lated to tourism activities has been the criteria to
situation causes management and commercializa- select websites. In this context, we have analyzed
tion problems that don’t help to emit a coherent the role of public agents that act as Overt induced
image of the whole destination to the possible I, following the Gartner’s nomenclature (1993) as
tourists. These scenarios can be avoided develop- we mentioned above.
ing networks for product commercialization that Data collection has been done taken into con-
includes all the Overt induced I agents. Is one of sideration the level of public institution searching
the easiest way to start a network, because in that the existing relationship among them. Three kinds
sense they can share marketing costs entering to of data have been collected. Firstly, it has been de-
scale economies. tected direct and indirect links that are placed into
We analyzed the public institutions’ websites of de website and which is the linked agent. Secondly,

1062
E-Tourism Image

Figure 6. Relational map of direct web links

it has been observed the quality of detected links as a link creator, which means that they promote
and it has determined their typology (advertising the tourism elements independently from the
links, friends’ links or partners’ links). Finally, it municipalities that they represent. (Figure 6)
has been also analyzed the e-marketing resources Jumping to the technological elements included
that have been used in the websites. into the webs that can help to emit a better image
In order to analyze data, it has been used UCI- five different types of them were analyzed, the type
NET 6.0 software (Borgatti, Everett & Freeman, of compelling web system, the multimedia systems
2002), which is a software specialized in social divided into sound, video and photography gal-
network analysis in a qualitative and quantitative leries, and finally the interaction with the tourists
way. In our case it has been used the application represented by interactive maps (Table 1).
to represent graphically the social network. What is really shocking is that the main Cata-
lan destination, only have little technological and
Data Analysis interactive elements to attract tourists. This is done
by the historical tourism tradition of the destination
Observing the network that can be drowned after
the analysis, is possible to assure that don’t exist a Table 1. Use of technological and interactive
common commercialization network in the Costa web resources
Brava destination. From the 41 analyzed webs only
10 have direct links with other promotion agent. Type N
The main problem is that only one of these links Advanced Compelling 18
is bidirectional. This means that the rest maybe Video 2
are not well developed links. Sound 1
Another element to extract from this relational Photographic galleries 11
map is the fact that only 2 of 8 regional institutions Interactive maps 4
appeared on it, and always as link receiver and not Total 36

1063
E-Tourism Image

Figure 7. Percentage of technological and interactive web resources

when an intensive promotion to attract tourists This integrated vision gives a greater innovation
never was needed. Most of the mature tourism capacity, in particular, regarding the destinations’
destinations have a similar problem. tourism agents; who have a broader view of the
One third of the tourism Web sites of the possibilities of maintaining and improving the
destination don’t have any of the mentioned ele- sustainable competitive advantage of the destina-
ments and nearly half of them only have one. This tion; allowing to consider the tourists’ key role
situation sorts out that the emitted image trough and how the tourism agents’ interact with them,
websites and technological elements are really as well as when this interaction takes place, and
poor. At least one of the webs that have three of what the basic tools that maintain this relation-
these elements is the one of the main DMO maybe ship are.
the most visited site. The problem is that in this In this context it is assumed, as justified earlier,
site there is not all the information related with that a tourism destination’s image is constructed
destination as we can imagine for the link system socially during a complex process in a seven-stage
showed before. (Figure 7) tourism experience. Moreover, the necessity to
manage this image is accepted as a method that
influences the tourists’ process of decision-making
Conclusion and for that reason it is necessary to pay special
attention to the agents who take part in this.
Discussion Nowadays, Internet is positioned as a relevant
tool to contribute to destination imagery formation
Initially, it has to be mentioned that it is important as well as and information source for tourists that
to observe demand and supply in an integrated way. can influence their decision-making. At the same

1064
E-Tourism Image

time, it is commonly accepted that city marketing induced agents, have not been considered. In
differs in many ways of destination marketing, future research, it would be necessary to revise
but if it is analyzed the common projected im- the conceptual model and include autonomous
age trough websites the division line disappears. and organic agents, to have a better proxy, tak-
So we can assume that a big city like Barcelona ing into account that tourists also interact with
can have the same number of official emitting autonomous and organic agents before traveling
image websites than a regional destination like to the destination.
Costa Brava. In the authors’ opinion, it could also be inter-
In the authors’ opinion, social network analy- esting to consider the difference between real and
sis is a perfect approach for studying tourism potential tourists in future revisions of the model.
destinations. In this sense, they have highlighted This reflection could have relevant implications,
the interaction among tourism agents to create a especially for observing the induced tourism im-
tourism product or service adequate to tourists’ age and searching for explanations for a tourist’s
needs, as one of the most important factors. One reasons to travel to a specific destination and not
of these interactions can be done in terms of public to others.
agents’ promotion for the whole destination; this
can be materialized though official websites and
links those appear in it. REFERENCES
As can be seen, in the case study, the promo-
tion through websites in Girona’s province does Aaker, D. (2001). Strategic market management.
not help to create a coherent and jointly destina- (6th ed.). New York, NY: John Wiley and Sons.
tion image. This situation can be avoided with Baloglu, S., & McCleary, K. W. (1999). A model
a better and wide interaction among the public of destination image formation. Annals of Tourism
agents, materializing it through their websites. In Research, 26(4), 808–889. doi:10.1016/S0160-
addition, probably, the low use of technological 7383(99)00030-4
and interactive resources in websites is done by
the hard sedimentation of tourism industry in this Baloglu, S., & Pekan, Y. A. (2006). The website
region, and in our opinion these elements will need design and Internet site marketing practices of
to be improved if tourism destination goes into a upscale and luxury hotels in Turkey. Tourism
declining process of their life cycle. Management, 27, 171–176. doi:10.1016/j.tour-
man.2004.07.003
Limitations and Future Research Becattini, G. (1979). Dal sttore industriale al
distretto industrialle. Alcune considerazione
The work that has been presented in this article
sull’unità dell’economia industriale. Revista di
opens a wide field of future research that takes
Economia e Politica Industriale, 1.
into consideration all the agents mentioned in
the presented model. This empirical analysis Beerli, A., & Martín, J. D. (2004). Factors in-
will contemplate if having a whole picture of the fluencing destination image. Annals of Tourism
market will contribute to really improving the Research, 31(3), 657–681. doi:10.1016/j.an-
planning and management of tourism destina- nals.2004.01.010
tions, as discussed.
Berger, P. L., & Luckmann, T. (1968). Social con-
In this first conceptualization the autonomous
struction of reality. Amorrortu: Buenos Aires.
and organic agents from Gartner’s model (1993),
who generate an uncontrolled tourism image by

1065
E-Tourism Image

Borgatti, S. P., Everett, M. G., & Freeman, L. Granovetter, M. S. (1985). Economic action and
C. (2002). Ucinet 6 for Windows: Software for social structure: The problem of embeddedness.
social network analysis. Harvard: Analytic Tech- American Journal of Sociology, 91, 481–510.
nologies. doi:10.1086/228311
Choi, S., Lehto, X. Y., & Morrison, A. M. (2007). Guia, J., Prats, L., & Comas, J. (2006). The desti-
Destination image representation on the web: nation as a local system of innovation: The role of
Content analysis of Macau travel related Web sites. relational networks. In L. Lazzeretti, & C. Petrillo
Tourism Management, 28, 118–129. doi:10.1016/j. (Eds.), Tourism local system and networking.
tourman.2006.03.002 Elseivier: Amsterdam.
Chon, K. S. (1991). Tourism destination im- Gunn, C. A. (1972). Vacationscape: Designing
age modification process: Marketing implica- tourist regions. Washington DC: Taylor and
tions. Tourism Management, 12(1), 68–72. Francis/University of Texas.
doi:10.1016/0261-5177(91)90030-W
Gunn, C. A. (1988). Vacationscape: Designing
Crompton, J. L. (1979). An assessment of the im- tourist regions. New York: Van Nostraud Rein-
age of Mexico as a vacation destination and the hold.
influence of geographical location upon the image.
Gunn, C. A. (1997). Vacationscape: Develop-
Journal of Travel Research, 18(4), 18–23.
ing tourist areas. Washington DC: Taylor and
Galí, N., & Donaire, J. A. (2005). The social con- Francis.
struction of the image of Girona: A methodological
Håkansson, H., & Snehota, I. J. (Eds.). (1995).
approach. Tourism Management, 26, 777–785.
Developing relationships in business networks.
doi:10.1016/j.tourman.2004.04.004
London: Routledge.
Gallarza, M. G., Gil, I., & Calderón, H. (2002).
Han, J. H., & Mills, J. E. (2006). Zero acquaintance
Destination Image – Towards a conceptual frame-
benchmarking at travel destination Web sites:
work. Annals of Tourism Research, 29(1), 56–78.
What is the first impression that national tourism
doi:10.1016/S0160-7383(01)00031-7
organizations try to make? International Journal
Gartner, W., & Hunt, J. (1987). An analysis of of Tourism Research, 8, 405–430. doi:10.1002/
state image change over a twelve year period jtr.581
(1971-1983). Journal of Travel Research, 16(2),
Hashim, H. H., Murphy, J., & Muhamad Hashim,
15–19. doi:10.1177/004728758702600204
N. (2007). Islam and online imagery on Malay-
Gartner, W. C. (1986). Temporal influence on sian tourist destination Web sites. Journal of
image change. Annals of Tourism Research, 13, Computer-Mediated Communication, 12(3), 16.
635–644. doi:10.1016/0160-7383(86)90006-X doi:10.1111/j.1083-6101.2007.00364.x
Gartner, W. C. (1993). Image formation process. Hite, J. M. (2003). Patterns of multidimensionality
Journal of Travel & Tourism Marketing, 2(2), among embedded network ties: A typology of rela-
191–215. doi:10.1300/J073v02n02_12 tional embeddedness in emerging entrepreneurial
firms. Strategic Organization, 1(1), 9–49.
Govers, R., & Go, F. M. (2004). Cultural identi-
ties constructed, imagined and experienced: A Hjalager, A. M. (2000). Tourism destinations and
3-gap tourism destination image model. Tourism, the concept of industrial districts. Tourism and
52(2), 165–182. Hospitality Research, 2(3), 199–213.

1066
E-Tourism Image

Holmen, E., Pedersen, A. C., & Torvatn, T. (2005). McIntosh, R. W., Goeldner, C. R., & Ritchie, J. R.
Building relationship for technological innovation. B. (2000). Turismo: Planeación, administración y
Journal of Business Research, 58(9), 1240–1250. perspectivas. Mexico D.F.: Limusa Willey
doi:10.1016/j.jbusres.2003.10.010
Miossec, J. M. (1977). L’image touristique
Hudson, S., & Ritchie, J. R. B. (2006). Promot- comme introduccion à la géographie du tourisme.
ing destinations via film tourism: An empirical Annales de Geographie, 55–70. doi:10.3406/
identification of supporting marketing initia- geo.1977.17568
tives . Journal of Travel Research, 44, 387–396.
Molina, F. X. (2001). European industrial dis-
doi:10.1177/0047287506286720
tricts: Influence of geographic concentration on
Jackson, J., & Murphy, P. (2002). Tourism destina- performance of the firm. Journal of International
tions as clusters: Analytical experiences from the Management, 7(4), 277–294. doi:10.1016/S1075-
World. Tourism and Hospitality, 4(1), 36–52. 4253(01)00048-5
Jenkins, O. H. (1999). Understanding and measur- Nelson, R. (1993). National innovation systems.
ing tourist destination images. International Jour- New York, NY: Oxford University Press.
nal of Tourism Research, 1, 1–15. doi:10.1002/
OMT. (1999). Desarrollo turístico sostenible:
(SICI)1522-1970(199901/02)1:1<1::AID-
Guia para administradores locales. Madrid:
JTR143>3.0.CO;2-L
OMT.
Johnson, G., & Scholes, K. (1999). Exploring
Porter, M. E. (1990). The competitive advantage
corporate strategy. Harlow, UK: Prentice Hall.
of nations. London: The McMillan Press.
Larsen, G., & George, V. (2004, February). The
Prats, L., & Balagué, J. (2005). Cohesión y sos-
social construction of destination image – A New
tenibilidad, elementos clave en la competitividad
Zeland firm example. Working Papers, 4/01.
del territorio turístico: El caso de la Costa Brava.
Lazerson, M., & Lorenzoni, G. (1999). The firms Retos Turísticos, 3(3), 8–15.
that feed industrial districts: A return to the Italian
Prats, L., Camprubí, R., & Comas, J. (2005). Net-
source. Industrial and Corporate Change, 8(2),
work ties relevance on the destination business
235–266. doi:10.1093/icc/8.2.235
relationships. In Universidade do Algarve (Eds.),
Lee, G. M., Cai, L. A., & O’Leary, J. T. (2006). Recent developments in tourism research. Faro:
An analysis of brand-building elements in the University of Algarve.
U. S. state tourism Web sites. Tourism Manage-
Prats, L., & Guia, J. (2005). The destination as
ment, 27, 815–828. WWW.Branding.States.US.
a local system of innovation. In C. Petrillo & J.
doi:10.1016/j.tourman.2005.05.016
Swarbrooke (Eds.), Networking and partnership
Lundvall, B. A. (Ed.). (1992). National systems of in destinations and development management (pp.
innovation. Towards a theory of innovation and 121-136). Arnhem: ATLAS.
interactive learning. London: Pinter Publishers.
Prats, L., Guia, J., & Molina, F. X. (2007). Tour-
Mazanec, J., & Schweiger, G. (1981). Improved ism local innovation systems or how tourism
marketing efficiency through multi-product brand destinations evolve. In M. Smith & L. Onderwater
names? An empirical investigation of image trans- (Eds.), Destinations revisited. Perspectives on
fer. European Research, 9(1), 32–44. developing and managing tourist areas (pp. 35-
53). Arnhem: ATLAS

1067
E-Tourism Image

Rallet, A., & Torre, A. (2004). Proximité et lo- Sorensen, F. (2004). Tourism experience innova-
calisation. Economie Rurale, 280. tion networks. Doctoral dissertation. Roskilde
University, Denmark.
Ritchie, J. R. (1993). Crafting a destination vision:
Putting de concept of residence responsive tourism Sundbo, J. (1998). The theory of innovation: En-
into practice. Tourism Management, 14, 279–289. trepreneurs, technology and strategy. Cheltenham,
doi:10.1016/0261-5177(93)90006-7 UK: Edward Elgar.
Roberts, J. (2001). Challenges facing service Telisman-Kosuta, N. (1989). Tourism destination
enterprises in a global knowledge-based econ- image. In S. F. Witt & L. Moutinho (Eds.), Tour-
omy: Lessons from the business services sector. ism marketing and management handbook (pp.
International Journal of Services Technology 557-561). Cambridge: Prentice Hall.
and Management, 2(3), 402–433. doi:10.1504/
Urry, J. (1990). The tourist gaze. London: Sage:
IJSTM.2001.001612
Wellman, B. (1997). An electronic group of virtu-
Selby, M., & Morgan, N. G. (1996). Reconstructing ally a social network. In S. Kiesler (Ed.), Culture
place image: A case study of its role in destina- of the internet. Mahwah: Lawrence Erlbaum.
tion market research. Tourism Management, 17,
287–294. doi:10.1016/0261-5177(96)00020-9

This work was previously published in Information Communication Technologies and City Marketing: Digital Opportunities
for Cities Around the World, edited by M. Gasco-Hernandez; T. Torres-Coronas, pp. 184-202, copyright 2009 by Information
Science Reference (an imprint of IGI Global).

1068
1069

Chapter 4.11
Successful Web-Based IT
Support Services:
Service Provider Perceptions of
Stakeholder-Oriented Challenges
Vanessa Cooper
RMIT University, Australia

Sharman Lichtenstein
Deakin University, Australia

Ross Smith
RMIT University, Australia

Abstract and concerns of other key stakeholders, and (2)


significant challenges exist in doing so. The main
Web-based self-service systems (WSSs) are in- contribution of the paper is the identification of the
creasingly leveraged for the delivery of after-sales key challenges involved. Important implications
information technology (IT) support services. for theory and practice are discussed.
Such services are offered by IT service provid-
ers to customer firms and increasingly involve
business partners. However little is known of INTRODUCTION
the challenges faced by IT service providers as a
result of the involvement of the other firms and The continued maturation of the Internet has
their employees (end-users). This paper reports been accompanied by a corporate shift from the
related findings from an interpretive study of IT provision of goods to the provision of services,
service provider perceptions in six multinational with parallel development of relevant new business
IT service provider firms (Cooper, 2007). The models and marketing paradigms (Rust, 2001).
findings highlight that, for IT service providers, Many businesses have developed E-services,
(1) it is important to consider and resolve the needs defined as the provision of services by electronic

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Successful Web-Based IT Support Services

networks such as the internet (Rust, 2001). Despite This article explores the context of managed
the increasing importance of E-services to busi- information technology (IT) support services. In
ness success, electronic commerce researchers this setting IT service providers employ WSSs
have been slow to investigate associated issues. to provide after-sales IT support to enterprise
As the E-services value chain requires different customers.
types of processes and offers greater flexibility in Key stakeholders comprise the service pro-
comparison with offline services, there are new vider firm and its employees, business partners
research challenges to be explored (Hofacker et and their employees, and enterprise customers
al., 2007). and their employees. As this article will show,
An important new source of value presented the involvement of the key stakeholders results
by E-services is supplementary E-services such in significant challenges for IT service providers
as electronic provision of pre- and post-sales cus- aiming to provide successful after-sales support
tomer support for purchased services and products by means of a WSS. These challenges will be
(Hofacker et al., 2007). Experts further suggest explored in the article by examining the IT service
that the successful provision of supplementary provider perspective.
E-services may be more important strategically to A knowledge transfer lens is used to explore
service providers and vendors than the quality of this topic as the transfer of after-sales IT support
originally-purchased services and products (Otim knowledge (such as IT solutions) from an IT ser-
& Grover, 2006; Piccoli et al., 2004). Marketing vice provider firm to a customer firm is central to
of supplementary services (offline and online) the concept of successful after-sales Web-based
can provide differentiation, improve customer support services (CSI 2002; Koh et al., 2004).
service, increase customer retention and lower This article draws on a large study investigating
service costs (Levenburg & Klein, 2006; Reich- the successful provision of managed after-sales
held & Schefter, 2000). IT support when facilitated by WSSs (Cooper,
This article focuses on the provision by ser- 2007). The perspectives of six large multinational
vice providers of supplementary E-services to IT service providers were obtained and analysed.
customer firms (“enterprise customers”) using the The views of IT service providers are important to
World Wide Web (“Web”). To leverage this market understand for improved service provision (Pitt,
successfully, vendors and service providers aim 1998). Our study focuses on the use of operational
to improve the implementation and delivery of E- IT support services, relating to (1) assembling
services by employing a systematic approach. One and operating the core IT environment, and (2)
such approach is a Net-based Customer Service providing key value-adding services such as the
System (NCSS) which has been described as “a Service (Help) Desk (Peppard, 2001).
network-based computerised information system Five further sections complete this article.
that delivers service to a customer either directly Section 2 provides a theoretical background by
(e.g. via a browser, PDA, or cell phone) or indirectly reviewing representative literature. Section 3
(via a service representative or agent accessing outlines the research design. Section 4 describes
the system)” (Piccoli et al., 2004 p.424). the key challenges relating to stakeholders,
This article focuses on the use of a key type of identified when an IT service provider transfers
NCSS based on a Web interface – a Web-based after-sales IT support-oriented knowledge to
Self-Service System (WSS). Self-service is gain- enterprise customers when WSSs are used to
ing importance in contemporary organisations facilitate service provision. Section 5 discusses
primarily for cost reduction reasons (Doyle, 2007). the key challenges. Section 6 summarises the

1070
Successful Web-Based IT Support Services

main points, draws conclusions, reflects on the mining tools which record past customer trans-
limitations of the findings and offers suggestions actions and analyse data to identify customer
for future research. segments, match products to customer profiles,
and better understand target demographics and
psychographic characteristics (Brohman et al.,
THEORETICAL BACKGROUND 2003). Customer service agents can utilise this
“customer intelligence” to potentially up-sell or
We first situate WSSs within a Customer Re- cross-sell products and services (Brohman et al.
lationship Management (CRM) context as the 2003). A WSS is an important type of operational
strategic goals of supplementary services - such eCRM application (Geib et al., 2005; Khalifa &
as after-sales support provision using WSSs - in- Shen, 2005) and is discussed below in the context
clude improving customer service and increasing of managed IT support services.
customer retention (Levenburg & Klein, 2006).
We then review the use of WSSs for after-sales IT WSS and After-Sales IT Support for
support provision to enterprise customers. Next Enterprise Customers
a stakeholder-oriented framework of successful
Web-based enterprise customer service drawn A WSS is a Web-based information system that en-
from earlier findings highlights the importance of ables organisations to move from labour-intensive
stakeholder relationships and related knowledge manual processes towards low-cost automated
flows. Finally, the section reviews the knowledge Web-based self-service (Pujari 2004). WSSs can
transfer process and the transfer of after-sales sup- facilitate the offering of customer support services
port knowledge from an IT support organisation for pre-sales, sales, and after-sales activities. They
to an enterprise customer. are underpinned by complex information systems,
complemented by a customer contact centre, and
Customer Relationship integrated with a multi-channel service strategy
Management (Negash et al. 2003).
WSSs offer important advantages to service
In recent years CRM has emerged as a poten- providers and customers (Geib et al., 2006; Pujari
tially powerful organisational strategy to enable 2004). Such advantages include electronically
a vendor or service provider to better identify leveraging the Web interface, customer/service-
and satisfy customer needs and retain customer provider (and customer/customer) interactions,
loyalty. Enhanced customer relationships may also knowledge management (KM) principles and
lead to improved customer-related operational self-service principles in order to capture and
effectiveness and a higher return on investment provide information and knowledge useful for
for the organisation (Barua et al., 2004). pre-sales, sales and after-sales support.
To improve the customer service experience In managed IT service environments, WSSs
and meet other CRM objectives, the aggrega- offer informational, transactional, and proactive
tion of data, information and knowledge about support services. Informational support includes
the customer is important. Specialised software “break-fix” support which provides customers
applications that perform electronic CRM experiencing technical problems with resolutions
(eCRM) have been developed for this purpose. to their problems. This type of support includes (1)
A common example of an eCRM application is unassisted support such as answers to Frequently
a Web interface, supported by database and data Asked Questions (FAQs) and the download of

1071
Successful Web-Based IT Support Services

software patches and (2) assisted support such cal success factors (CSFs) and challenges involved
as peer-to-peer online fora, e-mail and online should be identified and addressed.
chat. Other informational support includes the The study that is reported in this article iden-
provision of information and knowledge to assist tified a set of CSFs (Cooper et al, 2005, 2006a,
with enquiries and enable customers to access 2006b; Cooper, 2007) and also identified important
best-practice – for example, by the publication challenges, as perceived by IT service providers,
of White Papers. Transactional support includes many of which relate to other key stakeholders. In
case tracking, whereby the customer initially the next section we develop a stakeholder-oriented
documents their IT problem scenario, requests framework which situates WSSs in a broader en-
assistance from the support organisation, and terprise customer service context and highlights
is subsequently able to monitor the support or- the interactions with stakeholders.
ganisation’s progress in resolving this problem.
Proactive support includes the embedding of
problem detection support software on customer Stakeholder-oriented
end-user computers, and personalised messages Relational Web-based
directing the customer to potential product or Enterprise Customer Service
service purchases. In the study reported in this
article we consider primarily informational and A broad definition of stakeholders is “all those
transactional supplementary support services. parties who either affect or who are affected by
WSSs can reduce customer support expenses an organisation’s actions, behaviours and policies”
by empowering employees (Support Agents). (Mitroff, 1983). Management concerns about
These employees provide support in the form stakeholders arise because stakeholders have
of knowledge such as solutions to customers’ IT varying perspectives of the underlying problems
problems. WSSs boost Support Agent productivity and their ideal solutions may differ. It is important
by providing access to a comprehensive knowledge to business success that a firm resolve conflicting
base that includes many solutions, delivering con- stakeholder needs (Hatch, 1997).
venient and higher quality support, and increasing Experts report that the needs, roles, respon-
customer loyalty and retention. sibilities, relationships and other interactions of
Non-IT businesses outsource IT support serv- stakeholders are especially important to the suc-
ices to service providers because IT is not their core cess of business-to-business (B2B) commerce
competency, service levels are likely to improve, (Chua et al., 2005; Kandampully, 2003; Pan,
and Total Cost of Ownership should be reduced 2005; Ritter & Gemunden, 2003; Schultze &
(CRMInd, 2006; SSPA, 2004). Furthermore, client Orlikowski, 2004; Singh & Byrne, 2005). In B2B,
firms are often receptive to Web-based support businesses are increasingly interdependent and
delivery (SSPA, 2004). Thus the Web is considered the stakeholder issues must be carefully managed
by customer firms as an important channel for IT (Kumar & van Dissel, 1996). However stakeholder
support delivery. roles and responsibilities along the value chain
Despite the potential benefits of a WSS to are complex (Chi & Holsapple, 2005; Ritter &
service providers and enterprise customers, there Gemunden, 2003). Chi and Holsapple propose a
has been a long-reported dissatisfaction with Web- model of stakeholder collaboration in B2B high-
based self-service (Barnes et al., 2000; Meuter et lighting three behavioural processes: knowledge
al., 2003; Ragsdale, 2007). Clearly success from sharing, participative decision-making and con-
the use of a WSS cannot be assumed and the criti- flict governance. Other reasons for understanding

1072
Successful Web-Based IT Support Services

stakeholder issues in B2B commerce include the shows three key types of stakeholder organisations
need to manage stakeholder expectations (Singh which may be involved in support provision – a
& Byrne, 2005). support organisation (previously termed “ser-
Stakeholder relationships in services are vice provider”), business partner, and customer
different to mere service encounters and are as- organisation – and their interaction with one an-
sociated with emotions and expectations (Gutek other directly and indirectly via a WSS. At each
et al., 1999). Such relationships and associated organisation there are corporate entity represen-
knowledge flows are important in managed IT tatives (for example, managers) interacting with
services environments (Dahlberg & Nyrhinen, end-users. The framework clearly highlights the
2006; Xu, 2007) as the quality of knowledge interdependencies found in the multi-stakeholder
exchange influences the quality of outsourcing managed IT support environment.
relationships (Gong et al., 2007). Of interest to
this article, the adoption of WSSs can potentially Knowledge Transfer for IT Support
strengthen stakeholder relationships (Bhappu & of Enterprise Customers Using WSS
Schultze, 2006).
From the evidence above and a review of the The key knowledge process explored in this ar-
literature (Cooper, 2007) we have developed a ticle is inter-organisational knowledge transfer.
framework (Cooper et al., 2006a) that concep- We focus on reviewing staged processual and
tualises key stakeholders and relationships in a network models of knowledge transfer for reasons
B2B service context. Such a framework can be of relevance.
helpful in understanding complex knowledge Researchers advocating a staged processual
transfer from service providers to customer firms. approach to knowledge transfer argue that this
As mentioned earlier, such knowledge transfer can unlock the inner workings of the process and
is central to the provision of IT support. The enable a more nuanced identification of barriers
framework in Figure 1 depicts key relationships and enablers (Szulanski 1996). Staged inter-
between the main stakeholder types and a WSS. It organisational knowledge transfer models include

Figure 1. A stakeholder-oriented relational framework for web-based enterprise customer service


(Cooper et al, 2006a)

sO
c u

Wss

bP cO
c u c u

LEGEND
SO = Support Organisation c = Corporate Entity WSS = Web-based Self-service System
BP = Business Partner Organisation u = End-User Entity ___ = Relationship via WSS
CO = Customer Organisation - - - = Direct Relationship

1073
Successful Web-Based IT Support Services

Figure 2. Knowledge transfer in managed after-sales Web-based IT support (Cooper et al, 2006a)

C ustom er B usiness
P artner

Support
Organisation

S ervice D esk
S upport
A gents
WSS

W S S Interface and
P rocesses
S upport
E ngineers

K now ledge B ase

those developed by Cranefield and Yoong (2005) Lubatkin, 1998) between the two firms involved
and Nieminen (2005). Cranefield and Yoong’s in the transfer. Researchers proposing a network-
(2005) model comprises six stages: engaging; ing approach to knowledge transfer also adopt a
defining; seeking; articulating; integrating; and relational approach based on understanding how
disseminating. Among the benefits of this model is patterns of connections between individuals and
its identification of key organisational influences groups facilitate knowledge sharing and knowl-
on knowledge transfer during each stage. However edge transfer (Kakabadse et al., 2003).
the factors identified as most relevant to knowledge We next describe the knowledge transfer pro-
transfer in an inter-organisational context are the cess in the managed IT support context (Figure
need for fit between the transferred knowledge 2) drawing on literature and empirical data from
and the receiving organisation’s current organi- a case study (Cooper, 2007). IT service provid-
sational objectives and traditional discipline area, ers respond to enterprise customers’ after-sales
and the need to avoid use of “non-transferable enquiries, incidents and problems regarding core
examples” which cannot be readily transferred to IT products and services by providing and transfer-
other organisational contexts. Nieminen’s (2005) ring support-oriented knowledge. The knowledge
model of inter-organisational knowledge transfer is complex and based on solutions to the enquiries,
focuses on the role of a receiving organisation in incidents and problems reported.
enabling knowledge transfer. He writes of the Both tacit and explicit knowledge may be
need for a receiving organisation to have a range transferred. Tacit knowledge is knowledge that
of competencies among its employees in order to resides in the mind and is difficult to articulate.
reduce any significant absorptive gap (Lane & Explicit knowledge is a representation of knowl-

1074
Successful Web-Based IT Support Services

edge. An example of explicit knowledge in the and a focusing workshop (Rockart, 1979) with
managed after-sales IT support context is an IT participation by five senior customer service man-
solution stored in a knowledge base. agers and IT support managers. Case study data
As shown in Figure 2, when a customer firm were analysed using qualitative content analysis
experiences IT incidents or problems, IT profes- techniques (Mayring 2000) by two researchers
sionals at the firm may use the telephone channel working independently to identify the key chal-
or Web interface to obtain a solution. When the lenges relating to stakeholders. A set of CSFs for
telephone is chosen, first tier Support Agents the transfer of after-sales IT solution-oriented
identify potential solutions by accessing tacit knowledge to enterprise customers when a WSS
knowledge or by using the WSS to search the is used, and a set of key challenges, have been
solutions knowledge base. More complex prob- identified.
lems are escalated to more experienced Support In 2006 a focus group of Australian senior
Agents. Tiers of Support Engineers resolve the customer service and IT support managers from
most difficult problems by drawing on valuable an additional five large multinational IT service
tacit knowledge. New and evolving solutions provider organisations, all with successful WSSs,
are captured in the knowledge base as explicit was conducted, seeking to confirm and extend the
knowledge and organised for reuse. Successive findings from the ServIT case study. In the focus
efforts are made by Support Agents to address group discussion it became clear that, while the
related customer questions. Such efforts aim WSS of each company was successful, various
to assist the customer firm in institutionalising challenges remained. Two researchers working
support-oriented knowledge. Business partners independently confirmed and extended the set of
may assist in the knowledge transfer process by CSFs and key challenges working inductively from
providing partial solutions or other support and the transcript. Many of the challenges were found
may have access to the service provider’s WSS to be related to other stakeholders and it is these
knowledge base. challenges which are the focus of this article.
The quality of the findings was assured by
appropriate techniques. Achieving validity in
RESEARCH DESIGN qualitative research requires a fair, honest and
balanced account from the viewpoint of someone
Initially a comprehensive literature review was who lives in the social situation (Neuman, 2006).
undertaken. Next, in 2005 an in-depth case study Construct validity was assured in the study by
was conducted at the Australian headquarters of use of multiple sources of evidence. Participants
a large best-in-class multinational IT services were also given copies of interview summaries
organisation, “ServIT” (a pseudonym). ServIT and the outcomes from each phase of the research
had previously secured prestigious global awards were provided to participants for confirmation.
for its WSS. It defined a successful WSS as one Reliability was increased via a case study protocol
that delivers increased customer satisfaction while (Yin 2003) to document all procedures and prob-
reducing service provision costs and this defini- lems. Triangulation was achieved by collecting
tion of success was used throughout the project. and analysing data from a variety of sources at
Case data was collected from semi-structured ServIT and establishing consistency of results.
interviews with twelve senior customer service The results of the cross-organisational focus group
managers, IT support managers and support staff, added another level of triangulation.
observations of WSS use, document collection,

1075
Successful Web-Based IT Support Services

FINDINGS These six categories were believed by the IT


service provider participants in the study to apply
The six participating IT service providers utilised to all three stakeholder organisations (that is, those
WSS strategies to increase customer satisfaction shown in Figure 1). This finding is discussed fur-
while reducing the cost of IT support provision. ther in the next section, as it presented one of the
Each firm offered a suite of informational, trans- main challenges for IT service providers. Further
actional and proactive support services using details of the CSFs may be found in earlier publi-
WSSs. Twenty-seven CSFs were identified. While cations (Cooper et al, 2005, 2006a, 2006b).
the CSFs are not the focus of this article they are
summarised for reference in the Appendix. The
CSFs were also classified non-orthogonally into SERVICE PROVIDER PERCEPTIONS
six categories: OF STAKEHOLDER-ORIENTED
CHALLENGES
• Organisational Commitment and Readiness:
The organisation must manage the policies, This section summarises the eight major challeng-
processes and cultural issues which will af- es faced by IT service provider firms, pertaining
fect its ability and willingness to embrace to key stakeholders. Representative quotes from
Web-based Self-service. the ServIT case study and cross-organisational
• Manage for Strategic and Operational focus group illuminate the challenges. Company
Benefits: The WSS strategy must assist the names are pseudonyms for reasons of anonymity.
organisation in attaining its strategic and The reader is reminded that the findings have been
operational objectives. derived from the perspective of the IT support
• KM Capabilities and Processes: The or- firm only. A further comment on this limitation
ganisation must practice the principles of is made at the end of the article.
knowledge management and implement as- First, participants strongly believed that all
sociated knowledge management processes, stakeholder viewpoints should be considered
to maximise the benefits received from the by an IT support organisation when planning,
WSS strategy. implementing and managing a WSS for after-
• IT Infrastructure Capability: The organisa- sales IT support provision. Similar findings were
tion must have an adequate IT infrastructure made for electronic business settings more gen-
in place, to enable it to participate in Web- erally by Kandampully (2003) and Singh and
based Self-service. Byrne (2005). However our study highlighted
• Experience Management: The WSS should the potential for different stakeholder types to
manage the stakeholder’s experience, both hold unique perceptions. For example, while a
at the corporate and end-user level. The support organisation may find the transfer of IT
stakeholder experience will directly affect solutions to a customer firm highly desirable, it
satisfaction levels and therefore ongoing use was questioned by IT service provider participants
of the WSS. whether end-users at a customer enterprise would
• Content: The WSS must contain useful, feel the same way:
accurate and up-to-date content in order
to resolve the end-user’s support issue or I think the provider [support organisation] is
knowledge requirement. interested in transferring knowledge so [that]

1076
Successful Web-Based IT Support Services

they don’t have a problem any more and they can to a support organisation, when a WSS is used,
manage their costs and help the customers. [How- is not the same as cost-effectiveness from the
ever, the customer firm’s end-user is thinking] I perspective of an end-user at the customer firm.
am interested, not in receiving knowledge… [but] End-users are concerned with efficiency gains
I am interested in my problem being fixed. Don’t and usefulness of the knowledge gained from the
give me all this stuff [the details of the problem and WSS to perform their jobs (Cenfetelli et al., 2005),
solution]. Tell me what the problem is so I can fix rather than the financial costs involved.
it. I don’t want the transferring of any knowledge.
(Senior IT Architect, I-Systems) Cost effectiveness, from both the end-user of the
service and for ServIT, are equally important, but
This is an important finding deserving of I think you will find some subtleties in [how they
further research. If an end-user at a customer are] both explained as cost-effectiveness. For an
firm does not wish to learn from a provided IT end-user it’s ‘Can I actually get the results quickly
solution, how can a support organisation ensure from my perspective?’ From ServIT’s perspec-
such learning? Proposed cost and efficiency tive, it is ‘Can we actually reduce the cost of this
benefits to a support organisation stemming from service for our customers so that we are actually
knowledge transfer may have negative implica- making a profit? (Consulting Services Division
tions for customer satisfaction. It should be noted Manager, ServIT)
however that end-users at a customer firm have
specialised job roles with role-based knowledge Third, as described earlier, the six CSF catego-
needs. Thus, for example, while some end-users ries were found to apply to all three stakeholder
may only desire a resumption of IT operations organisations. This finding supports prior re-
using a supplied solution, other personnel such as search by Schultze and Bhappu (2005) who note
Database Administrators are likely to be highly that customers are often partly responsible for
interested in learning about a solution and gain- a service provider’s success. Clearly, a support
ing more general support knowledge which could organisation will have very little control over
be useful at a future time. On the other hand if whether a customer organisation and business
a professional at an enterprise customer has full partner organisation do, in fact, address the six
knowledge of a solution the customer organisa- CSF categories. As a result, performance mea-
tion may become increasingly independent of the surement and management of WSS strategies in
support organisation. Ultimately the success of a B2B context will be challenging.
the support organisation can be affected. Thus Kurnia and Johnston (2002) note the impor-
there may be a level of knowledge which a support tance of industry capability in internet trading
organisation may wish to retain and not transfer for corporate adoption of an inter-organisational
to customer firms. system. Thus if corporate customers and business
Second, participants questioned whether the partners in a managed IT support situation do not
different stakeholder types might interpret the manage the CSFs, the support organisation’s WSS
requirements for CSFs (Appendix) differently. For may not be successful. As electronic commerce
example, for CSF-9 Ease of use participants men- environments are increasingly interdependent
tioned that regular end-users of the WSS interface (Kumar & van Dissel, 1996; Kurnia & Johnston
would prefer an efficient interface, whereas novice 2002), a support organisation may benefit from
users would prefer easy-to-use interfaces. As a sec- providing further assistance to business partners
ond example, for CSF-1 Cost-effectiveness some and customer organisations to better manage the
participants mentioned that cost-effectiveness key factors.

1077
Successful Web-Based IT Support Services

One promising strategy may be for a support end-user trusts the Web-site and the organisation
organisation to conduct education and training behind it, this trust forms part of the relationship,
programs with partner and customer organisations with the WSS simply providing the connection:
to increase their awareness and understanding
of CSFs. Such education and training could be Although it is a piece of software, it is the front
provided by online support, training, newsletters end of a company. For instance, if I feel I have a
and so on. Indeed participants also identified the positive relationship with my Netbank Web-site
importance of education and training in terms because I trust it, then that is part of the relation-
of developing relationships with the customer ship. (IT Customer Consultant, DistSystems)
and ultimately meeting key objectives for CRM.
Thus education and training programs will not Others, however, saw Web-based relationships
only provide partner and customer firms with as perhaps one-way only, whereby it was ques-
the required knowledge to increase their aware- tioned whether there is a relationship between the
ness of WSSs but will provide an opportunity end-user and the providing organisation, when the
for the support organisation to better understand end-user uses the WSS anonymously.
the needs of the customer and business partners Fifth, it was also suggested that stakeholder
in a WSS context. This finding supports current relationships can be more complex than Figure
literature highlighting the importance of develop- 1 allows. Participants observed that relationships
ing relationships with enterprise customers when with stakeholder types may vary.
providing support services (Peppers & Rogers,
2001; Pujari, 2004). We can have the relationship with either both
A second strategy that is more challenging the customer and the business partner or we can
would be to ensure, when developing external have it directly with the customer or we could
service contracts, that customer and business part- just have it with the business partner…I mean
ner organisations are obliged to meet minimum there are just so many different combinations that
standards relating to CSFs. Any failure to address that relationship can actually take. (IT Customer
CSFs at customer firms and business partners Consultant, DistSystems)
would affect the ability of a support organisation
to service them effectively. However such a strat- Consequently in some instances Figure 1
egy may be incompatible with the development should be modified. For example, for support-
of improved relationships with business partners ing some customer firms a business partner is
and customer firms. It is also important to recall involved, while for supporting other customers
that this study only investigated the IT support there are no business partners involved.
organisation perspective. While IT support or- The complexity of relationships would also
ganisations identified the CSFs as important also increase as stakeholder organisations increase in
for partner and customer firms to address, studies size. Multiple relationships would be developed at
of partner and customer perspectives of CSFs the individual, departmental and corporate levels.
may provide quite different results. This question This finding is significant as while it has been
should be explored in future research. acknowledged in a growing body of literature
Fourth, the interesting issue was raised of that developing relationships with partners is
whether relationships are possible—or even important (e.g. Vlachopoulou & Manthou, 2003),
enabled—by a WSS. A two-way relationship was there is very little literature concerning partner-
posited by several participants, whereby if an unit-based relationships.

1078
Successful Web-Based IT Support Services

Sixth, intellectual property (IP), security and organisation that does not have the same privacy
privacy issues were found problematic for stake- policy as you, and then [if they] were to use that
holder-oriented reasons. Inter-organisational firms information in such a way that it violates some-
in collaborative relationships must protect certain body’s privacy, who is ultimately responsible? That
knowledge which may be strategic (Solitander is a huge legal question that is being tested now.
2006). In the present study, service provider Does it come back to the original person or the
firms were concerned about the loss of IP in the person who let the information go? (Emerging
leverage of user fora for customer support. In such Technology Consultant, OpSys)
fora, end-users may provide solutions to reported
problems however it can be unclear who owns the Seventh, understanding the complexities of
IP that is the solution. Scholars have noted that in the business partner alliance was an issue. Not
customer co-production of service, as illustrated all service providers participating in the study
by this situation, new models of digital governance agreed with the distinction between a business
and customer-based innovation are needed (Rai partner relationship and a customer/supplier re-
& Sambamurthy, 2006). lationship. For example, a participant from one
IT support organisations, business partners and organisation claimed that his organisation did not
customers frequently operate across international have business partners, while others claimed that
borders. Support organisations must consider is- their organisations were, in fact, partners of his
sues of security, privacy and IP within the context organisation. It was advocated that the distinction
of off-shore environments (Rai & Sambamurthy, between a business partner and customer/supplier
2006; Tafti, 2005). The IT service providers in this relationship, surrounds whether the relationship
study expressed uncertainty about the ability of is a one-off transaction, or whether it is an on-
national legal systems to deal with such complex going relationship. In an enterprise IT services
issues. For example, customer end-users may be context, however, there is an apparent blurring of
unaware of the location where their personal in- relationships. In some projects, organisations may
formation is stored and retrieved. Further, while be considered business partners, while in others
customer organisations may be aware that, in they would be considered competitors:
some instances, third parties such as a support
organisation’s business partners have access to DistSystems, DataCorp, OpSys and I-Systems can
their stored personal information, they may not say, we have been working together as business
be aware of the eventual use of this information partners and as competitors and sometimes even
or its accessibility. A perceived potential privacy on the same bid…on the one hand you compete, in
violation may lead to unforeseen competitor the next 30 seconds, I might be talking to OpSys
problems – for example, where business partners about something we are competitors about, and
compete with a customer enterprise. Sensitive in another thing we are working together on, you
information, knowledge and IP may indeed be switch hats, you switch alliances, it is just the
compromised (Rai & Sambamurthy, 2006). way things are these days…. (Senior Systems
Consultant, I-Systems)
A really interesting one that is going through the
legal battles now and it comes back to privacy. Some of the complexities of business partner
As a main [support] organisation, you have in- alliances surround intellectual property and se-
formation that could be of a private nature. If you curity and privacy, discussed earlier. Lei (1997)
then make the information available to another argues that, with respect to strategic alliances,

1079
Successful Web-Based IT Support Services

regardless of the various types of legal structures ServIT invests significant resources in its legal
which may be put in place, over time companies team and explicitly states in the terms and condi-
will absorb and internalise skills, regardless of tions of using the WSS and in support contracts
the amount of formal, legal ownership that is that ServIT will not be responsible for any deg-
demarcated by the alliance structure. Further, radation of a customer’s systems, if an enterprise
Ferdinand and Simm (2007) suggest a need for customer decides to act upon information derived
increased research on illegal inter-organisational from the WSS. Another challenge is that while
knowledge transfer. Participants in our research ServIT monitors fora content, it relies more on
study also expressed a need for greater under- a “merit” based system whereby forum users
standing of this complex area. They believed it are allocated merit points by original posters (of
likely that associated concerns would continue in questions) when they provide valuable responses.
the foreseeable future. A system of points aggregation motivates users
Eighth, managing customer contributions to to share valuable knowledge and thus contribute
service was considered an important challenge. to E-service provision. However, the researchers
The growing literature on co-production high- found from forum observations that this scheme
lights the productivity benefits as well as the occasionally led to highly successful users mov-
managerial challenges that arise when customers ing on to create their own sites independently
become “partial employees” (Benapudi & Leone, of the service provider, such was their fame and
2003). The advantages of co-production include following.
decreased cost of service provision and greater Other potentially negative impacts of customer
control and autonomy for the customer, while co-contribution include potential defamation
disadvantages include the difficulty of control- in the fora. While the terms and conditions of
ling service quality when customers are actively posting to online fora state that users should not
involved in the production process (Schultze & post defamatory statements, such defamation
Bhappu, 2005). sometimes eventuates.
In the study IT service providers raised con-
cerns relating to the contribution of knowledge
by customers to their own service fulfillment, DISCUSSION
optimisation and improvement. At ServIT when
customers interacted and shared resolution- The findings above suggest a need for industry
knowledge in online for a the knowledge was not sector improvement. For the IT support industry
captured permanently. Further, the online fora to learn to do things better, learning at the in-
were open to a variety of end-users causing new dustry level is needed. Prior studies suggest the
issues of accuracy and liability: importance of business learning in a network
context (c.f. Knight, 2002). This would entail
ServIT has ventured into the hosting of forums. new collaborative electronic business projects
Solutions can be provided by non-ServIT people focused on IT support provision involving IT
and that’s a potential conflict between the reli- service providers, business partners and customer
ability of our knowledge and the fact that we organisations. According to Cameron’s (2005)
are opening up [knowledge] to end-users, which review of prior relevant studies, there are four
could have a good result, but there is a danger important influences to consider for successful
that we are facilitating the [incorrect] solution. electronic business collaboration: motivation,
(IT Manager, ServIT) capability, communication and coordination.

1080
Successful Web-Based IT Support Services

First, organisations must be motivated to partici- reach critical mass, and power imbalances among
pate in collaborative projects. Second, the desire participants (Driedonks et al., 2005).
to increase organisational capability (skills and The risk of reduced service quality was identi-
knowledge) can be a powerful motivator. Third, fied in the practice of customer co-contribution to
it is important to communicate about the value of service. This concern was noted more generally for
such collaborative projects within organisations. online service provision by Schultze and Bhappu
Fourth, industry groups can assist in coordinating (2005). There may be ways for IT service firms
such projects for successful conclusions. to better monitor customer contributions via user
As many of the key challenges identified in fora and capture the high quality solutions (which
the previous section centre on the potential for currently are not captured). Related questions of
stakeholder conflict, the development of individual knowledge ownership must be resolved. However
IT service contracts (including service require- regulatory issues at different geographic locations
ments) with customer firms and business partners cloud the resolution of such questions.
requires attention. This process should entail Indeed there were several areas identified
greater clarification and negotiation of stakeholder where IT service providers suffer from inad-
needs. The objectives of each organisation should equate regulatory support. The security, privacy
be articulated during joint planning, which is an and IP issues experienced in the often-offshore
important aspect of business collaboration in managed IT support environment highlight the
supply chains (Holsapple & Jin, 2007). Holsapple need for clarification and awareness of relevant
and Jin note several other important collaborative regulations and laws at industry, national and
decisions typically made in a supply chain. The international levels. IT support companies also
findings from our study suggest a need for focused seek to understand how to share knowledge with
studies seeking to identify key problem points business competitors within strategic alliances
in decision-making where stakeholder conflicts while maintaining a competitive advantage. New
are influential so that the overall process can theories are sought to underpin such knowledge
be improved to avoid or manage such conflicts. sharing in an increasingly collaborative global
Chi and Holsapple (2005) propose a model of business environment.
stakeholder collaboration in B2B highlighting The six IT service provider companies in our
three behavioural processes: knowledge shar- study uniformly noted that the key objective for
ing, participative decision-making and conflict use of a WSS for after-sales IT support was to
governance. Our study provides support for the increase customer satisfaction while reducing
importance of such processes. support costs. This objective was expected to be
The findings and discussion above also suggest achieved by relational as well as transactional
a need for supporting infrastructure. Electronic methods. However as the study showed, the suc-
marketplaces offer a recognised structure which cessful accomplishment of both goals can be
may be useful to support both collaborative problematic (also found by Bunduchi, 2005).
IT support projects and specific supply chains Research is beginning to appear on the enabling
(Markus & Christiaanse, 2003). However there are of relationships in service provision when the
also challenges in using electronic marketplaces internet is used to facilitate service provision.
successfully. In a recent case study of an elec- In a business to consumer (B2C) context, Sigala
tronic marketplace, the key influences affecting (2007) explored online travel service provision and
its success were the loss of social capital, nature found that the communication aspect of the online
of communication channels used, time taken to service played a key role in relationship develop-

1081
Successful Web-Based IT Support Services

ment. Where provider-consumer communication • All stakeholder viewpoints should be con-


was enhanced by use of relevant communication sidered by an IT support organisation when
tools, relationships and client satisfaction were planning, implementing and managing a
improved. Thus IT support organisations may WSS for after-sales IT support provision;
find that the path to satisfying the relational as- • Different stakeholder types might interpret
pect of WSS success is by better leveraging the the requirements for CSFs differently;
internet’s communication tools and customising • The six CSF categories identified apply to
communication-oriented content. all three stakeholder organisations, however
Some of the challenges identified in our study a support organisation will have very little
relate to the process of knowledge transfer. For control over whether a customer organisa-
example, it was noted that a customer firm end- tion and business partner organisation do,
user may not be interested in institutionalizing in fact, address these categories. Thus per-
transferred IT solutions. Cranefield and Yoong formance measurement and management
(2005), in identifying key challenges in inter- of WSS strategies in a B2B context will be
organisational knowledge transfer, highlighted difficult;
the need for a fit between the knowledge received • It is problematic whether relationships are
and the receiving organisation’s objectives. If a possible, or even enabled by a WSS;
customer firm does not prioritise the institution- • Stakeholder relationships can be more com-
alization of IT solutions throughout the firm, its plex than Figure 1 allows;
employees will not make the effort to learn the • Intellectual property, security and privacy
solutions transferred from the service provider issues can be problematic for stakeholder-
and will simply apply them to resolve the initial oriented reasons;
problem. This suggests a need for IT service • Understanding the complexities of the busi-
providers to educate their customers about the ness partner alliance can be an issue; and
importance of institutionalizing IT solutions to • Managing customer contributions to service
their organisation. Nieminen (2005) noted that is an important challenge.
for knowledge transfer to take place, a receiving
organisation must be capable of absorbing shared The findings demonstrate that best-in-class
knowledge. Such capability may be missing IT service providers face diverse challenges to
from the customer firms which are receiving IT better understand and resolve potential conflicting
solutions. stakeholder needs in this context.
Theoretically this article provides numerous
insights into the key challenges faced by IT service
CONCLUSION providers, relating to the different stakeholders,
in the provision of B2B after-sales IT support
This article has identified and discussed eight services via WSSs. The article also highlights a
major stakeholder-oriented challenges in the need for new theories which integrate WSS strat-
provision of managed after-sales IT support ser- egies across the multiple stakeholders involved.
vices via WSSs to enterprise customers, from the Our findings further suggest that relationship
perspective of multinational IT service providers. development by WSSs is poorly understood by
Specifically: the companies involved and that further research
is needed to develop new understandings.

1082
Successful Web-Based IT Support Services

This article also assists IT service providers Barua, A., Konana, P., Whinston, A.B. and Yin,
by recommending that all stakeholder viewpoints F. (2004). An Empirical Investigation of Net-
and issues should be considered when planning, enabled Business Value. MIS Quarterly. 28(4),
implementing and managing WSSs in the man- December, 585-620.
aged after-sales IT support context. Addressing
Benapudi, N. and Leone, R.P. (2003). “Psycho-
diverse stakeholder needs may be particularly
logical Implications of Customer Participation
challenging in some areas such as security, privacy
in Co-production”, Journal of Marketing. 67(1),
and IP. Such emerging challenges highlight the
January, 14-28.
complexities of working with business partners
which are also considered competitors. Greater Bhappu, A.D. and Schultze, U. (2006). The Role
collaboration is needed with better supporting of Relational and Operational Performance in
infrastructure and regulation. Business-to-Business Customers’ Adoption of
While the findings from this article are limited Self-Service Technology. Journal of Service
by the context (managed after-sales IT support) Research. 8(4), 372 – 385.
and scope (the IT support organisation perspective
Brohman, M.K., Watson, R.T., Piccoli, G. and
only was studied), they are indicative of possible
Parasuraman, A. (2003). Data Completeness: A
concerns that other types of service providers may
Key to Effective Net-based Customer Service
have in offering supplementary E-services using
Systems, Communications from the Association
WSSs. Thus the findings provide a foundation
for Computing Machinery. 46(6), 47-51.
for exploration in other settings. Investigating
the customer and business partner perspectives Bunduchi, R. (2005). Business Relationships in
would also provide valuable balance to the views Internet-based Electronic Markets: The Role of
expressed and analysed in this article. Goodwill Trust and Transaction Costs. Informa-
tion Systems Journal. 15 (4), 321–341.
Cameron, J. (2005). Ten Concepts for an eBusiness
ACKNOWLEDGMENT
Collaborative Project Management Framework. In
Proceedings of eIntegration in Action: 18th Bled
This article is a significantly extended and updated
eConference, June 6-8 2005, Bled, Slovenia.
version of a article published in the Proceedings
of the 20th Bled eConference (Bled07), Bled, Slo- Cenfetelli, R., Benbasat, I. and Al-Natour, S.
venia, 4-6 June 2007. The authors are grateful to (2005). Information technology mediated cus-
the anonymous reviewers whose valuable advice tomer service: A functional perspective. In Pro-
helped improve the quality of this article. ceedings of ICIS 2005, Las Vegas, USA.
Chi, L. and Holsapple, C. (2005). Understanding
Computer-mediated Interorganisational Col-
REFERENCES
laboration: A Model and Framework. Journal of
Knowledge Management. 9(1), 53-75.
Barnes, J.G., Dunne, P.A. and Glynn, W.J. (2000).
Self-service and Technology: Unanticipated and Chua ,C.E.H., Khoo, H.M., Straub, D.W., Kadiyala,
Unintended Effects on Customer Relationships. S. and Kuechler, D. (2005). The Evolution of E-
Handbook of Services Marketing and Manage- commerce Research: A Stakeholder Perspective.
ment, Swartz, T.A. and Iabocobucci, D. (eds.), Journal of Electronic Commerce Research. 6(4),
Sage Publications, Thousand Oaks, California, 262-280.
89-102.

1083
Successful Web-Based IT Support Services

Cooper, V.A. (2007). Knowledge Transfer in En- Dahlberg, T. and Nyrhinen, M. (2006). A New
terprise IT Support Provision using Web-based Instrument to Measure the Success of IT Out-
Self-Service, unpublished PhD Thesis, Deakin sourcing. In Proceedings of the 39th Hawaii
University, Melbourne, Australia. International Conference on System Sciences
(HICSS’2006), IEEE Society Press.
Cooper, V.A., Lichtenstein, S. and Smith, R.
(2006a). Enabling the Transfer of Information Doyle, S. (2007). Self-service Delivery and the
Technology Support Knowledge to Enterprise Growing Roles of Channels. The Journal of Database
Customers Using Web-based Self-service sys- Marketing & Customer Strategy Management. 14(2),
tems: Critical Success Factors from the Support January , 150-159.
Organisation Perspective, in Proceedings of Sev-
Driedonks, C., Gregor, S. and Wassenaar, A.
enteenth Australasian Conference on Information
(2005). Economic and Social analysis of the
Systems (ACIS 2006), Adelaide, Australia.
Adoption of B2B Electronic Marketplaces: A
Cooper, V., Lichtenstein, S. & Smith, R. (2006b) Case Study in the Australian Beef Industry.
Knowledge transfer in enterprise information International Journal of Electronic Commerce.
technology support using web-based self-service 9(3), Spring, 49-72.
systems, International Journal of Technology
Ferdinand, J. and Simm, D. (2007). Re-theorizing
Marketing, 1(2), 145-170.
External Learning: Insights from Economic and
Cooper, V.A., Lichtenstein, S. & Smith, R. (2005) Industrial Espionage. Management Learning.
Emerging Issues in After-sales Enterprise Infor- 38(3), 297-317.
mation Technology Support Using Web-based
Geib, M., Kolb, L. and Brenner, W. (2006). Col-
Self-service Systems, in Proceedings of Sixteenth
laborative Customer Management. Financial Ser-
Australasian Conference on Information Systems
vices Alliances. In (Fjermestad, J. and Romano,
(ACIS 2005), University of Technology Sydney,
N.C. (eds)), Electronic Customer Relationship
30 November – 2 December, 2005, Sydney.
Management, Armonk, New York.
Cranefield, J. and Yoong, P. (2005). Organisational
Gong, H., Tate, M. and Alborz, S. (2007). Manag-
Factors Affecting Inter-organisational Knowledge
ing the Outsourcing Marriage to Achieve Success.
Transfer in the New Zealand State Sector: A
In Proceedings of Hawaii International Confer-
Case Study. The Electronic Journal for Virtual
ence on the System Sciences (HICSS’07), IEEE
Organisations and Networks. 7, December.
Society Press.
[CRMInd] CRMIndustry.com (2006). 2006 State
Gutek, B., Bhappu, A.D., Liao-Troth, M.A. and
of IT Outsourcing. Summary report, <http://www.
Cherry, B. (1999). Distinguishing between Service
crmindustry.com/industry_research/outsourcing.
Relationships and Encounters. Journal of Applied
htm> (Accessed 03 June 2008).
Psychology. 84(2), 218-233.
[CSI] Consortium Service Innovation (2002).
Hatch, M.J. (1997). Organisation Theory: Mod-
Getting Started with KCS. Official Consortium
ern, Symbolic and Postmodern Perspectives.
Service Innovation Site, <http://www.thinkhdi.
Oxford:UK, Oxford University Press.
com/files/pdfs/GettingStartedKCS.pdf> (Ac-
cessed 03 June 2008). Hofacker, C. F., Goldsmith, R.G., Bridges, E.
and Swilley, E. (2007). E-Services: A Synthesis
and Research Agenda. Journal of Value Chain
Management. 1(1/2), 13-44.

1084
Successful Web-Based IT Support Services

Holsapple, C.W. and Jin, H. (2007). Connecting Lei, D.T. (1997). Competence-building, Technol-
Some Dots: E-commerce, Supply Chains and ogy Fusion and Competitive Advantage: The Key
Collaborative Decision Making. Decision Line. Roles of Organisational Learning and Strategic
38(5), October, 14-21. Alliances. International Journal of Technology
Management. 14(2/3/4), 208-37.
Kakabadse, N.K., Kakabadse, A. and Kouzmin,
A. (2003). Reviewing the Knowledge Manage- Levenburg, N.M. and Klein, H.A. (2006). Deliv-
ment Literature: Towards a Taxonomy. Journal of ering customer services online: identifying best
Knowledge Management Practice. 7(4), 75-91. practices of medium-sized enterprises. Informa-
tion Systems Journal. 16(2), 135-155.
Kandampully, J. (2003). B2B Relationships and
Networks in the Internet Age. Management Deci- Markus, M.L. and Christiaanse, E. (2003). Adop-
sion. 41(5), 443-451. tion and Impact of Electronic Marketplaces.
Information Systems and E-Business Management. 1(2),
Khalifa, M. and Shen, N. (2005). Effects of
January, 139-155.
Electronic Customer Relationship Management
on Customer Satisfaction: A Temporal Model. In Mayring, P. (2000). Qualitative Content Analysis,
Proceedings of the 38th Annual Hawaii Interna- Forum Quality Social Research, (1)2. <http://
tional Conference on System Sciences (HICSS’05), www.qualitative-research.net/fqs-texte/2-00/2-
IEEE Society Press. 00mayring-e.htm> (Accessed 3 June 2008).
Knight, L. (2002). Network Learning: Exploring Mitroff, I.I. (1983). Stakeholders of the Organisa-
Learning by Interorganizational Networks. Hu- tional Mind. Jossey-Bass. San Francisco, CA.
man Relations. 55(4), 427-454.
Meuter, M.L., Ostrom, A., Bitner, M.J. and
Koh, C., Ang, S. and Straub, D.W. (2004). IT Roundtree, R. (2003). The Influence of Technology
Outsourcing Success: A Psychological Contract Anxiety on Consumer Use and Experiences with
Perspective. Information Systems Research. 15(4), Self-service Technologies. Journal of Business
356-373. Research. 56(11), 899-906.
Kumar, K. and van Dissel, H.G. (1996). Sus- Negash, S., Ryan, T. and Igbaria, M. (2003). Qual-
tainable Collaboration: Managing Conflict and ity and Effectiveness in Web-based Customer
Cooperation in Interorganisational Systems. MIS Support Systems. Information and Management.
Quarterly, 20(3), 279-300. 40(8), September, 757-768.
Kurnia, S. and Johnston, R. (2002). A Review of Neuman, W.L. (2006). Social Research Meth-
Approaches to EC-enabled IOS Adoption Stud- ods: Qualitative and Quantitative Approaches,
ies. In Proceedings of the 35th Annual Hawaii Sixth Edition. Allwyn & Bacon, Boston, Mas-
Conference on System Sciences (HICSS’02), IEEE sachusetts.
Society Press.
Nieminen, H. (2005). Organisational Receptivity
Lane, P.J. and Lubatkin, M. (1998). Relative – Understanding the Inter-organisational Learn-
Absorptive Capacity and Inter-organisational ing Ability, Electronic Journal of Knowledge
Learning, Strategic Management Journal. Management. 3(2), 107-118.
19(5), 461-477.

1085
Successful Web-Based IT Support Services

Otim, S. and Grover, V. (2006). An Empirical Reichheld, F.F. and Schefter, P. (2000). E-Loyalty:
Study on Web-based Services and Customer Loy- Your Secret Weapon on the Web. Harvard Busi-
alty. European Journal of Information Systems. ness Review. 78(4), July-August, 105-112.
15(6), 527-541.
Ritter, T. and Gemunden, H.D. (2003). Inter-
Pan, G.S.C. (2005). Information Systems Project organisational Relationships and Networks: an
Abandonment: A Stakeholder Analysis. Inter- Overview. Journal of Business Research. 56,
national Journal of Information Management. 691-697.
25(2), 173-184.
Rockart, J.F. (1979). Chief Executives Define their
Peppard, J. (2001). Bridging the Gap Between own Data Needs. Harvard Business Review. 57(2),
the IS Organisation and the Rest of the Business: March-April, 81-93.
Plotting a Route. Information Systems Journal.
Rust, R.T. (2001). The Rise of E-Service, Journal
11(3), 249‑260.
of Service Research, 3(4), May, 283-284.
Peppers, D. and Rogers, M. (2001). One to One
Schultze, U. and Bhappu, A.D.P. (2005). Incorpo-
B2B: Customer Development Strategies for the
rating Self-Serve Technology into Co-production
Business-to-Business World, Currency Double-
Designs. International Journal of E-Collabora-
day, New York.
tion. 1(4), 1-23.
Piccoli, G., Brohman, M.K., Watson, R.T. and
Schultze, U. and Orlikowski, W.J. (2004). Practice
Parasuraman, A. (2004). Net-based Customer
Perspective on Technology-mediated Network
Service Systems: Evolution and Revolution in
Relations: The Use of Internet-based Self-serve
Web-site Functionalities. Decision Sciences.
Technologies. Information Systems Research.
35(3), 423-455.
15(1), 87-106.
Pitt, L.F., Berthon, P. and Lane, N. (1998). Gaps
Sigala, M. (2007). Investigating the Internet’s
Within the IS Department: Barriers to Service
Impact on Interfirm Relations: Evidence from the
Quality. Journal of Information Technology.
Business Travel Management Distribution Chain.
13(3), 191-200.
Journal of Enterprise Information Management.
Pujari, D. (2004). Self-service with a Smile? Self- 20(3), 335-355.
service Technology (SST) Encounters Among
Singh, M. and Byrne, J. (2005). Performance
Canadian Business-to-Business. International
Evaluation of e-Business in Australia. Electronic
Journal of Service Industry Management. 15(2),
Journal of Information Systems Evaluation.
200-219.
8(2).
Ragsdale, J. (2007). Self-Service Success Contin-
Solitander, M. (2006). Balancing the Flows:
ues to Decline; How Web 2.0 Can Help, Service
Managing the Intellectual Capital Flows in Inter-
and Support Professional Association, US.
organisational Projects. Electronic Journal of
Rai, A. and Sambamurthy, V. (2006). Editorial Knowledge Management. 4(2), 197-206.
Notes – The Growth of Interest in Services Man-
SSPA (2004). 2005 Support Demand Research
agement: Opportunities for Information Systems
Series. Tech Strategy Partners and Service &
Scholars. Information Systems Research. 17(4),
Support Professionals Association.
327 – 331,

1086
Successful Web-Based IT Support Services

Szulanski, G. (1996). Exploring Internal Sticki- Xu, L. (2007). Outsourcing and Multi-party
ness: Impediments to the Transfer of Best Practice Business Collaborations Modelling. Journal of
Within the Firm. Strategic Management Journal. Electronic Commerce in Organisations. 5(2),
17, Special Issue, 27-43. 77-96.
Tafti, M.H.A. (2005). Risk Factors Associated with Yin, R.K. (2003). Case Study Research, Design
Offshore IT Outsourcing. Industrial Management and Methods, Fourth Edition. Sage Publications,
& Data Systems. 105(5), 549-560. Newbury Park.
Vlachopoulou, M. and Manthou, V. (2003).
Partnership Alliances in Virtual Markets. Inter-
national Journal of Physical Distribution and
Logistics Management. 33(3), 254-267.

1087
Successful Web-Based IT Support Services

APPENDIX.

CSFs for Knowledge Transfer from a Support Organisation to a Customer


Organisation using WSSs (A)
Critical Success Factor Description
The cost equation for providing/using web-based self-service must
CSF-1 Cost Effectiveness be better, or at least not worse, than providing/using non-web-based
self-service.
CSF-2: Provision of Additional Ser- Current WSS transactions are used proactively as an opportunity to
vices and Cross-Selling Opportunities offer the customer organisation additional advice and services
A sufficient number of end-users must proactively contribute suf-
CSF-3: Critical Mass: Knowledge ficient knowledge content to the WSS knowledge base, to encourage
Content and Knowledge Contributors all parties to initially use, and to continue to use, the WSS as a means
of resolving their support issues or information requirements
The WSS must provide the functionality and knowledge required to
CSF-4: Usefulness: Provision of
meet the objectives of all stakeholders. For example, for the end-user
Knowledge Which Meets User Re-
customer, it should resolve a specific technical or business problem,
quirements
or provide other required knowledge resources.
Use of the WSS to resolve a support issue or provide other knowledge
resources must be perceived as efficient by all parties. This is inclu-
CSF-5: Ability to Provide Efficiency
sive of not only the performance of the WSS tool but the surrounding
processes for using the WSS.
The providing organisation, relevant business partners and the cus-
CSF-6: Access, Connectivity, Avail-
tomer organisation must have sufficient technology infrastructure in
ability and Performance
place, to enable all parties to participate in web-based self-service.
The WSS must have an effective Information Architecture and Search
CSF-7: Effective Information Archi- Engine such that the information system that organises and retrieves
tecture and Search Engine knowledge in the knowledge base is perceived as effective by end-
users.
All stakeholders using WSS must feel secure, private and confident in
all aspects of WSS transactions including the stored data components
CSF-8: Security, Privacy and Assur-
of transactions. Issues surrounding information security and informa-
ance
tion privacy, and the need to keep confidential related company
secrets (intellectual property) must be addressed.
An end-user must perceive that use of the WSS does not demand
CSF-9: Ease of Use/Usability
excessive cognitive or ergonomic effort.
The first few end-user experiences with the WSS must result in a
CSF-10: Early Positive Experience positive outcome, where end-user needs are met and they feel valued,
in order for the end-user to adopt WSS long term.
Using the WSS on an ongoing basis must result in a positive outcome,
where corporate customer needs and all types of end-users’ needs are
CSF-11: Positive Experience
met and they also feel valued. A positive experience is closely related
to customer organisation/end-user satisfaction
The customer organisation/end-user must feel confident that the solu-
tion provided by the WSS will resolve their issue and will not result
CSF-12 Confidence in Solution
in further issues. They must also have self-confidence in their own
ability to apply the offered solution.
The support organisation (and relevant business partners) must
CSF-13: Customer Focus: Under-
understand the individual business and technical needs of individual
stand Customer and their Require-
customer organisations and their end-users. With this understanding,
ments
WSS must be tailored to meet those individual needs.

1088
Successful Web-Based IT Support Services

APPENDIX.

CSFs for Knowledge Transfer from a Support Organisation to a Customer


Organisation using WSSs (B)
Critical Success Factor Description
The relationship between the support organisation, business partners
and the customer organisation must be one which supports open com-
CSF-14: Positive Relationship
munication and trust. This positive relationship should exist at both
the corporate and end-user levels.
Additional assistance, or education and training in respect to how
CSF-15: Provision of Additional Sup-
to use the WSS must be provided by the support organisation as
port: Education & Training
requested by end-users.
Management within the support organisation, business partner and
customer organisations must have an understanding of the work
processes and conditions which will affect the ability and willingness
CSF-16: Employee Focus of employees to adopt the WSS and associated strategies. With this
understanding, management must focus on meeting the needs of their
employees where possible, in order to maximise employee productiv-
ity and the benefits received from the WSS strategy.
The support organisation should foster an environment that recogn-
ises that WSS is part of the way it wants to conduct business. In addi-
CSF-17: Culture
tion, an open, sharing culture is needed. The culture should extend to
customer organisations and business partners.
CSF-18: Marketing and Awareness of Marketing programs which raise awareness of and support for, the
Web-based Service adoption of WSS, must be in place.
Knowledge capture processes to ensure that valuable knowledge is
created and captured into the WSS knowledge base by end-users must
CSF-19: Knowledge Creation, Cap-
be in place. Knowledge reuse processes to ensure that this knowledge
ture and Re-Use
is subsequently accessed and re-used by end-users, must also be in
place.
Processes must be in place to ensure the accuracy of the knowledge
CSF-20: Knowledge Validation which is captured into the WSS knowledge base and to ensure that
once it is captured, it is frequently reviewed and updated for currency.
Processes must be in place to ensure that the structure and format of
CSF-21: Knowledge Storage/Retrieval
captured knowledge facilitate findability.
The knowledge must be presented in a form which maximizes the
CSF-22 : Presentation of Knowledge
understanding acquired by end-users.
CSF-23: Measurement & Feedback Sufficient measurement and feedback methods for assessing the ef-
of WSS fectiveness of the WSS strategy must be in place.
There must be alignment and integration between the WSS and other
CSF-24: Alignment and Integration channels’ support processes, as well as with related business pro-
cesses, in the context of the business/industry environment.
The capability for the end-user or WSS to over-ride transactions
initially made via the WSS, must be in place, whereby if an end-user
CSF-25: WSS Override and Recovery is not finding a satisfactory resolution via the WSS, the transaction is
directed to an alternative mode of service delivery (e.g. a chat session
or telephone call).
A process must be in place whereby an end-user can easily re-initiate
CSF-26: Ease of Re-initiation a support transaction to re-locate a previously retrieved resolution or
other knowledge resource.
Top management must provide ongoing support and commitment to
CSF-27 Top Management Support
the WSS and associated strategies.

This work was previously published in International Journal of E-Services and Mobile Applications, Vol. 1, Issue 1, edited by
A. Scupola, pp. 1-20, copyright 2009 by IGI Publishing (an imprint of IGI Global).

1089
1090

Chapter 4.12
Mailing Lists and Social
Semantic Web
Sergio Fernández
Fundación CTIC, Spain

Diego Berrueta
Fundación CTIC, Spain

Lian Shi
Fundación CTIC, Spain

Jose E. Labra
University of Oviedo, Spain

Patricia Ordóñez de Pablos


University of Oviedo, Spain

Abstract to identify persons through the mailing lists of open


source communities. These techniques have been
Electronic Mailing lists are a key part of the Inter- tested using a dataset automatically extracted from
net. They have enabled the development of social several online open source communities.
communities who share and exchange knowledge
in specialized and general domains. In this chapter INTRODUCTION
the auhtors describe methods to capture some of that
knowledge which will enable the development of Early forms of electronic mailing lists were invented
new datasets using Semantic Web technologies. In almost as soon as electronic Mail (e-Mail) and are
particular, the authors present the SWAML project, a cornerstone of Internet, allowing a lot of people
which collects data from mailing lists. They also to keep up to date on news related with their inter-
describe smushing techniques that normalize RDF ests. Besides direct messaging between individuals,
datasets capturing different resources that identify mailing lists exist as private or public forums for
the same one. They have applied those techniques information exchange in communities with shared
interests. Mailing list archives are compilations
DOI: 10.4018/978-1-60566-272-5.ch004

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Mailing Lists and Social Semantic Web

of the previously posted messages that are often SOCIAL SEMANTIC WEB
converted into static HTML pages for their pub-
lication on the web. They represent a noteworthy The Semantic Web vision tries to develop new
portion of the contents that are indexed by web ways to integrate and reuse the information
search engines, and they capture an impressive published on the web. To that end, the W3C has
body of knowledge that, however, is difficult to developed several technologies, like RDF, which
locate and browse. enable to add metadata descriptions that contain
The reason for this difficulty can be traced back meaningful values and global properties to re-
to the translation procedure that run to transform sources. The resulting metadata forms a graph
the e-mail messages into static HTML pages. model which can be easily linked with other graphs
This task is fulfilled by scripts that create static (Berners-Lee, 2006) incrementing the knowledge
HTML pages for each message in the archive. In represented by the original graph. Those values
addition, some indexes (by date, by author, by and properties formalize the knowledge of a par-
thread) are generated and usually split by date ticular. In 2004, the W3C consortium developed
ranges to avoid excessive growth. OWL (Patel-Schneider et al, 2004), a web ontol-
On the one hand, this fixed structure reduces ogy language which facilitates the definition of
the flexibility when users explore the mailing list those formalizations, called ontologies. Based on
archives using their web browsers. On the other description logics, OWL has been adopted as the
hand, most of the meta-data that were associated to standard ontology language with several available
each e-mail message are lost when the message is editors, reasoners and tools. There have been also
rendered as HTML for presentational purposes. a number of ontologies developed in OWL for dif-
We propose to use an ontology and RDF ferent purposes and with different level of detail,
(Resource Description Framework, Klyne 2004) from generic to domain-specific ones.
to publish the mailing list archives into the (Se- On the other hand, in the last years, the concept
mantic) Web, retaining the meta-data that were of Web 2.0 has attracted a lot of interest. One of the
present in the messages. Additionally, by doing key aspects of Web 2.0 applications is the social
so, the information can be merged and linked to part of the web. Users are not considered as mere
other vocabularies, such as FOAF (Brickley and consumers of information, but also as produc-
Miller, 2005). ers. People want to share knowledge, establish
The rest of the chapter is organized as follows: relationships, and even work together using web
in section 2 we describe the main developments of environments. It is necessary to develop people-
Social Semantic Web related with mailing lists. In oriented web technologies which can represent
section 3, we explain several techniques to collect people interests and that enable the integration and
RDF datasets from mailing lists and other social reuse of people related information in the same
sources. Section 4 contains a description of the way that the semantic web vision advocates. These
SWAML project that collects those RDF datasets technologies can be seen as social semantic web
from mailing lists. In section 5, we describe several and we expect that there will be more and more
applications that consume that data. In section 6, applications making use of them.
we discuss some experiments that we have done One of the first developments is the FOAF
over those datasets. Finally, in section 7 we present vocabulary, which represents basic properties of
some conclusions and future work. people, like their name, homepage, etc. as well as
the people they know. FOAF descriptions are very

1091
Mailing Lists and Social Semantic Web

flexible and can be extended to other domains. mailing list can be represented by an instance of
There are already web portals which export their sioc:Forum (a subclass of Forum might be used
user profiles in FOAF format and the number of instead, although it is not required). Messages
FOAF applications is increasing. sent to the list and their replies become instances
Apart from FOAF, there are other ontologies of sioc:Post.
related to the social semantic web. In particular, Finally, people involved into the list are
SIOC (Semantically-Interlinked Online Com- instances of sioc:User. The SIOC ontology
munities), provides a vocabulary to interconnect provides a property to link forums and users,
different discussion methods such as blogs, web- namely sioc:has_subscriber. We argue that be-
based forums and mailing lists (Breslin 2005, ing subscribed to a mailing list is just one of the
Breslin 2006). Although we will apply mainly roles a user can play with respect to a forum.
SIOC to mailing-lists, it has a wider scope than Moreover, the list of subscribers is often avail-
just mailing lists, and generalizes all kinds of able only to the system administrator for privacy
online discussion primitives in the more abstract reasons. On the other hand, it is easy to collect the
sioc:Forum concept. Each forum represents an set of people who post to the list, i.e., the people
online community of people that communicate actively involved in the forum. Depending on the
and share a common interest. The goal of SIOC settings, the latter may be a subset of the former,
is to interconnect these online communities. in particular in those mailing lists that forbid post-
Other relevant concepts of the ontology are ing privileges to non-subscribers. Ideally, these
sioc:User and sioc:Post, which model respectively two different semantics would be captured using
the members of the communities and the content new properties. However, for practical reasons,
they produce. Instances of these three classes and to avoid privacy issues, we consider just the
(forums, users and posts) can be linked together already existent sioc:has_subscriber property, and
using several properties. we populate it with the set of active members of
The SIOC ontology was designed to express a forum. Consequently, inactive members of the
the information contained both explicitly and forum remain hidden, but this does not represent
implicitly in Internet discussion methods. Sev- a problem due to the open world assumption.
eral software applications, usually deployed as Additionally, the Dublin Core (Dublin Core
plug-ins, are already available to export SIOC Metadata Element Set, Version 1.1, 2006) and
data from some popular blogging platforms and Dublin Core Terms vocabularies are used to
content management systems. The effort, however, capture meta-data such as the message date
is focused on web-based communities (blogs, (dcterms:created) and title (dc:title).
discussion forums), while little has been done Given the distributed nature of RDF, it is ex-
so far to extend the coverage to legacy non-web pected that there will be different RDF datasets
communities, such as mailing lists and Usenet describing aspects of the same resources. The term
groups. smushing has been defined as the process of nor-
SIOC classes and properties are defined in malizing an RDF dataset in order to unify a priori
OWL, and their instances can be expressed in different RDF resources which actually represent
RDF. Therefore, they can be easily linked to other the same thing. The application which executes a
ontologies. The obvious choice here is FOAF, data smushing process is called a smusher. The
which provides powerful means to describe the process comprises two stages:
personal data of the members of a community. First, redundant resources are identified;
Mailing lists can be easily described by instan- then, the dataset is updated to reflect the re-
tiation of the SIOC classes and properties. Each cently acquired knowledge. The latter is usually

1092
Mailing Lists and Social Semantic Web

achieved by adding new triples to the model to This is particularly true when there is no obvious
relate the pairs of redundant resources. The OWL immediate benefit of exporting SIOC data.
property owl:sameAs is often used for this pur- On the other hand, methods which do not
pose, although other properties without built-in require direct access to the database and can
logic interpretations can be used as well (e.g.: operate on resources already published on the
ex:hasSimilarName). Redundant resources can web are non-intrusive. One technique is the use
be spotted using a number of techniques. In this of cooked HTML views of the information, the
chapter, we explore two of them: (1) using logic same ones that are rendered by web browsers for
inference and (2) comparing labels. human consumption. An example could be RSS/
Atom feeds, which have become very popular in
the recent years. They can be easily translated
COLLECTING DATA INTO THE into SIOC instances using XSLT stylesheets
SOCIAL SEMANTIC WEB (for XML-based feeds) or SPARQL queries (for
RSS 1.0, which is actually RDF). Unfortunately,
Since SIOC is a recent specification, its adoption these feeds often contain just partial descriptions.
is still low, and only a few sites export SIOC data. Another technique is the use of public APIs. The
There exist a number of techniques that can be Web 2.0 trend has pushed some social web sites to
used to bootstrap a network of semantic descrip- export (part of) their functionality through APIs in
tions from current social web sites. We classify order to enable their consumption by third-party
them in two main categories: intrusive and non- mash-ups and applications. Where available, these
intrusive techniques. APIs offer an excellent opportunity to create RDF
On the one hand, methods which require views of the data. A shared aspect of these sources
direct access to the underlying database behind is their ubiquitous availability through web pro-
the social web site are intrusive techniques. The tocols and languages, such as HTTP and XML.
web application acts as the controller and pub- Therefore, they can be consumed anywhere, and
lishes different views of the model in formats thus system administrators are freed of taking care
such as HTML and RSS. In terms of this pattern, of any additional deployment. In contrast, they
publishing SIOC data is as simple as adding a cannot compete with the intrusive approaches in
new view. From a functional point of view, this terms of information quality, as their access to the
is the most powerful scenario, because it allows data is not primary.
a lossless publication due to the direct access to
the back-end database. The SIOC community
has contributed a number of plugins for some SWAML Project
popular web community-building applications,
such as Drupal, WordPress and PhpBB2. Mail- SWAML (Fernández et al, 2008) is a Python tool
ing lists are also covered by SWAML, which is that reads mailing list archives in raw format,
described in the next section. There is, however, typically stored in a “mailbox” (or “mbox”),
a major blocker for this approach. All these soft- as defined in RFC 4155 (Hall 2005). It parses
ware components need a deployment in the server mailboxes and outputs RDF descriptions of the
side (where the database is). This is a burden for messages, mailing lists and users as instances of
system administrators, who are often unwilling to the SIOC ontology. Internally, it re-constructs the
make a move that would make it more difficult to structure of the conversations in a tree structure,
maintain, keep secure and upgrade their systems. and it exploits this structure to produce links be-
tween the posts. This script is highly configurable

1093
Mailing Lists and Social Semantic Web

Figure 1. SIOC post example in RDF/XML

and non-interactive, and has been designed to be The second link among messages is established
invoked by the system task scheduler. This low- between a post and its immediate successor (or
coupling with the software that runs the mailing predecessor) in chronological order. It is worth
list eases its portability and deployment. to note that this link is not strictly necessary, be-
SWAML could be classified as an intrusive cause the following (or preceding) message can
technique because it requires access to the pri- be obtained by sorting by date the sequence of
mary data source, even if in this case it is not a posts. However, this is a rather expensive opera-
relational database but a text file (for instance, the tion, because the whole set of posts is required
approach followed by mle (Michael Hausenblas at in order to perform the sorting. The open world
al., 2007) is considered completely non-intrusive). assumption makes this query even more chal-
Anyway, it is worth mentioning that some servers lenging. Therefore, considering that browsing to
publish these text files (mailboxes) through HTTP. the previous or next message is a common use
Therefore, sometimes it is possible to retrieve the case, and the complete set of posts can be very
mailbox and build a perfect replica of the primary large or even unavailable, we introduced two
database in another box. In such cases, SWAML new properties, next_by_date and prev_by_date.
can be used without the participation of the system These properties where eventually accepted into
administration of the original web server. the SIOC ontology. An RDF representation of a
There are many ways in which a mailing list sample message is shown in Figure 1.
message might be related with other messages. SWAML is essentially a mailbox parser and
However, we consider just two scenarios. The first translator implemented in Python. Its output is a
one links a post with its replies (sioc:has_reply). number of SIOC instances (Forum, Posts and Us-
Actually, due to sequential layout of the messages ers) in a set of RDF files. SWAML can be invoked
in the most widely used format to store mailing by the system task scheduler.
list archives (mailbox), it is easier to generate Parsing the mailbox and rebuilding the discus-
the inverse property (sioc:reply_of). Anyway, the sion threads may be sometimes tricky. Although
has_reply property can be generated either by a each mail message has a supposedly unique iden-
description logics reasoner or by performing two tifier in its header, the Message-ID, defined by
passes over the sequence. RFC 2822 (Resnick, 2001), in practice its unique-

1094
Mailing Lists and Social Semantic Web

Figure 2. Buxon is an end-user application that consumes sioc:Forum instances, which in turn can be
generated from mailboxes using SWAML.

ness cannot be taken for granted. Actually, we have different representations of the same information
found some messages with repeated identifiers resource should share a unique URI. Fortunately,
in some mailing lists, probably due to non-RFC the HTTP protocol supports this feature by using
compliant or ill-configured mail transport agents. “content-negotiation”. Clients of the protocol can
Therefore, SWAML assumes that any reference to declare their preferred formats in the headers of
a message (such as those created by the In-Reply- an HTTP request using the Accept header. Web
To header) is in fact a reference to the most recent servers will deliver the information in the most
message with that ID in the mailbox (obviously, suited available format, using the Content-type
only previous messages are considered). Using header of the HTTP response to specify the actual
this rule of thumb, SWAML builds an in-memory format of the returned delivered content. MIME
tree representation of the conversation threads, so types such as text/html and application/rdf+xml
sioc:Posts can be properly linked. are used as identifiers of the requested and avail-
Actually, SWAML goes further than just a able formats.
format-translation tool. A dedicated subroutine Setting up the content negotiation in the
that runs as part of the batch execution but may be server-side usually requires some tuning of the
also separately invoked on any sioc:Forum, tries web server configuration. It also depends on some
to find a FOAF description for each sioc:User. choices made by the publisher of the information,
One important requirement of the semantic such as the namespace scheme for the URIs or the
web is to be an extension (and not a replacement) fragmentation of the information. In (Miles et al,
of the current document-based web. Ideally, each 2006) there is a list of some common scenarios,
user agent must be able to retrieve the information which are described to great detail, and configu-
in their format of choice. For instance, current ration examples for the Apache web server are
web browsers prefer (X)HTML documents, be- provided. The most suitable scenarios (or recipes,
cause they can be rendered and presented to the as they are called) to publish mailing list metadata
end user. However, semantic web agents require are the fifth and sixth, i.e., multiple documents
information to be available in a serialized RDF available both in HTML and RDF.
format, such as RDF/XML or N3. Furthermore,

1095
Mailing Lists and Social Semantic Web

Figure 3. A sample htaccess configuration file for Apache generated by SWAML. These two rules redi-
rect the request to the proper file based on the content negotiation field of the HTTP request. Some lines
have been wrapped for readability.

The fifth scenario is extensively described in satisfied without retrieving a significant overload
the referred source, and it has been implemented of unneeded information. However, scattering
in SWAML. At the same time RDF and HTML the metadata across a myriad of small files has
files are written, SWAML also produces htaccess some disadvantages. For instance, the number of
local configuration files for Apache. One of these resources that must be retrieved to fulfill a single
configuration file is shown in Figure 3, while a query is greatly increased. Therefore, storing the
sample request/response negotiation is depicted RDF graph in a specialized database is an appeal-
in Figure 4. ing alternative.
RDF metadata generated by SWAML can Fortunately, a common protocol to access se-
grow to a large size for lists with a high traffic mantic repositories using SPARQL as the query
and several years of operation, where there are language is available (Clark 2006) and is gain-
tens of thousands of messages. The partition of ing support by the RDF databases. This protocol
the information might be an issue in such cases. exposes a simple API to execute and retrieve the
On the one hand, information chunks are preferred results of SPARQL queries (at the present moment,
to be small so any conceivable use case can be SPARQL is a read-only query language, although

Figure 4. An HTTP dialog with content negotiation

1096
Mailing Lists and Social Semantic Web

Figure 5. Sample Apache web server rewrite rule to translate HTTP request into SPARQL queries using
a Sesame RDF repository. The last line has been wrapped for readability.

there are proposals to extend it with full CRUD even the presence of a data repository instead of
capabilities such as those of SQL). This abstract static files is kept completely hidden to the client.
query API may be realized by different means, At the same time, by avoiding the undesirable
such as SOAP bindings (described by a WSDL data replication, this technique helps to keep the
2.0 interface) and HTTP bindings. The former information consistent. The most representative
enables interoperability with web service frame- feature of our proposal is that it does not require
works, while the latter can be exploited without any kind of server side script or application to
the full-blown web service machinery. translate the queries, because the data repository
Web service endpoints which implement the can serve the information directly in the format
SPARQL protocol are sprouting on the web, some desired by the client.
of them pouring huge amounts of data into the We have implemented this technique using the
semantic web. We argue that metadata of large Apache web server and Sesame 2.0 RDF repository
mailing lists can be conveniently exposed as (Broekstra et al, 2006). Figure 6 reproduces the
SPARQL endpoints. That means to effectively hand-made htaccess file (as opposed to the ones
translate the decision on data selection to the client that are automatically produced by SWAML).
(Pan 2006), and therefore minimizing the number Unfortunately, Of course, the rewrite rule must
of requests and the data overload. For instance, be fired only when RDF data is requested, while
the client agent can retrieve all the headers of the requests for HTML must go through it.
messages in a given date range, but skip the body We note, however, that our proposal presents
of the messages, saving a considerable amount some security-related issues. In particular, it is eas-
of bandwidth. ily vulnerable to SPARQL-injection. Therefore,
However, non SPARQL-aware agents still we strongly discourage the use of this technique
need to access the information. This is the sce- in production environments. Nevertheless, some
nario of the sixth scenario (recipe) of the above changes in the regular expressions are possible in
cited document, but unfortunately this one is order to prevent this kind of attack.
still being discussed. We propose a simple solu- There is another different approach to publish-
tion based on URL rewriting of the requests in ing metadata: to embed it into the HTML content.
order to translate conventional HTTP requests W3C is pushing two complementary technologies,
for resources into SPARQL queries that dynami- RDFa (Adida & Birbeck, 2007) and GRDDL
cally generate an RDF subgraph that contains the (Connolly, 2007), which respectively encode into,
requested information about the resource. The and extract RDF data from XHTML documents.
rewriting mechanism, the SPARQL query and We have also explored this path. SWAML gener-

1097
Mailing Lists and Social Semantic Web

Figure 6. A single message rendered as XHTML code with RDFa and GRDDL markup by SWAML.

ates simple XHTML pages for each message to the archives of mailing lists as if they were using
illustrate the usage of both RDFa and GRDDL. their desktop mail application. Buxon takes the
We must remark that these pages are just a proof- URI of a sioc:Forum instance (for example, a
of-concept of the semantic enrichment, and they mailing list exported by SWAML, although any
lack many of the fancy features and complex sioc:Forum instance is accepted) and fetches the
templates of the already-existent applications data, retrieving additional files if necessary. Then,
which generate plain HTML. it rebuilds the conversation structure and displays
the familiar message thread list (see Figure 7).
CONSUMING MAILING Buxon also gives users the ability to query
LIST METADATA the messages, searching for terms or filtering the
messages in a date range. All these queries are
Buxon internally translated to SPARQL (Prud’hommeaux
& Seaborne, 2007) to be executed over the RDF
Buxon is a multi-platform desktop application graph. Newer versions of Buxon can send the
written in PyGTK. It allows end users to browse sioc:Forum URI to PingTheSemanticWeb.com, a

1098
Mailing Lists and Social Semantic Web

Figure 7. Buxon browsing SIOC-Dev mailing list.

social web service that tracks semantic web docu- plotted using a map web service such as Google
ments. That way, Buxon contributes to establish Maps (Figure 8).
an infrastructure that lets people easily create, It is also possible execute visualize the mes-
find and publish RDF documents. sages in a time line view using the Timeline
DHTML widget by the MIT SIMILE project using
Other Browsers and Clients a query like the one we propose in Figure 9.

The SIOC RDF data can be explored and queried


using any generic RDF browser, such as Tabulator Experimentation
(Berners-Lee et al., 2006). The most interesting
applications appear when instances of sioc:User A corpus of RDF data with many foaf:Person
are linked to FOAF descriptions of these users. instances was assembled by crawling and scrap-
For instance, it is trivial to write a query to ob- ping five online communities. There is a shared
tain the geographical coordinates of members topic in these communities, namely open source
of a mailing list and to codify them into a KML development; hence we expect them to have a
file (Ricket 2006), provided they describe their significant number of people in common. We
location in their FOAF file using the basic geo continue the work started in Berrueta et al (2007)
vocabulary (Brickley 2006). The KML file can be to mine online discussion communities, and we

1099
Mailing Lists and Social Semantic Web

Figure 8. Plotting the geographical coordinates of the members of a mailing list using KML and Google
Maps.

extend it to new information sources. More de- from this directory of open source projects
tails are described in Berrueta et al We use the and developers as Linked Data. Due to API
following sources: usage restrictions, we could only get data
about the +12,000 oldest user accounts.
• GNOME Desktop mailings lists: All the • Debian packages: Descriptions of Debian
authors of messages in four mailing lists packages maintainers were extracted from
(evolution-hackers, gnome-accessibility- apt database of Debian packages in the
devel, gtk-devel and xml) within the date main section of the unstable distribution.
range July 1998 to June 2008 were export-
ed to RDF using SWAML. Instances generated from these data sources
• Debian mailing lists: All the authors of were assigned a URI in a different namespace
messages in four mailing lists (debian- for each source. Some of these data sources do
devel, debian-gtk-gnome, debian-java and not directly produce instances of foaf:Person,
debian-user) during years 2005 and 2006 but just instances of sioc:User. An assumption
were scrapped from the HTML versions of is made that there is a foaf:Person instance for
the archives with a set of XSLT style sheets each sioc:User, with the same e-mail address and
to produce RDF triples. name. These instances were automatically created
• Advogato: This community exports its when missing. This assumption obviously leads
data as FOAF files. We used an RDF to redundant instances of foaf:Person which will
crawler starting at Miguel de Icaza’s pro- be later detected by the smusher.
file. Although Advogato claims to have The ultimate goal of our experiments is to
+13,000 registered users, only +4,000 were exercise the smushing processes described previ-
found by the crawler. ously against a realistic dataset. Two million RDF
• Ohloh: The RDFohloh (S. Fernández, triples were extracted from the sources described
2008) project exposes the information above, and put into OpenLink Virtuoso server

1100
Mailing Lists and Social Semantic Web

Figure 9. SPARQL query to extract the information required to visualize a time line of the messages
posted to any sioc:Forum instance.

which provides not only an effective triple store, On the other hand, a more strict interpretation
but also a SPARQL endpoint that was used to of how Mailman works is made with respect to
execute queries using scripts. the GNOME mailing lists, where identical e-mail
We evaluated two smushers: the first one address posting in different mailing lists are as-
smushed foaf:Person instances assuming that sumed to belong to a priori different users. In the
foaf:mbox_sha1sum is an IFP; the second one second case, we rely on the smushing process to
smushed the same instances comparing their merge the identities of these users.
foaf:name labels for string strict equality, without Although they must be handled with extreme
any normalization. Both smushers were imple- care due to the issues afore-mentioned, the com-
mented using SPARQL CONSTRUCT rules. bined results of the two smushing processes are
The newly created owl:sameAs triples were put consistent with the expected ones. For instance,
in different named graphs. These links were ana- there is a very high overlap between the Debian
lyzed to find co-occurrences of people in different developers (maintainers of Debian packages)
communities. and the Debian mailing lists. Obviously, Debian
Some communities use the e-mail address as developers are a relatively small group at the core
their primary key to identify its users. However, of the Debian community, thus they are very active
other communities use a different primary key, thus in its mailing lists. Another example is the overlap
allowing users to repeat their e-mail addresses. between Advogato and GNOME mailing lists.
For instance, a small number of users have regis- Advogato is a reputation-based social web site
tered more than one account in Advogato with the that blossomed at the same time that the GNOME
same e-mail (these accounts have been manually project was gaining momentum. Advogato was
reviewed, and they seem to be accounts created passionately embraced by the GNOME devel-
for testing purposes). opers, who used Advogato to rate each others’
Our data acquisition process introduces a development abilities.
key difference between how user accounts are We also studied whether there are some people
interpreted in Debian mailing lists and GNOME that are present in many of the communities at
mailing lists. The former considers e-mail address the same time. We chose communities which
as globally unique, i.e., the same e-mail address are closely related to each other, consequently,
posting in different Debian mailing lists is assumed we expected a high number of cross-community
to belong to the same user. subscribers. There are several people who are

1101
Mailing Lists and Social Semantic Web

present in many communities. We can conclude features that exceed what is now offered by static
that almost all the most active open source de- HTML versions of the archives on the web.
velopers in our dataset are core members of the Secondly, the crawlers of the web search
Debian community. Another interesting fact is engines can use the enhanced expressivity of the
that only a few people among the top members of RDF data to refine search results. For instance,
the communities consistently use a single e-mail precise semantic descriptions of the messages
address and just one variant of their names. This permit to filter out repeated messages, advance
fact proves the difficulty of the smushing process, in the fight against spam, or introduce additional
but also its usefulness. filter criteria in the search forms.
Another consequence of no lesser importance
is that each e-mail message is assigned a URI that
Conclusion and future work can be resolved to a machine-readable description
of the message. This actually makes possible to
There are a lot of ongoing efforts to translate data link a message like any other web resource, and
already reachable on the web into formats which therefore enriches the expressivity of the web.
are semantic web-friendly. Most of that work fo- Integration of the SWAML process with popu-
cuses on relational databases, micro-formats and lar HTML-based mailing list archivers, such as
web services. However, at the time of this writing Hypermail or Pipermail, would be a giant push
and to the best of our knowledge, e-mail was almost to speed up the adoption of SWAML. It is well
excluded from the Semantic Web. Our project, in known that one of the most awkward problems
combination with the generic SIOC framework, of any new technology is to gain a critical mass
fills this gap, conveniently providing an ontology of users. The semantic web is not an exception.
and a parser to publish machine-readable versions A good recipe to tackle this problem is to inte-
of the archives of the countless mailing lists that grate the new technology into old tools, making
exist on the Internet. a smooth transition without requiring any extra
Furthermore, the SWAML project fulfills a effort from users. Merging the SWAML process
much-needed requirement for the Semantic Web: into the batch flow of tools such as Hypermail
to be able to refer to semantic versions of e-mail would allow users to generate both RDF and
messages and their properties using resource URIs. production-quality, semantically enriched HTML
By re-using the SIOC vocabulary for describing versions of the archives.
online discussions, SWAML allows any semantic So far, no semantic annotation relative to the
web document (in particular, SIOC documents) meaning of the messages is considered. Obvi-
to refer to e-mail messages from other discus- ously, such information can not be automatically
sions taking place on forums, blogs, etc., so that derived from a RFC 4155-compliant mailbox.
distributed conversations can occur across these However, it is conceivable that it could be added
discussion media. Also, by providing e-mail mes- by other means, such as social tagging using
sages in RDF format, SWAML is providing a rich folksonomies, or parsing the metadata added by
source of data, namely mailing lists, for use in the authors of the messages using micro-formats
SIOC applications. or RDFa when posting in XHTML format. The
The availability of these data leads to some inherent community-based nature of mailing lists
benefits. In the first place, data can be fetched can be exploited to build recommendation systems
by user applications to provide handy browsing (Celma 2006).
through the archives of the mailing lists, providing We have also explored smushing techniques
to spot redundant RDF instances in large datasets.

1102
Mailing Lists and Social Semantic Web

We have tested these techniques with more than ACKNOWLEDGMENT


36,000 instances of foaf:Person in a dataset au-
tomatically extracted from different online open The authors would like to express their gratitude
source communities. We have used only public to Dr. John Breslin and Uldis Bojārs from DERI
data sources, consequently, these instances lack Galway, whose support and contributions have
detailed personal information. been a great help to this work. Also thanks to
We are aware of the extreme simplicity of Ignacio Barrientos for his contribution packaging
our experimentation using label comparison. In SWAML for Debian GNU/Linux.
our opinion, however, it contributes to show the
potential of this smushing technique. We note
that it is possible to have more usages for it, for REFERENCES
instance, smushing not just by people’s names, but
also by their publications, their organizations, etc. Adida, B., & Birbeck, M. (2008). RDFa Primer.
Surprisingly, the named-based smushing finds a Technical Report, W3C Working Draft.
high number of redundant resources even if the Berners-Lee, T. (2006). Linked Data Design Is-
comparison strategy for labels (names) is very sues. Available at http://www.w3.org/DesignIs-
simplistic (in this case, case-sensitive string equal- sues/LinkedData.html
ity comparison). More intelligent comparison
functions should lead to a higher recall. In this Berners-Lee, T., et al. (2006). Tabulator: Exploring
direction, we are evaluating some normalization and Analyzing linked data on the Semantic Web.
functions for names. We have also evaluated Proceedings of the 3rd International Semantic Web
classical information retrieval comparison func- User Interaction Workshop (SWUI06) workshop,
tions that take into account the similarity of the Athens, Georgia.
strings (e.g., Levenshtein); nevertheless, their
Berrueta, D. et al, (2008). Best practice recipes for
applicability to compare people’s names is open
publishing RDF vocabularies. Technical Report,
to discussion.
W3C Note.
We believe that the ratio of smushing can be
further improved if the dataset is enriched with Berrueta, D., Fernández, S., & Shi, L. (2007).
more detailed descriptions about people. Experi- Bootstrapping the Semantic Web of Social On-
ments are being carried out to retrieve additional line Communities. In Proceedings of workshop
RDF data from semantic web search engines as on Social Web Search and Mining (SWSM2008),
a previous step to smushing. co-located with WWW2008, Beijing, China.
We have implemented a smusher application
Bojārs, U., & Breslin, J. (2007). SIOC Core
for persons, and we intend to use it to further
Ontology Specification. Available at http://rdfs.
investigate the potential for the optimization of
org/sioc/spec/.
the smushing process. The way in which these
techniques are translated into actual algorithms Breslin, J., et al. (2005). Towards Semantically-
is critical to achieve a promising performance of Interlinked Online Communities. Proceedings
the smushing process, especially for very large of the 2nd European Semantic Web Conference,
datasets. In parallel, increasing the precision of ESWC 2005, Heraklion, Crete, Greece.
smushing will require to study how to enable
Breslin, J. (2006). SIOC: an approach to connect
different smushing strategies to interrelate and
web-based communities. International Journal of
reciprocally collaborate.
Web Based Communities, 2(2), 133–142.

1103
Mailing Lists and Social Semantic Web

Brickley, D. (2006). Basic geo (WGS84 lat/long) Hausenblas, M., & Rehatschek, H. (2007). mle:
vocabulary. Technical report, W3C Informal Enhancing the Exploration of Mailing List Ar-
Note. chives Through Making Semantics Explicit. Se-
mantic Web Challenge 07, Busan, South Korea.
Brickley, D., & Miller, L. (2005). FOAF Vocabu-
lary Specification. Technical report. Klyne, G., & Carroll, J. J. (2004). Resource
Description Framework (RDF): Concepts and
Broekstra, J. (2006). Sesame: A generic archi-
abstract syntax. Technical report, W3C Recom-
tecture for storing and querying RDF and RDF
mendation.
Schema. In Springer . Lecture Notes in Computer
Science, 2342, 54–68. doi:10.1007/3-540-48005- Pan, Z., et al. (2006). An investigation into the
6_7 feasibility of the semantic web. Technical Report
LU-CSE-06-025, Dept. of Computer Science and
Celma, O. (2006). FOAFing the music: Bridg-
Engineering, Lehigh University.
ing the semantic gap in music recommendation.
Proceedings of the 5th International Semantic Patel-Schneider, P. F., Hayes, P., & Horrocks, I.
Web Conference, Athens, USA. (2004). OWL Web Ontology Language: Semantics
and Abstract Syntax. Recommendation, W3C,
Clark, K. G. (2008). SPARQL protocol for RDF.
February.
Technical report, W3C Recommendation.
Prud’hommeaux, E., & Seaborne, A. (2008).
Connolly, D. (2007). Gleaning Resource Descrip-
SPARQL Query Language for RDF. Technical
tions from Dialects of Languages (GRDDL). Tech-
report, W3C recommendation.
nical report, W3C Candidate Recommendation.
Resnick, P. (2001). RFC 2822 - internet message
Fernández, S., Berrueta, D., & Labra, J. E. (2008).
format, Technical report, The Internet Society.
A Semantic Web Approach to Publish and Con-
sume Mailing Lists. IADIS International Journal Ricket, D. (2006). Google Maps and Google Earth
on WWW/Internet, 6, 90-102. integration using KML. In American Geophysical
Union 2006 Fall Meeting.
Fernándrez, S. (2008). RDFohloh, a RDF Wrap-
per of Ohloh. Proceedings of 1st workshop on Shi, L., Berrueta, D., Fernández, S., Polo, L., &
Social Data on the Web (SDoW2008), collocated Fernández, S. (2008). Smushing RDF instances:
with 7th International Semantic Web Conference, Are Alice and Bob the same open source devel-
Karlsruhe, Germany. oper? Proceedings of 3rd ExpertFinder workshop
on Personal Identification and Collaborations:
Hall, E. (2005). RFC 4155 - the application/
Knowledge Mediation and Extraction (PICKME
mbox media type. Technical report, The Internet
2008), collocated with 7th International Semantic
Society.
Web Conference, Karlsruhe, Germany.

This work was previously published in Social Web Evolution: Integrating Semantic Applications and Web 2.0 Technologies,
edited by M. D. Lytras; P. Ordonez de Pablos, pp. 42-56, copyright 2009 by Information Science Reference (an imprint of IGI
Global).

1104
1105

Chapter 4.13
Communicative Networking and
Linguistic Mashups on Web 2.0
Mark Pegrum
University of Western Australia, Australia

Abstract glance at Web 1.0 and Web 3.0 can be helpful in an


understanding of Web 2.0, the term popularized by
This chapter discusses the application of a range Tim O’Reilly through the first Web 2.0 Conference
of Web 2.0 technologies to language education. It in 2004 (O’Reilly, 2005) and now commonly used
argues that Web 2.0 is fundamentally about network- to describe the current state of the web.
ing, community building, and identity negotiation. The retrospective term Web 1.0 refers to the
Given the textual nature of the Web, all of this is initial information-oriented web, authored by a
made possible primarily through the medium of small number of people for a very large number of
language. Consequently, Web 2.0 is ideally suited users. Consisting mainly of static webpages, it of-
to the teaching of language and literacy. To be most fered little room for interactivity. Educational uses
effective, this requires a broadly social constructivist largely fell into two categories: information retrieval
pedagogical approach as well as a willingness to (as in webquests) or rote training (drill exercises).
work with the messy reality of linguistic “mashups,” While there were some clear benefits in terms of
the hybrid uses of languages, codes, and media student autonomy, use of authentic materials and
which inform Web 2.0. exposure to multiliteracies, and while problem-
based learning and guided discovery approaches to
Web 1.0 were not unknown, it was most often used
Introduction in ways corresponding to traditional transmission
or behaviourist models of pedagogy.
There continues to be widespread confusion and Web 3.0, a speculative term describing a possible
apprehension about the effects of the Internet and future version of the web, refers most commonly
new technologies on education. Recent discussions to the semantic web, where software agents will
of the web in versions ranging from 1.0 to 3.0 have collate and integrate information to give intelligent
done little to alleviate this situation, with at least one responses to human operators, and/or the geospatial
spurious reference to Web 6.0 (Motteram & Ioannou- web, where location will be used to index informa-
Georgiou, 2007) making the point that labels and tion. These are, however, long-term projections,
numbers are not the important thing. However, a whose educational implications are impossible to
assess at present.
DOI: 10.4018/978-1-60566-190-2.ch002

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Communicative Networking and Linguistic Mashups on Web 2.0

In between is the presently dominant Web 2.0, connections, building communities and shaping
also known as the social web, which comprises a their own self-representations online. In this way,
loose grouping of newer generation social tech- language and literacy educators can play a key
nologies whose users are actively involved in role in the collaborative enterprise that is Web
communicating and collaborating with each other 2.0. It is important to acknowledge, however, that
as they build connections and communities across effective use of Web 2.0 requires a rethinking of
the world, negotiating their online identities in the approaches to literacy and pedagogy which may
process. What happened, as Davies puts it, was have traditionally seemed unproblematic, but
that “society got more technical while software which are less than ideally suited to the new on-
got more social” (2003, p. 5). The 2007 Horizon line environment — or the wider world in which
Report describes Web 2.0’s social networking sites it is embedded.
as being “fundamentally about community” (New This chapter begins by examining recent
Media Consortium, 2007, p. 12), while Jimmy changes in conceptions of literacy and pedagogy
Wales (2007), founder of Wikipedia, has linked which may enable educators to better frame
Web 2.0 to the new digital literacies concerned with their use of Web 2.0. It then goes on to discuss
“inclusion, collaboration and participation”. In common Web 2.0 tools and their applications to
brief, Web 2.0 technologies, from blogs and wikis language education, focusing firstly on collab-
through social networking sites and folksonomies orative technologies such as discussion boards,
to podcasting and virtual worlds, are all about blogs and wikis; secondly on social networking
communicative networking. Such networking is technologies; thirdly on information linking tech-
likely to become increasingly important as a digital nologies like folksonomies and RSS; and fourthly
native ethos takes over from a digital immigrant on cutting-edge technologies such as podcasting,
one (Prensky, 2001), as more technologies become m-learning and virtual worlds. Finally, the chapter
available to those with little specialist expertise in explores some of the main limitations of Web 2.0
IT, and as today’s technologies converge to form in education, in a discussion which ranges across
ever more versatile hybrids. pedagogical, social, sociopolitical and philosophi-
Web 2.0 has many applications in education, cal issues. Drawing these threads together, the
both current and potential, but its greatest impact conclusion offers recommendations for language
may well be in subjects which foreground language and literacy educators who wish to use Web 2.0
and communication. After all, given the textual more extensively in their teaching.
nature of the web, all the connections made online
and all the communities established there are en-
abled primarily through the medium of language. Changing Literacies
As a result, for language and literacy educators, the and Pedagogies
advent of Web 2.0 presents great opportunities: to
decentralize the role of the classroom (Coleman, It has been clear for some time that traditional
2007), escape the language lab, and engage with print literacy alone is no longer sufficient to
the younger generation of digital natives on their allow people to operate effectively in society.
own territory. It is a territory whose geography is Web 2.0 greatly exacerbates the problematical
forged through language and whose key naviga- aspects of this situation. As a result, there is an
tion tools are literacies. Teachers can help their urgent need to pluralize the concept of literacy,
students develop greater language competence and as has been claimed in recent work on literacies
additional linguistic tools to navigate Web 2.0, and multiliteracies (Barton & Hamilton, 2000;
as the students engage in the process of making Cope & Kalantzis, 2000; Kist, 2004; Street, 1994;

1106
Communicative Networking and Linguistic Mashups on Web 2.0

Unsworth, 2001). It is important to challenge the Fortunately, there is a range of appropriate


focus on “formalised, monolingual, monocultural, pedagogical tools at hand. While Web 1.0 lent itself
and rule-governed forms of language” inherent to transmission pedagogies and behaviourist drills,
in print literacy pedagogy (New London Group, working effectively with Web 2.0 demands a more
2000, p. 9). In their place, multiliteracies should be constructivist orientation. Social constructivist
promoted and developed to facilitate the naviga- pedagogy, with its roots in the work of Vygotsky
tion of “our culturally and linguistically diverse and carrying influences from Dewey and progres-
and increasingly globalised societies” as well as sivism, views social interaction as the source of
“the burgeoning variety of text forms associated all learning. Acknowledging and valuing students’
with information and multimedia technologies” pre-existing knowledges and multiple perspec-
(ibid.). The multiliteracies paradigm can thus refer tives, it helps students deconstruct and reconstruct
to multiple cultural and linguistic codes on the one these as they engage actively and collaboratively
hand, and to multiple media on the other. Both in building new understandings through scaffolded
aspects, but particularly the latter, are reflected learning experiences (Dalgarno, 2001; Finger,
in the rapidly multiplying treatises on computer, Russell, Jamieson-Proctor & Russell, 2007, p.
electronic and hypertext literacies (Dudfield, 119; Jonassen, 1992). As Hoppe, Joiner, Milrad
1999; Kern, 2006; Selber, 2004; Warschauer, and Sharples (2003) state, “there is an impera-
1999, 2003; Wray, 2004). tive to move from a view of e- and m-learning
In short: in the Web 2.0 environment, there as solely delivery mechanisms for content”—the
is a dynamic fusion of media and a rich blend transmission approach typical of Web 1.0—and to
of cultures, languages and, within languages, embrace contemporary pedagogy with its “high
evolving codes and registers. While English may valuation of active, productive, creative and col-
be the default lingua franca, it is less a single laborative learning methods [which go] much
international English and more a loosely con- beyond the ‘absorption’ of codified information”
catenated assemblage of World Englishes. And (p. 255; italics in original). It might be argued that
generational differences ensure that, even among a constructivist approach is becoming ever more
speakers of single varieties of English, there is a relevant in a world where, as Warschauer (2007)
bewildering mixture of modes of self-expression. indicates, “[t]he ability to draw on rote answers
Indeed, the multilingualism and multiliteracies is inadequate” because “yesterday’s answers
which underpin Web 2.0 parallel the increasingly are outdated faster than ever” (p. 42). What is
productive mixing of pre-existing video, graphics, relevant in such a world is the ability to seek out
music and text commonly referred to as mashups information through networks of contacts, and to
(a term derived from the hip hop practice of mix- collaboratively build understanding with others
ing songs to create new hybrids). “Linguistic engaged in similar pursuits. The social networking,
mashups,” then, would seem to be in the nature dialogue building and collaborative knowledge
of international socialization and online network- construction tools of Web 2.0 are uniquely suited
ing: the emphasis is on communication, which to preparing students for this world.
involves sophisticated aggregations of multiple Another useful perspective is provided by the
media drawing on increasingly porous cultural communities of practice paradigm, where learning
and linguistic codes. Web 2.0 is not about neat is conceived of as “social participation,” meaning
definitions or clear borders. Rather, its users must that people engage in the “process of being active
find ways to work with the global cacophony of participants in the practices of social communities
voices which make up its textual fabric. and constructing identities in relation to these com-

1107
Communicative Networking and Linguistic Mashups on Web 2.0

munities” (Wenger, 1998, p. 4; italics in original). ing communication in “the messy real world of
Communities of practice have, in fact, been defined cultural flows and mixes” (Pegrum, forthcoming
as “networked learning systems” which connect 2008a) — one whose messiness is exponentially
“all participants and learning system components increased by the technological affordances and
across multiple levels of practice and inquiry” communicative possibilities of Web 2.0.
(Quinton, 2006, p. 563). This is precisely the kind In the new millennium, the notion of identity
of educational networking that can be fostered by has also emerged as a major focus of research
Web 2.0 applications. As students begin to use in language pedagogy, thanks in large part to
these tools, they are not only gaining important the work of Norton (2000), who observes that
future skills but may well find themselves enter- “an investment in the target language is also an
ing, as legitimate peripheral participants, the investment in a learner’s own identity” (p. 11).
very communities of practice in which they will Pavlenko and Blackledge (2004) foreground the
eventually become full participants. It is implicitly questions of power and empowerment which
a community of practice orientation that Holmes, underpin identity concerns:
Tangney, FitzGibbon, Savage and Mehan (2001)
ascribe to when they express the hope that, in a individuals are agentive beings who are constantly
“communal constructivist” approach to new tech- in search of new social and linguistic resources
nologies, “students will not simply pass through which allow them to resist identities that position
a course like water through a sieve but instead them in undesirable ways, produce new identities,
leave their own imprint in the development of and assign alternative meanings to the links be-
the course, their school or university, and ideally tween identities and linguistic varieties. (p. 27)
the discipline” (p. 1).
In language teaching itself, the last decade Ricento (2005) goes even further in describ-
of the twentieth century witnessed a move away ing:
from the ideals of the communicative approach
— which, having dealt with some of the key limi- the central role of language in the negotiation
tations of preceding approaches, came to create of a person’s sense of self at different points in
its own problems — and towards a conception of time and in different contexts, and in allowing a
intercultural communicative competence. While person access (or lack thereof) to powerful social
continuing to recognize the importance of the networks that give learners the opportunity to
communicative element, the intercultural com- speak. (p. 898)
municative competence movement has rejected
any insistence on the imitation of native speaker Web 2.0 places an even greater premium on
models along with the accompanying goal of inte- such issues for language teachers and learners:
gration into a target culture. Rather, the language it elevates to the level of a constituting principle
learner is encouraged to move into a “third place” the notion that identity is constructed through
(Kramsch, 1993) between cultures; from here, he language.
or she will be able to explore his/her own culture In sum, if the limitations of a “single-mode,
as well as other cultures, which are not seen as single-language, single-culture literacy” (Pegrum,
static entities into which full integration might be forthcoming 2008a) were always apparent to some,
possible, but rather as multiple, contradictory and they are all the more obvious in our shrinking
in flux (Byram, 1997; Corbett, 2003; Kramsch, world, where members of the net generation are
1998; Phipps & Gonzalez, 2004). Intercultural simultaneously bound together and yet differenti-
competence is thus very much about negotiat- ated from each other through their use of Web 2.0

1108
Communicative Networking and Linguistic Mashups on Web 2.0

tools. What Pennycook (2007) has recently writ- judgement. Most students, Hubbard (2004) notes,
ten in regard to the rapidly globalizing culture of can “profit from some formal, sustained training
hip hop—original source of the mashup—applies in how to take operational competence in a given
equally to students’ desire for linguistic and cul- computer application and transfer that into learn-
tural self-realisation on Web 2.0: ing competence” (p. 51). More than this, students
need to learn critical literacy skills to sort through,
If we believe that education needs to proceed by evaluate and prioritize the masses of data with
taking student knowledge, identity and desire into which they are confronted, turning information
account, we need to engage with multiple ways of into understanding (McFarlane, Roche & Triggs,
speaking, being and learning, with multilayered 2007; Pesce, 2007). They also need a grasp of the
modes of identity at global, regional, national powerful linguistic and media options at their
and local levels. disposal for shaping their identities and engag-
ing with others online. It is a fallacy to think that
educators in this new virtual world are no more
Unless we get in touch with this as educators, the than facilitators. As has been widely argued in the
flow will pass us by. … Languages will flow and literature about online learning, and in line with
change around us, new combinations of languages social constructivist pedagogical models, teach-
and cultures will be put together, texts will be ers must be prepared to play a central organizing,
sampled and mixed in ever new juxtapositions. guiding and mentoring role (Garrison & Anderson,
Students are in the flow; pedagogy needs to go 2003; Pegrum, 2007; Warschauer, 2007).
with the flow. (p. 158) In doing so, they have a golden opportunity
to engage with their students. They can support
Of course, it is not only about multiple Eng- the latter’s online self-presentations and endorse
lishes, but multiple languages. It is not only about their community building by helping to enhance
multiple texts, but multiple textualities. It is time, their language and literacy skills. At the same
as Canagarajah (2003) has suggested, to begin time, teachers should be open to learning from
teaching the “fluid literacies” (p. xi) essential their students about their digital lifestyles — and
for navigation and negotiation in this new hybrid in the process, teachers may well find their own
world: language and literacy skills enhanced in unex-
pected ways. Collaboration which brings together
Rather than developing mastery in a ‘target teachers’ pedagogical and critical expertise and
language,’ we should strive for competence in a students’ technological and practical expertise
repertoire of codes and discourses. Rather than is the only way to unlock the full educational
simply joining a speech community, we should potential of Web 2.0.
teach students to shuttle between communities.
Not satisfied with teaching students to be context-
sensitive, we should teach them to be context- Thinking Collaboratively
transforming. (p. xiii)
Much of Web 2.0 is devoted to fostering com-
Few can doubt that students are part of this munities of interest or practice which nurture
world already, on the web and beyond it. But collaborative thinking. As such, it effectively il-
that does not mean they are fully accomplished lustrates the potential, noted by Kaye in the early
navigators, have all the language and literacy skills days of computer-mediated communication, for
they need, or always exercise appropriate critical the “weaving together of ideas and information

1109
Communicative Networking and Linguistic Mashups on Web 2.0

Figure 1. Sample discussion thread replies from Third Space Trial 1, Feb. 2007

from many peoples’ [sic] minds” (1989, p. 3). cussion than newer voice alternatives (whether
This principle underpins asynchronous discussion synchronous VoIP or asynchronous voiceboards).
boards (DBs), in some ways a spiritual precursor See Figure 1 for an example of threaded postings
of Web 2.0, along with the more multifaceted in an international Master’s forum for language
blogs, wikis and hybrid blikis (or blokis), all of teachers. Used in conjunction with face-to-face
which may contain in-built discussion or com- classes, DBs may help cater to differing learning
ments features. styles and needs. For example, they allow more
Being text-based, asynchronous DBs are time to be spent on composition of contributions
natural vehicles for the development of writing by less extroverted or non-native students; the
skills, while there is some limited evidence they time-independence of DBs may thus “mitigate
may also support the development of oral skills the effects of certain inequalities” (Locke, 2007,
(Burgmer, 2006, p. 96; Levy & Stockwell, 2006, p. 188). It has also been widely claimed that DB
p. 182). It has been widely observed that writing exchanges typically display a high level of cog-
on the Internet, because of its conversational na- nitive sophistication (e.g., Garrison & Anderson,
ture, often takes the form of a hybrid code, mixing 2003, p.26; Heckman & Annabi, 2005; Hiltz &
together features of speech and writing with its Goldman, 2005, p.6). This may be because “[t]
own peculiar elements (Crystal, 2001a, 2001b; he historical divide between speech and writing
cf. Al-Sa’Di & Hamdan, 2005, on synchronous has been overcome with the interactional and
chat). It is worth bearing in mind, then, that DBs reflective aspects of language merged in a single
may not only help students learn about standard medium” (Warschauer, 1999, p.6). This particular
spoken and written language, but about hybridized aspect of online hybridity would certainly seem
language uses of the kind with which they need to have major advantages.
to be familiar in order to enter fully into many When they involve multilingual or multicul-
online environments. tural cohorts of students, DBs may equally promote
If structured carefully, asynchronous DBs can the development of intercultural competence. In
promote the formation of learning communities the ongoing Third Space in Online Discussion
where students, reacting to and building on each research project, which involves language teachers
other’s ideas in branching discussion threads, enrolled in Master’s courses at the University of
collaboratively construct their understandings Western Australia and Canterbury Christ Church
of the subject matter at hand — all through the University, UK, discussion forums (like that seen in
medium of written language, which is probably Figure 1) are being analyzed as educational “third
more conducive to reflective educational dis- spaces” which exist in the interstices between

1110
Communicative Networking and Linguistic Mashups on Web 2.0

students’ cultural and educational experiences, and can present more nuanced versions of themselves.
where there is ample space for the deconstruction As Kazan indicates, “[w]ithin cyberspace, writers
and reconstruction of pedagogical, linguistic and have flexibility in how they construct a self and the
cultural knowledge and understanding (Pegrum & more strategies they acquire, the more flexibility
Bax, 2007). It is apparent that, as Zieghahn (2001) they have” (2007, p. 264). The task for teachers
realized some years ago, “the online environment is to help students make more “informed rhetori-
offers a unique medium through which to reflect cal decisions” (ibid.), which will allow them not
upon individual cultural position and on inter- only to shape their online identities as they wish,
cultural communication” (p. 144). While most but also to “develop a public voice about issues
educational DBs necessarily operate in a single they care about” and so come to understand “their
lingua franca, multilingual forums are possible in literacies as citizenship skills as well as avenues
some language learning situations. Linguistically to entertainment” (Rheingold, 2007).
as well as culturally, then, DBs can help educators Wikis are even more strongly oriented towards
respond to Canagarajah’s aforementioned plea to collaboration than blogs since they are effectively
teach students to shuttle between communities. co-operatively authored websites. They turn the
In the process, their sense of their online — and element of collective intelligence implicit in
perhaps offline — selves may be shaped through blogging communities into a structural principle.
their interactions with peers. Students are able to engage in a form of process
While blogs — described by Doctorow (2002) writing in which they draft and redraft work
as “outboard brain[s]” — can function as reflective collaboratively, each contributor adding to and
diaries, they can also be conversational centre- modifying the work of peers. With a private wiki,
pieces: readers may leave comments for a blog’s feedback can be received from the class teacher
author and each other, thereby forging connections and peers, or, with a public wiki, from the entire
and community around topics of mutual interest. Internet. As Mitchell (2005) notes, it has even
Students can certainly join the conversations on been suggested that wikis are an example of “the
others’ blogs, but they can equally set up their tried and trusted system of peer review taken to
own. Receiving feedback on blog entries from a new level” (p.120).
peers and teachers can facilitate knowledge con- One option is for students to contribute to pre-
struction as well as perspective shifts as they go existing wikis such as Wikipedia or, for learners,
about developing their online personas. Indeed, Simple English Wikipedia, thereby entering into
with fully public blogs, students can potentially established communities of practice. Alternatively,
receive feedback from anyone on the entire Internet dedicated class wikis can be set up on subjects of
and may, as a result, invest themselves more fully relevance or interest, and in time new communities
in writing and publishing tasks. of practice may form around these. Even a course
Because blogs can be multilingual (allowing constrained by a tight, exam-oriented syllabus can
some mixing of the mother tongue with the target exploit wiki technology: under the guidance of
language), multimodal (allowing pictures, video the teacher, each individual’s or group’s research
and audio to support written text), and carefully could feed into a network of student-constructed
designed (drawing on technical knowledge and documents reviewing material to be covered in
artistic flair), students at even the lowest levels of the exam. This might include vocabulary ac-
linguistic proficiency need not feel the work they companied by definitions and examples; gram-
are creating fails to capture or express important as- mar points accompanied by explanations and
pects of their identities or beliefs. At higher levels, illustrations; or set literature accompanied by
as students’ linguistic competence develops, they summaries and quotations. Once again, there is

1111
Communicative Networking and Linguistic Mashups on Web 2.0

ample opportunity for multilingual, multimodal, important is how you relate, emotionally, and
technically sophisticated and artistically creative physically, to that world.
presentation. The more sophisticated the wiki, the
greater the students’ facility with multiliteracies Social networking sites, with MySpace and
will need to be — or become. Facebook being by far the most popular, allow
each user to set up an online identity, or profile,
and to keep in touch with friends and acquain-
Social Networking tances by constantly updating this profile while
regularly viewing others’ profiles; new contacts
Social networking technologies also promote can be established through mutual acquaintances
collaborative thinking, many of them effectively or shared interests. Since 2006, Facebook has
harnessing the power of collective intelligence, used a news feed system to keep users updated
but the accent is on the networking aspect. It has on changes to the profiles of their contacts. Typi-
been suggested that Facebook, for example, “puts cally, social networking sites integrate a range
the social community first, with content — includ- of other communication channels, which may
ing, but not limited to, educational content—being include email, instant messaging (IM) and even
the medium of exchange” (Downes, 2007). Some blogs, with facilities for sharing photos, videos
observers claim that virtual networks are replac- and audio files. There is a fine line separating
ing the gradually disappearing or increasingly these sites from social sharing services, such as
inaccessible public spaces in which young people Flickr for photos or YouTube for videos. Facebook
formerly gathered (boyd, 2006). These networks allows the integration of Flickr photos as well as
are intimately bound up with selfhood; the sense del.icio.us tags (see below) into profiles, while it
of empowerment that comes from the crafting of is now also possible for users to assemble friends
personal identity on social sites (Coghlan, 2007) and acquaintances from the virtual world Second
goes hand in hand with negotiating membership Life alongside their other contacts.
of the groups of friends and acquaintances who Social networking sites are perhaps the most
congregate there. The potential effects on language maligned feature of Web 2.0, mainly due to
education are an extension of the paradigm shift fears of Internet predation but also because of
neatly captured at the start of the millennium by concerns over time spent online, as well as the
Kramsch, A’Ness & Lam (2000, p. 97) in their possible degeneration of literacy skills as the
comments on language learning through partici- digital natives communicate ever more rapidly in
pation in informal online interaction: ever more truncated “netspeak.” Yet the reality is
that students are already using social networking
The kind of language experience … in which rules sites and educators have the choice to work with
are learned first and then put to use in conversa- or against them. The advantage of the former
tion, has given way to a learning by doing, and strategy is that it is possible to openly address
learning to meet the demands of doing in specific concerns over Internet safety or time spent online,
contexts, to solve immediate problems together attempting to provide guidance in such areas.
in the small culture of communities of practice This might be extended to include a focus on
(Holliday, 1999; Lave & Wenger, 1991; Uber what Barney (2007, p. 279) refers to as “critical
Grosse & Leto, 1999; Wenger, 1999). Rather than technological literacy”: asking questions about
an object of reverence or study in itself, language the presuppositions and blind spots, the benefits
is viewed as a tool which brings people together and drawbacks, in short, the “affordances and
and creates intimacy (Harmon, 1999). What is … denials” of different technologies. Helping

1112
Communicative Networking and Linguistic Mashups on Web 2.0

students adopt a critical distance to all technolo- portunity to explore with students the nature and
gies will do them a much greater service in the uses of netspeak, when and where it is appropri-
long run than simply closing down all discussion ate, and how to codeswitch between netspeak and
in the classroom, leaving them to conduct their more standardised language forms. One of the
explorations, unguided, in their own time. main reasons for the widely criticized spread of
At the same time, there are many education- netspeak into more traditional domains of literacy
ally beneficial aspects of social sites which can may well be students’ ignorance of codeswitch-
be more fully exploited. According to recent US ing or their inability to carry it out appropriately.
statistics, some 59% of 9-17 year olds say they Teachers’ failure to explicitly address this area
talk about topics broadly related to education with students can only limit the latter’s repertoire
on social networking sites, while 50% claim to of literacies and constrain their ability to access
discuss schoolwork (National School Boards As- and move between linguistic communities.
sociation, 2007, p.1). Thus, whatever educators
may think, students have already appropriated
social networking as a constructivist learning Information Linking
tool. However, educators could certainly do more
to encourage the use of this tool for groupwork Folksonomies are a step beyond social sharing.
outside the classroom. The potential for language Relying very much on the principle of collective
learning partnerships is undoubtedly great. Lak- intelligence, they are a way of indexing distributed
shimi (2007), discussing her English language knowledge, which is then typically presented in the
students’ use of the social networking site Orkut, semi-organic form of a tag cloud, as seen in Figure
comments: “Students who have been incommuni- 2. In essence, they allow information linking with
cado in the classroom are so interactive on Orkut a social element, because people (the “folk”) have
that it leaves me wondering if Orkut would be a a central organizing role, which gives rise to rich
better teacher than I am in helping students learn “person-mediated serendipity” (Lambe, 2006,
to use English to be socially interactive” (n.p.). n.p.). After all, people who use the same tags are
Interaction, of course, is precisely the motivation: likely to have similar interests; and a folksonomy
the wish to communicate and participate, with allows tags to be traced to users, and those users’
language being an essential tool. other tags to be explored. The potential for “col-
Social sharing sites offer the additional pos- laborative information discovery” (Alexander,
sibility of posting individual or collaborative work 2006, p. 36) may be exploited by students working
to the web, with students viewing each other’s together to create class folksonomies dependent on
materials and, for example, commenting on their criteria negotiated and evaluated by the students
peers’ photographed posters (Flickr), PowerPoint themselves. This could even involve the tagging
slides (Slideshare), presentations (YouTube) or of the students’ own material posted on wikis
short films posted to blogs (such as the English or social sharing sites. Given the usefulness of
Advertising Class). As Coghlan (2007) observes well-constructed folksonomies, they might also
with regard to student-created advertisements on be consulted by members of wider communities
the last of these sites, some examples may involve of practice on the Internet and could provide a
little traditional language use, but there is a lot of means of entry into such communities; as Wenger
learning potential in the areas of “multiliteracy, (1998) reminds readers, learning communities
digital literacy and e-literacy.” should not be isolated but should “use the world
The communication on social networking and around them as a learning resource and be a
social sharing sites offers, finally, a unique op- learning resource for the world” (p. 275). In all

1113
Communicative Networking and Linguistic Mashups on Web 2.0

Figure 2. Extract from E-language Tag Cloud (http://e-language.wikispaces.com/e-learning-tagcloud)

cases, tagging, like indexing of any kind, requires both native and non-native feeds would lead to a
a high level of facility with the language being rich patchwork of first and additional language
used for classification. With sufficient scaffolding, usage, approximating in some ways the multi-
folksonomy building can function as a literacy dialectal reality of today’s world. Awareness of
enhancement exercise. multiliteracies can be enhanced through feeds
RSS (Really Simple Syndication) feeds provide which distribute audio or video content in addi-
automatic updates of syndicated content — rang- tion to or in place of written text.
ing from blog entries to podcasts—from sites to
which a user subscribes. Many homepage, blog
and wiki services now make it very easy to include Mashup Frontiers
selected RSS feeds on webpages. Drawing in feeds
from other sites in this way amounts to the incor- Some of the greatest educational promise is to
poration of others’ views and perspectives, leading be found in the areas of podcasting, vodcasting,
to the co-construction of knowledge within a new m-learning and virtual worlds, all of which offer
frame. At the same time, as Anderson (2006) notes considerable language learning opportunities, es-
with respect to blog feeds, distribution of content pecially for those prepared to work with multiple
by RSS allows “public review, argument and literacies and language mashups.
resolution of topic issues by students globally— M-learning refers to education involving
in the process creating outstanding international mobile technology. The best-known example is
learning opportunities” (p. 146). podcasting, where syndicated audio files, poten-
Incoming feeds naturally entail a constant tially with accompanying text or image files, are
stream of information flowing into a desktop downloaded from the web and transferred to a
aggregator or webpage. The language could portable device such as an iPod or MP3 player, thus
be that of native speakers; thus, learners could facilitating “time and place shifting to access the
conceivably subscribe to media or blog feeds in content” (Molina & 2006 EDUCAUSE Evolving
languages they know or are learning, and would Technologies Committee, 2006, p. 122). Listening
be exposed to extensive authentic input. There to podcasts is widely perceived as advantageous
is also an argument, however, for subscribing to for learning foreign languages or even brushing
non-native language feeds. For example, TESOL up on grammar, vocabulary or style in one’s
students working in a World Englishes paradigm first language. Surveying a selection of national
might find it beneficial to subscribe to feeds from iTunes stores on the randomly selected date of
Kachru’s outer or expanding circles. Incorporating 17 October, 2007, for instance, it was found that

1114
Communicative Networking and Linguistic Mashups on Web 2.0

the majority of the 25 most popular educational HARP, conducted at Harvard University in the
podcasts in each country were related to foreign USA (Harvard University, n.d.).
language learning or first language improvement: Virtual worlds are perhaps the most striking re-
24 in Spain, 22 in Germany and Switzerland, 21 alization of the possibilities of Web 2.0. The avatars
in Australia, New Zealand and the UK, 20 in which inhabit them are certainly Web 2.0’s clear-
Canada and Ireland, 19 in France, Sweden and est example of the potential for identity creation,
the US, and 17 in Italy. shaping and development. These worlds are very
M-learning can also involve regularly send- much about networking. Within them, avatars’
ing students digestible chunks of information via understandings of their new environment are con-
mobile phones, as has been done, for example, structed largely through their engagement — their
with Italian vocabulary accompanied by quiz- sharing and building of knowledge — with other
zes at Griffith University in Australia (Levy & avatars. Externally, virtual worlds are supported
Kennedy, 2005). However, there is the potential by and increasingly integrated with blogs, wikis,
for greater levels of interactivity than this, as and social networking sites. Operating around
suggested in a recently proposed definition of and through these sites are distributed knowledge
m-learning as “the processes of coming to know systems where, as in the gaming communities
through conversations across multiple contexts discussed by Williamson and Facer (2004), the
among people and personal interactive technolo- key information is found “in the interconnections
gies” (Sharples, Taylor & Vavoula, 2007, p.225). between the ‘nodes’ (the people, texts, tools and
For example, students can work individually or, technologies) in the network, rather than with
better still, collaboratively to create podcasts or isolated individuals” (p. 266, with reference to
even vodcasts – as video podcasts are usually Gee). In a comment which captures something of
known – for publication to the web. Moblogging, the richness of the virtual/non-virtual interface, the
or mobile blogging, allows students to use devices best-known of these worlds, Second Life (SL), has
like mobile phones to post text, audio or video files been described as “a playground [and] a crucible
to blogs. Peers and teachers can then respond to for ideas about how people can augment their in-
these postings in traditional text or mixed-media teraction through constructive, and constructivist,
formats, addressing the communicative intent play/work/whatever” (Stevens, 2007, n.p.).
while possibly also critiquing features of language Since the rollout of voice technology to SL in
or composition. In many cases, spoken language mid-2007, in-world avatar-to-avatar interactions
will be foregrounded, thus helping to balance out can involve a mixture of spoken and written lan-
the heavy emphasis on written text still typical of guage not unlike that found in the real world. This
the web, including Web 2.0. Sometimes there may creates valuable opportunities for students to try
be room for multiple linguistic codes and registers out new language, building up confidence and flu-
if not multiple dialects or languages. More sophis- ency before embarking on real world encounters.
ticated versions of m-learning involve participants Language teachers have been quick to pick up
interacting with real-world environments and each on this potential, with the inaugural SLanguages
other with the aid of GPS-enabled phones and other Colloquium taking place on 23 June, 2007, and
portable devices, which may provide instructions bringing together around 50 educators from across
and information as well as a variety of communi- the globe; a snapshot of the opening talk by Gavin
cation channels; salient examples range from the Dudeney is shown in Figure 3. Language teaching
MOBIlearn Uffizi Gallery trial in Florence, Italy is already underway in SL, with English classes
(Sharples, Taylor & Vavoula, 2007, pp.236-242) on offer, for example, through The English Village
to the Handheld Augmented Reality Project, or and Languagelab.com. SL also offers immersive

1115
Communicative Networking and Linguistic Mashups on Web 2.0

Figure 3. Inaugural SLanguages Colloquium on EduNation in Second Life, 23 June 2007. Reproduced
by kind permission of Gavin Dudeney, EduNation.

linguistic experiences outside formal classes, a evening” (personal communication, 11 Oct. 2007).
point emphasised at the inaugural in-world Festival Vance Stevens (2007) quotes a comment about
of European Languages in 2007, which promoted SL which hints at intriguing language education
the idea of learners seeking out target language possibilities: “Yesterday a cheerful Italian gave me
areas of SL in which to practise their skills. a Babbler translator so we started teaching each
A certain degree of linguistic versatility is ad- other Italian and Hungarian using English as the
vantageous for anyone wishing to develop a fuller common language, which was real fun, especially
SL presence, since language is the glue which holds that we were figure ice-skating meanwhile” (n.p.).
together any community which establishes itself Participation in such conversations — and teach-
there. Community, in fact, has been described as ing scenarios — requires a willingness to engage
the killer app of SL (Yowell, cited in Panganiban, with the unruliness of linguistic globalization as
2007). Different languages, and certainly different reflected through virtual world encounters. It de-
dialects and registers, are necessary for effective mands a capacity to codeswitch and a facility with
participation in a range of contexts and communi- intercultural communicative competence skills: in
ties, with an increased linguistic repertoire being a short, the agility to shuttle between linguistic and
concomitant of increased community involvement cultural communities. This, in turn, reads like a
and wider social networking. This is not unlike set of lesson aims compiled from recent thinking
the real world, except it is now possible to cross on language pedagogy. While a single target lan-
linguistic and cultural boundaries without leaving guage will necessarily remain the focus of most
one’s desk. language lessons — and can be supported with SL
Anecdotal evidence suggests multilingual in- immersion experiences — there is no reason why
teractions in SL are becoming more common. A students should not occasionally be exposed to
striking example of a four-person, five-language multilingual, multicultural interactions, especially
(Catalan, English, French, Portuguese and Span- as these are likely to become ever more central not
ish) conversation has been described by Gavin only to the SL microcosm, but to the wider web,
Dudeney, who writes of “the ease with which and indeed the world which lies beyond it.
some of us switched between the languages we
knew, and typed furiously to reformulate things we
thought one of the others wouldn’t understand into
a language they would,” resulting in “a very rich

1116
Communicative Networking and Linguistic Mashups on Web 2.0

Limitations of Web their current state of “continuous partial attention”


2.0 in Education (Stone, 2006), technology users will lose the ability
to focus clearly as well as the will to occasionally
This chapter has discussed the potential of Web power down their multifarious communication
2.0 for education generally and language education channels and make time for reflection—a crucial
specifically. However, it will take time for current part of education (Pegrum, 2005). And, even while
practices to become more widespread and for the acknowledging the benefits of constructivism, it
potential of Web 2.0 to be fully realized. This might be asked whether it is possible or desirable
requires further “normalization” of computing, to teach everything in a constructivist manner all
so that the majority of educators eventually come of the time. It is important to maintain balance in
to regard it with neither fear nor awe, but see it all of the above areas.
as simply providing a set of tools which may be If students are already spending a lot of time
used in the service of particular pedagogical goals online, added educational demands should have a
(Bax, 2003; Chambers & Bax, 2006). Teacher clear value. The identity issues permeating online
training has a major role to play in demystifying presence are complex and delicate, and educators
computing. Specifically, this entails providing should beware of aggravating narcissistic tenden-
teachers with appropriate pedagogical frameworks cies which may be nourished by social networking
for e-learning; an overview of the range of tools (Ryan, 2007). Teachers must also face the fact
available; and adequate technological skills so they may not be welcome to approach students
that they do not feel intimidated by their students’ on some sites and through some channels; sen-
know-how and, moreover, have the confidence sitivity is needed in negotiating educational uses
to draw on the latter’s technological expertise to with students.
complement their own pedagogical expertise. In Collaborative work raises questions of author-
addition, Web 2.0 provides very serviceable tools ship and ownership, while non-participation is
for building social constructivist professional often not an option, as Conrad (2002) has noted
development forums, and it is possible to imagine with regard to virtual learning environments: “you
that in time “Web 2.0 may well become the biggest cannot run and you cannot hide. Online life is a
training institution in the world” (Consultants-E, fishbowl existence” (p.208; italics in original).
2007). This point will only be reached, however, There is some cause for concern over privacy
if intensive preparatory work is carried out by on social networking sites like Facebook (boyd,
today’s teacher training institutions. 2008). What is more, a lot of online material is
While learning about the advantages of Web preserved indefinitely so that, as Friedman (2006)
2.0, teachers must equally come to understand warns in a more general context, “whatever you
that e-learning is not, in and of itself, automati- do, whatever mistakes you make, will be search-
cally constructivist or pedagogically progressive able one day” (p. 185).
(Pegrum, forthcoming 2008b), and demands for Of course, the continued presence of a digital
speed, flexibility and cost saving can easily lead divide—or, more accurately perhaps, a digital
to impoverished content delivery systems. As spectrum (Haythornthwaite, 2007)—means that
suggested earlier, some creativity is needed to not everyone around the world, or within any
work within the constraints of rigid syllabi or given society, has equal access to the Internet.
assessments. As rewarding as it may be, well- While the rapid spread of mobile technologies
designed online learning will normally require partially alleviates this situation, it is not the end
a heavy investment of time and energy by both of the issue: in recent years, the digital divide
staff and students. There is also a danger that, in has come to be seen less in terms of access to

1117
Communicative Networking and Linguistic Mashups on Web 2.0

technology and more in terms of skills and pat- Vodafone’s InsideOut, which allows calls between
terns of use (ibid.; Warschauer, 2003) or, in short, the physical world and Second Life.
digital literacy. It should also be remembered that For this reason, despite initial evidence which
“global communication technologies are cultural points to a lowering of cognitive performance
artifacts that are produced by and productive and efficiency through multitasking (Baron,
of socio-historically located subjects” (Belz & 2008; Wallis, 2006), it is possible that students
Thorne 2006, p. xviii), and that they carry the who monitor multiple IM channels while writing
Anglo-centric and, more broadly, Western values assignments or who send text messages during
of their creators (Ess, 2007; Goodfellow, 2003; lectures are engaging in what, for them, is “a
Reeder, Macfadyen, Chase & Roche, 2004). natural way to interact and construct their own
Students from varying linguistic, cultural, ethnic, learning” (Reddekopp, 2006). Through practice,
religious, social and educational backgrounds may they may have adapted to such behaviour (Baron,
have their reasons for not wishing to participate 2008). What if, moreover, such a melding of
in some or all online activities—reasons whose learning, networking and identity building could
legitimacy is often eclipsed in Western secular give rise to lateral connections and a more holistic
education. Compromises must be sought with mode of education? In the absence of empirical
students who, for example, may struggle with the evidence, these reflections are necessarily specu-
radically egalitarian nature of social networking lative. However, it is important not to close off
technologies or who, as Sabre (2007) notes, might new possibilities before they are fully apparent,
be uncomfortable with virtual worlds because of thereby perhaps losing valuable educational
religious prohibitions on graphic representations opportunities—and losing students’ allegiance
of humans. along the way. While the digital natives have much
But perhaps the greatest single issue for would- to learn about language and literacy from an older
be Web 2.0 educators may be an inability to step generation of teachers, the teaching profession as
outside traditional philosophical and sociopolitical a whole has much to learn from its digital native
frames of reference. This could mean an inability students, especially here at the technological and
to see outside the frame of Enlightenment ratio- social frontier of Web 2.0.
nalism and objectivism and to grasp the socially
constructed nature of knowledge and learning, a
fundamental flaw in Keen’s timely if hyperbolic Conclusion
critique, The Cult of the Amateur (2007). It might
mean an inability to value collaboration and com- The technologies covered in this chapter — discus-
munity on their own terms outside of a capitalist sion boards, blogs, wikis, social networking, social
paradigm of competition, as seen in Tapscott sharing, folksonomies, RSS, podcasting, vodcast-
and Williams’ otherwise informative Wikinomics ing, m-learning and virtual worlds — comprise a
(2006). It could mean an inability to perceive that, representative Web 2.0 list, but one which is both
for the net generation, the notion of a prophylactic incomplete and unstable. New technologies and
divide between “virtual” and “real” life makes applications are constantly appearing, while there
little sense: like the radio or the telephone for is an overall tendency towards functional conver-
older generations, the virtual is just another part gence. Yet, however this list might look a few years
of the real (cf. Davies, 2003; Thorne, 2006, p. from now, it is likely to still be informed by the
20). In fact, the connections between them are fundamental features this chapter has described as
becoming ubiquitous, as seen in services such as underpinning Web 2.0: communicative network-

1118
Communicative Networking and Linguistic Mashups on Web 2.0

ing, community building and identity negotiation, users of language(s) within — and beyond — the
performed through hybrid codes, multiple media digital environment, they need to adopt an open,
and linguacultural mashups. exploratory and flexible attitude. They need to
Writing of Web 2.0, McIntosh (2006) suggests appreciate and work with the social orientation of
that “[t]he reason these social technologies work is Web 2.0. They need to become comfortable with
because they are social. But they are also changing linguistic and media mashups and actively foster
the way that we socialise” (p. 72). As has been the codeswitching and shuttling skills demanded
seen, socializing and networking on Web 2.0 are by the untidy realities of globalization, on- and
very much dependent on language. Web 2.0 is, offline. And, while continuing to provide the same
after all, “a means whereby just about anyone can level of educational input and guidance as good
contribute to an ongoing ‘conversation’ in which teachers have always done, they need to trust the
knowledge is both discovered and constructed as digital natives to help them map what, for educa-
it goes on” (Freedman, 2006, p.13), and there can tion, is still largely uncharted territory.
be no conversation without language. It is little
wonder, then, that Crystal (2001a, 2001b) has
called the Internet a linguistic revolution; that References
Macfadyen & Doff (2005, following Cicognani)
have claimed that cyberspace must be viewed in Al-Sa’Di, R. A., & Hamdan, J. M. (2005).
linguistic terms; or that some observers feel Web Synchronous online chat English: Computer-
2.0 comes close to realizing Tim Berners-Lee’s mediated communication. World Englishes, 24(4),
original idea of the web as a “read-write medium” 409–424.
(Lee & Berry, 2006, p. 20). Alexander, B. (2006). Web 2.0: A new wave of
It has been suggested in this chapter that innovation for teaching and learning? EDUCAUSE
language and literacy educators are in an ideal Review, 41(2), 32-44. Retrieved November 29,
position to exploit the linguistic nature of Web 2.0. 2007, from http://www.educause.edu/ir/library/
This requires a conception of literacy – indeed, pdf/erm0621.pdf
of multiliteracies – which is appropriate to Web
2.0 and the increasingly interconnected world of Anderson, T. (2006). Interaction in learning and
which it is both a symbol and a product. It re- teaching on the educational Semantic Web. In C.
quires a suitable pedagogical base for e-learning, Juwah (Ed.), Interactions in online education:
drawing on social constructivism, communities of Implications for theory and practice (pp.141-155).
practice, intercultural communicative competence London: Routledge.
and identity studies. It requires a familiarity with
Barney, D. (2007). The question of education in
the advantages and drawbacks of each Web 2.0
technological society. In J. Lockard & M. Pegrum
tool, coupled with an ability to tailor such tools
(Eds.), Brave new classrooms: Democratic educa-
to particular cultural contexts. It requires some
tion and the Internet (pp. 271-284). New York:
reflection on how to address pedagogical, social,
Peter Lang.
sociopolitical and philosophical limitations on the
use of Web 2.0 in education. In all of the above, Baron, N. (2008). Adjusting the volume: Technol-
teacher training has an important role to play. ogy and multitasking in discourse control. In J.E.
Beyond this, if language and literacy educa- Katz (Ed.), Handbook of mobile communication
tors are to fully exploit the potential of Web 2.0 studies (pp.177-194). Cambridge, MA: MIT
as a platform to enhance language teaching and Press.
to help their students become more sophisticated

1119
Communicative Networking and Linguistic Mashups on Web 2.0

Barton, D., & Hamilton, M. (2000). Literacy Coleman, J. (2007). Eeyore and the pixel drop-
practices. In D. Barton, M. Hamilton & R. Ivanič out: What’s wrong with technology-enhanced
(Eds.), Situated literacies: Reading and writing in learning? Paper presented at New and Emerging
context (pp. 7-15). London: Routledge. Technologies in ELT, Loyola College, Chennai,
India, 3-5 Aug.
Bax, S. (2003). CALL – Past, present and fu-
ture. System, 31, 13–28. doi:10.1016/S0346- Conrad, D. (2002). Inhibition, integrity and
251X(02)00071-4 etiquette among online learners: The art of
niceness. Distance Education, 23(2), 197–212.
Belz, J. A., & Thorne, S. L. (2006). Introduction:
doi:10.1080/0158791022000009204
Internet-mediated intercultural foreign language
education and the intercultural speaker. In J.A. Cope, B., & Kalantzis, M. (Eds.). (2000). Mul-
Belz & S.L. Thorne (Eds.), Internet-mediated tiliteracies: Literacy learning and the design of
intercultural foreign language education (pp. viii- social futures. London: Routledge.
xxv). Boston: Heinle & Heinle. boyd, d. (2006).
Corbett, J. (2003). An intercultural approach to
Identity production in a networked culture: Why
English language teaching. Clevedon: Multilin-
youth heart MySpace. Paper presented to American
gual Matters.
Association for the Advancement of Science, 19
February. Retrieved November 29, 2007, from Crystal, D. (2001a). Language and the Internet.
http://www.danah.org/papers/AAAS2006.html Cambridge: Cambridge University Press.
boyd, d. (2008). Facebook’s privacy trainwreck:
Crystal, D. (2001b). A linguistic revolution?
Exposure, invasion, and social convergence.
Education Communication and Information,
Convergence, 14(1), 13-20.
1(2), 93–97.
Burgmer, C. (2006). Computer-mediated com-
Dalgarno, B. (2001). Interpretations of con-
munication (CMC): A beneficial tool for modern
structivism and consequences for computer as-
language learning? Unpublished M. Phil. dis-
sisted learning. British Journal of Educational
sertation, University of Dundee, UK.
Technology, 32(2), 183–194. doi:10.1111/1467-
Byram, M. (1997). Teaching and assessing inter- 8535.00189
cultural communicative competence. Clevedon:
Davies, W. (2003). You don’t know me but…
Multilingual Matters.
Social capital and social software. London: The
Canagarajah, S. (2003). Foreword. In G. Smither- Work Foundation. Retrieved November 29, 2007,
man & V. Villanueva (Eds.), Language diversity from http://www.theworkfoundation.com/Assets/
in the classroom: From intention to practice PDFs/you_dontknowme.pdf
(pp. ix-xiv). Carbondale, IL: Southern Illinois
Doctorow, C. (2002). My blog, my outboard brain.
University Press.
O’Reilly WebDev Center, 31 May. Retrieved No-
Chambers, A., & Bax, S. (2006). Making CALL vember 29, 2007, from http://www.oreillynet.com/
work: Towards normalisation. System, 34, 465– lpt/a/javascript/2002/01/01/cory.html
479. doi:10.1016/j.system.2006.08.001
Downes, S. (2007). Places to go: Facebook.
Coghlan, M. (2007). Language learning in a Innovate, 4(1). Retrieved November 29, 2007,
connected world. Paper presented at New and from http://www.innovateonline.info/index.
Emerging Technologies in ELT, Loyola College, php?view=article&id=517
Chennai, India, 3-5 Aug.

1120
Communicative Networking and Linguistic Mashups on Web 2.0

Dudfield, A. (1999). Literacy and cyberculture. Heckman, R., & Annabi, H. (2005). A content
Reading Online. Retrieved November 29, 2007, analytic comparison of learning processes in online
from http://www.readingonline.org/articles/dud- and face-to-face case study discussions. Journal
field/main.html of Computer-Mediated Communication, 10(2).
Retrieved November 29, 2007, from http://jcmc.
Ess, C. (2007). Liberal arts and distance education:
indiana.edu/vol10/issue2/heckman.html
Can Socratic virtue and Confucius’ exemplary
person be taught online? In J. Lockard & M. Pe- Hiltz, S. R., & Goldman, R. (2005). What are
grum (Eds.), Brave new classrooms: Democratic asynchronous learning networks? In S.R. Hiltz
education and the Internet (pp. 189-212). New & R. Goldman (Eds.), Learning together online:
York: Peter Lang. Research on asynchronous learning networks
(pp.3-18). Mahwah, NJ: Lawrence Erlbaum.
Finger, G., Russell, G., Jamieson-Proctor, R.,
& Russell, N. (2007). Transforming learning Holmes, B., Tangney, B., FitzGibbon, A., Savage,
with ICT: Making it happen. Frenchs Forest: T., & Mehan, S. (2001). Communal constructiv-
Pearson. ism: Students constructing learning for as well as
with others. Dublin: University of Dublin Depart-
Freedman, T. (2006). Introduction. In T. Freedman
ment of Computer Science. Retrieved November
(Ed.), Coming of age: An introduction to the new
29, 2007, from https://www.cs.tcd.ie/publications/
world wide web (pp.13-14). Ilford: Terry Freed-
tech-reports/reports.01/TCD-CS-2001-04.pdf
man. Retrieved November 29, 2007, from http://
fullmeasure.co.uk/Coming_of_age_v1-2.pdf Hoppe, H. U., Joiner, R., Milrad, M., & Sharples,
M. (2003). Guest editorial: Wireless and mobile
Friedman, T. L. (2006). The world is flat: The
technologies in education. Journal of Computer
globalized world in the twenty-first century (2nd
Assisted Learning, 19(3), 255–259. doi:10.1046/
ed.). London: Penguin.
j.0266-4909.2003.00027.x
Garrison, D. R., & Anderson, T. (2003). E-learning
Hubbard, P. (2004). Learner training for effective
in the 21st century: A framework for research and
use of CALL. In S. Fotos & C. Browne (Eds.),
practice. London: Routledge Falmer.
New perspectives on CALL for second language
Goodfellow, R. (2003). Literacies, technologies, classrooms (pp. 45-68). Mahwah, NJ: Lawrence
and learning communities: Speaking and writing Erlbaum.
in the virtual classroom. Paper presented at Eu-
Jonassen, D. H. (1992). Evaluating constructivistic
roCALL 03, Limerick, Ireland, 5 Sept. Retrieved
learning. In T. M. Duffy & D. H. Jonassen (Eds.),
November 29, 2007, from http://iet.open.ac.uk/
Constructivism and the technology of instruction:
pp/r.goodfellow/Euroc03/talk.htm
A conversation (pp. 137-148). Hillsdale, NJ:
Harvard University. (n.d.). Handheld augmented Lawrence Erlbaum.
reality project (HARP). Retrieved November
Kaye, A. (1989). Computer-mediated commu-
29, 2007, from http://isites.harvard.edu/icb/icb.
nication and distance education. In R. Mason &
do?keyword=harp
A.R. Kaye (Eds.), Mindweave: Communication,
Haythornthwaite, C. (2007). Digital divide and computers and distance education (pp. 3-21).
e-learning. In R. Andrews & C. Haythornthwaite Oxford: Pergamon.
(Eds.), The Sage handbook of e-learning research
(pp.97-118). Los Angeles: Sage.

1121
Communicative Networking and Linguistic Mashups on Web 2.0

Kazan, T. S. (2007). Braving the body: Embodi- Lee, S., & Berry, M. (2006). Effective e-learning
ment and (cyber-)texts. In J. Lockard & M. Pe- through collaboration. In T. Freedman (Ed.), Com-
grum (Eds.), Brave new classrooms: Democratic ing of age: An introduction to the new world wide
education and the Internet (pp. 251-269). New web (pp. 19-28). Ilford: Terry Freedman. Retrieved
York: Peter Lang. November 29, 2007, from http://fullmeasure.
co.uk/Coming_of_age_v1-2.pdf
Keen, A. (2007). The cult of the amateur: How
today’s Internet is killing our culture. London: Levy, M., & Kennedy, C. (2005). Learning Ital-
Nicholas Brealey. ian via mobile SMS. In A. Kukulska-Hulme &
J. Traxler (Eds.), Mobile learning: A handbook
Kern, R. (2006). Perspectives on technology in
for educators and trainers (pp. 76-83). London:
learning and teaching languages. TESOL Quar-
Routledge.
terly, 40(1), 183–210.
Levy, M., & Stockwell, G. (2006). CALL dimen-
Kist, W. (2004). New literacies in action: Teach-
sions: Options and issues in computer assisted
ing and learning in multiple media. New York:
language learning. Mahwah, NJ: Lawrence Er-
Teachers’ College Press.
lbaum.
Kramsch, C. (1993). Context and culture in
Locke, T. (2007). E-learning and the reshaping of
language teaching. Oxford: Oxford University
rhetorical space. In R. Andrews & C. Haythorn-
Press.
thwaite (Eds.), The Sage handbook of e-learning
Kramsch, C. (1998). The privilege of the intercul- research (pp. 179-201). London: Sage.
tural speaker. In M. Byram & M. Fleming (Eds.),
Macfadyen, L. P., & Doff, S. (2005). The language
Language learning in intercultural perspective:
of cyberspace: Text, discourse, cultural tool. In C.
Approaches through drama and ethnography
Ghaoui (Ed.), Encyclopedia of human-computer
(pp. 16-31). Cambridge: Cambridge University
interaction. Hershey, PA: The Idea Group. Re-
Press.
trieved November 1, 2006, from http://homepage.
Kramsch, C., A’Ness, F., & Lam, W. S. E. (2000). mac.com/leahmac/LM/Docs/C65MacfadyenD-
Authenticity and authorship in the computer- off.pdf
mediated acquisition of L2 literacy. Language
McFarlane, A., Roche, N., & Triggs, P. (2007).
Learning & Technology, 4(2). Retrieved 29 Nov.,
Mobile learning: Research findings. Report to
2007 from http://llt.msu.edu/vol4num2/kramsch/
BECTA. Coventry: BECTA. Retrieved Janu-
default.html
ary 14, 2008, from http://partners.becta.org.uk/
Lakshimi, S. K. C. (2007). The role of teachers and upload-dir/downloads/page_documents/research/
the modern tool of online community networks in mobile_learning_july07.pdf
facilitating social interaction. Paper presented at
McIntosh, E. (2006). Podcasting and wikis. In T.
New and Emerging Technologies in ELT, Loyola
Freedman (Ed.), Coming of age: An introduction to
College, Chennai, India, 3-5 Aug.
the new World Wide Web (pp.71-75). Ilford: Terry
Lambe, P. (2006). Folksonomies and rich serendip- Freedman. Retrieved November 29, 2007, from
ity. Green Chameleon, 20 Oct. Retrieved Novem- http://fullmeasure.co.uk/Coming_of_age_v1-2.
ber 29, 2007, from http://www.greenchameleon. pdf
com/gc/blog_detail/folksonomies_and_rich_ser-
endipity/

1122
Communicative Networking and Linguistic Mashups on Web 2.0

Mitchell, P. (2005). Wikis in education. In J. Klobas Pegrum, M. (2005). Speed kills: Slowing down
et al., Wikis: Tools for information, work and col- online language teacher training. In B. Beaven
laboration (pp. 119-147). Oxford: Chandos. (Ed.), IATEFL 2005: Cardiff Conference selections
(pp. 156-158). Canterbury: IATEFL.
Molina, P. G. 2006 EDUCAUSE Evolving
Technologies Committee. (2006). Pioneering Pegrum, M. (2007). Socrates and Plato meet neo-
new territory and technologies. EDUCAUSE liberalism in the virtual agora: Online dialog and
Review, 41(5), 112-134. Retrieved January 14, the development of oppositional pedagogies. In J.
2008, from http://www.educause.edu/ir/library/ Lockard & M. Pegrum (Eds.), Brave new class-
pdf/erm0659.pdf rooms: Democratic education and the Internet
(pp. 13-34). New York: Peter Lang.
Motteram, G., & Ioannou-Georgiou, S. (2007).
Are teachers fit for Web 6.0? Paper presented at Pegrum, M. (2008a). Film, culture and iden-
New and Emerging Technologies in ELT, Loyola tity: Critical intercultural literacies for the
College, Chennai, India, 3-5 Aug. language classroom. Language and Inter-
cultural Communication, 8(2), 136–154.
National School Boards Association [USA].
doi:10.1080/14708470802271073
(2007). Creating and connecting: Research
and guidelines on online social—and educa- Pegrum, M. (forthcoming, 2008b). From hype
tional —networking. Alexandria, VA: National to hope: 10 myths of e-learning. In S. Ioannou-
School Boards Association. Retrieved Novem- Georgiou & P. Pavlou (Eds.), Learning technolo-
ber 29, 2007, from http://www.nsba.org/site/ gies in the language classroom: Research and
docs/41400/41340.pdf reflections. Canterbury: IATEFL.
Norton, B. (2000). Identity and language learn- Pegrum, M., & Bax, S. (2007). Catering to diver-
ing: Gender, ethnicity and educational change. sity through asynchronous online discussion: Link-
Harlow: Longman. ing teachers across continents. Paper presented at
Diversity: A Catalyst for Innovation — The 20th
O’Reilly, T. (2005). What is Web 2.0: Design pat-
English Australia Conference, Sydney, Australia,
terns and business models for the next generation
13-15 Sept.
of software. O’Reilly, 30 Sept. Retrieved January
14, 2008, from http://www.oreillynet.com/pub/a/ Pennycook, A. (2007). Global Englishes and
oreilly/tim/news/2005/09/30/what-is-web-20. transcultural flows. London: Routledge.
html?page=1
Pesce, M. (2007). Challenges and opportunities:
Panganiban, R. (2007). SLCC Day 1: ‘Second Peer-produced knowledge and Australian educa-
Life has the killer app, which is community’. The tion. Education.au. Retrieved November 29, 2007,
Click Heard Round the World, 25 Aug. Retrieved from http://blogs.educationau.edu.au/gputland/
November 29, 2007, from http://www.rikomatic. wp-content/uploads/2007/08/challenges-and-
com/blog/2007/08/slcc-day-1-seco.html opportunities.doc
Pavlenko, A., & Blackledge, A. (2004). Introduc- Phipps, A., & Gonzalez, M. (2004). Modern lan-
tion: New theoretical approaches to the study of guages: Learning and teaching in an intercultural
negotiation of identities in multilingual contexts. In field. London: Sage.
A. Pavlenko & A. Blackledge (Eds.), Negotiation
of identities in multilingual contexts (pp. 1-33).
Clevedon: Multilingual Matters.

1123
Communicative Networking and Linguistic Mashups on Web 2.0

Prensky, M. (2001). Digital natives, digital Sabre, G. (2007). SL education vs personal beliefs.
immigrants. On the Horizon, 9(5). Retrieved The Metaverse Journal, 10 Sept. Retrieved Janu-
November 29, 2007, from http://www.marcpren- ary 14, 2008, from http://www.metaversejournal.
sky.com/writing/Prensky%20-%20Digital%20 com/2007/09/10/sl-education-vs-personal-
Natives,%20Digital%20Immigrants%20-%20 beliefs/
Part1.pdf
Selber, S. A. (2004). Multiliteracies for a digital
Quinton, S. (2006). A brief critique on the future age. Carbondale, IL: Southern Illinois University
of learning: Assessing the potential for research. In Press.
D.L. Fisher & M.S. Khine (Eds.), Contemporary
Sharples, M., Taylor, J., & Vavoula, G. (2007).
approaches to research on learning environments:
A theory of learning for the mobile age. In R.
Worldviews (pp. 543-578). Singapore: World
Andrews & C. Haythornthwaite (Eds.), The Sage
Scientific Publishing.
handbook of E-Learning research (pp. 221-247).
Reddekopp, C. (2006). Multitasking and student London: Sage.
learning: Possible explanations for a growing
Stevens, V. (2007). Second Life and online collabo-
phenomenon. Occasional Papers in Educational
ration through peer-to-peer distributed learning
Technology. Saskatoon: University of Saskatch-
networks. Paper presented at METSMaC Confer-
ewan. Retrieved November 29, 2007, from http://
ence, Abu Dhabi, UAE, 17-19 Mar. Retrieved
www.usask.ca/education/coursework/802papers/
November 29, 2007, from http://www.homestead.
redekopp/index.htm
com/prosites-vstevens/files/efi/papers/metsmac/
Reeder, K., Macfadyen, L. P., Chase, M., & metsmac_secondlife.htm
Roche, J. (2004). Falling through the (cultural)
Stone, L. (2006). Attention: The real aphrodi-
gaps? Intercultural communication challenges
siac. Paper presented at ETech, San Diego, 6-9
in cyberspace. In F. Sudweeks & C. Ess (Eds.),
Mar. Retrieved November 29, 2007, from http://
Fourth International Conference on Cultural At-
radar.oreilly.com/archives/2006/03/etech_linda_
titudes towards Technology and Communication
stone_1.html
(pp.123-134). Perth: Murdoch University.
Street, B. V. (1994). Struggles over the meaning(s)
Rheingold, H. (2007). Vision of the future. Paper
of literacy. In M. Hamilton, D. Barton & R. Ivanič
presented to Education.au, 2 October. Retrieved
(Eds.), Worlds of literacy (pp.15-20). Clevedon:
November 29, 2007, from http://www.educa-
Multilingual Matters.
tionau.edu.au/jahia/Jahia/pid/521
Tapscott, D., & Williams, A. D. (2006). Wikinom-
Ricento, T. (2005). Considerations of identity in L2
ics: How mass collaboration changes everything.
learning. In E. Hinkel (Ed.), Handbook of research
New York: Portfolio.
in second language teaching and learning (pp.
895-910). Mahwah, NJ: Lawrence Erlbaum. The Consultants-E. (2007, Nov.). A potted his-
tory of technology in teaching – present. The
Ryan, Y. (2007). Do you YouTube? Wanna come to
Consultants-E Newsletter. Retrieved January 14,
MySpace? Musings on narcissism in the 21st cen-
2008, from http://www.theconsultants-e.com/
tury student. Paper presented at First Year in Higher
newsletter/The%20Consultants-E%20Newslet-
Education Conference, QUT, Brisbane, Australia.
ter%20-%20November%202007.pdf
Retrieved November, 29, 2007, from http://www.
fyhe.qut.edu.au/past_papers/papers07/final_pa-
pers/Do%20you%20YouTube%20_2_.pdf

1124
Communicative Networking and Linguistic Mashups on Web 2.0

The New London Group. (2000). A pedagogy Williamson, B., & Facer, K. (2004). More than
of multiliteracies: Designing social futures. In ‘just a game’: The implications for schools of
B. Cope & M. Kalantzis (Eds.), Multiliteracies: children’s computer games communities. Edu-
Literacy learning and the design of social futures cation Communication and Information, 4(2/3),
(pp. 9-37). London: Routledge. 253–268.
The New Media Consortium/The EDUCAUSE Wray, D. (Ed.). (2004). Literacy: Major themes
Learning Initiative. (2007). The horizon report: in education. Vol. 4: New literacies: The impact
2007 edition. Austin, TX: The New Media Consor- of technology. London: Routledge Falmer.
tium. Retrieved November 29, 2007, from http://
Zieghahn, L. (2001). ‘Talk’ about cul-
www.nmc.org/pdf/2007_Horizon_Report.pdf
ture online: The potential for transforma-
Thorne, S. L. (2006). Pedagogical and praxiologi- tion. Distance Education, 22(1), 144–150.
cal lessons from Internet-mediated intercultural doi:10.1080/0158791010220109
foreign language education research. In J.A. Belz
& S.L. Thorne (Eds.), Internet-mediated inter-
cultural foreign language education (pp.2-30).
KEY TERMS
Boston: Heinle & Heinle.
Unsworth, L. (2001). Teaching multiliteracies Codeswitching: This term refers to the use of
across the curriculum: Changing contexts of text more than one language or language variety in a
and image in classroom practice. Buckingham: given context, for example to aid communication
Open University Press. or to signal aspects of identity.
Continuous Partial Attention: According
Wales, J. (2007). Challenging how knowledge is to Linda Stone, citizens of Western societies
created. Education.au. Retrieved November 29, increasingly live in a state of continuous partial
2007, from http://www.educationau.edu.au/jahia/ attention, as they continuously monitor multiple
Jahia/home/challenging communication and information channels in an
Wallis, C. (2006). The multitasking generation. attempt not to miss anything. She argues that this
Time, 10(April), 46–53. is a post-multitasking behavior motivated less by
the need to save time or be efficient than by the
Warschauer, M. (1999). Electronic literacies: desire to always be connected to the network.
Language, culture and power in online education. Folksonomy: An index produced in a bottom-
Mahwah, NJ: Lawrence Erlbaum. up manner by adding user-generated tags to
Warschauer, M. (2003). Technology and social webpages of interest through a service such as
inclusion: Rethinking the digital divide. Cam- del.icio.us. The resulting list of tags is known as
bridge, MA: MIT Press. a folksonomy and may be displayed in the form
of a tag cloud, in which more prominent tags are
Warschauer, M. (2007). The paradoxical future of shown in larger and darker type.
digital learning. Learning Inquiry, 1(1), 41–49. Mashup: This term, which stems from the hip
doi:10.1007/s11519-007-0001-5 hop practice of mixing music and/or lyrics from
Wenger, E. (1998). Communities of practice: different songs to create new hybrids, can refer to
Learning, meaning, and identity. Cambridge: web applications which combine data from dif-
Cambridge University Press. ferent sources or, more commonly, to digital files

1125
Communicative Networking and Linguistic Mashups on Web 2.0

which mix together pre-existing video, graphics, Third Place: This term is used by Claire
music, text, etc, in new combinations. Kramsch to refer to the space between cultures
Social Constructivism: Social constructivism which language learners may reach as they develop
is a theory of learning which draws heavily on the intercultural (communicative) competence.
work of the Soviet psychologist Lev Vygotsky Web 1.0: A retrospective term which emerged
(1896-1934). It suggests that learners add to and after the advent of Web 2.0, Web 1.0 refers to
reshape their mental models of reality through the original, information-oriented version of the
social collaboration, building new understandings World Wide Web. Created by Tim Berners-Lee
as they actively engage in learning experiences. in 1989/1990, it consisted of largely static web-
Scaffolding, or guidance, is provided by teachers pages developed by a small number of authors
or more experienced peers in the learner’s zone of for consumption by a large audience.
proximal development, that is, the zone between
what a learner can achieve independently and
what s/he may achieve with support.

This work was previously published in Handbook of Research on Web 2.0 and Second Language Learning, edited by M. Thomas,
pp. 20-41, copyright 2009 by Information Science Reference (an imprint of IGI Global).

1126
1127

Chapter 4.14
Adoption of Web Services
in Digital Libraries:
An Exploratory Study
Fatih Oguz
Valdosta State University, USA

Abstract Introduction

This chapter describes a research study with an With the advent of the Internet and specifically the
objective to explore and describe decision factors World Wide Web (WWW) application, means of
related to technology adoption. The study utilized accessing data and information have changed for-
theories of diffusion of innovations and communities ever. The Internet brought great opportunities for
of practice as frameworks and a case study of Web libraries as well as dilemmas and problems, such
services (WS) technology in the digital library (DL) as technology choice and readiness.
environment to develop an understanding of the Digital libraries (DL) were envisioned as
decision-making process. A qualitative case study network-accessible repositories in the 1990s. Now,
approach was used to investigate the research prob- DLs extend the classical brick-and-mortar library
lems and data was collected through semistructured concept, bring value to society, and transform in-
interviews, documentary evidence (e.g., meeting formation landscape by improving and changing
minutes), and a comprehensive member check. the means of knowledge access, creation, use, and
Face-to-face and phone interviews were conducted discovery across disciplines, regardless of temporal
with respondents from five different DL programs and geographical barriers (Larsen & Watctlar, 2003;
in the U.S., selected based on distinctive character- Reddy & Wladawsky-Berger, 2001).
istics (e.g., size of the DL program). Findings of the The speed of technological advances in infor-
research suggest that the decision-making process is mation technologies (IT) in the last 10 years has
a complex procedure in which a number of factors enabled DLs to provide innovative resources and
are considered when making WS adoption decisions. services to people. The information landscape is
These factors are categorized as organizational, changing as a result of the revolutionary develop-
individual, and technology-specific factors. ments in IT, incompleteness of content on Internet,
ever increasing digital content along with the
evolution of networked technologies and applica-
DOI: 10.4018/978-1-60566-042-4.ch014 tions, lack of standards, ineffective information

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Adoption of Web Services

retrieval mechanisms, and minimal cataloging. information throughout and beyond the enterprise
These factors present challenges to the future of (Adams, Gisolfi, Snell, & Varadan, 2002). Besides
DL development efforts (Borgman, 1999; Reddy recognizing heterogeneity of networked resources
& Wladawsky-Berger, 2001). and applications as a fundamental ingredient, WS
The concept of Web services (WS) has emerged are independent of platform and the development
as the next generation of Web-based technology environment can be packaged and published on
for exchanging information. This effort began the Internet. Also, WS enable just-in-time integra-
with the submission of the SOAP 1.1 to the World tion and interoperability with legacy applications
Wide Web Consortium (W3C) (Barefoot, 2002). (Oguz & Moen, 2006).
WS are self-contained applications that can be The development and widespread deployment
described, published, invoked, and located over of more intelligent knowledge environments that
the Internet (or any network). Once a Web service not only support scholarly inquiry and commu-
is deployed, other applications can discover and nication but also that are open, accessible to all,
invoke the service. WS provide a programmable and transparent in their operation remains as a
interface for other applications without requiring fundamental challenge for DL practitioners and
custom programming and proprietary solutions researchers.
regardless of the operating systems and program- DL applications need to have some room to
ming languages to share information as opposed accommodate future technological innovations
to providing users with a graphical user interface regardless how they are built, using off-the-shelf
(Boss, 2004). software vs. custom-built, and thus decision
According to the W3C, a Web service is defined makers who include managers, coordinators, de-
as a software system designed to support interop- signers, and developers need to make important
erable machine-to-machine interaction over a decisions at some point in time to adopt or reject
network by using XML for sending and receiving an innovation, including a specific technology,
messages (Booth, Haas, McCabe, Newcomer, application, framework or idea related with DLs.
Champion, Ferris, et al., 2004). Simplicity and Decision makers who need information about an
flexibility of XML made it a definitive standard innovation may seek this information through
for data transmission and storage. XML is an open both informal and formal communication channels
standard and can be accessed and processed by while making such critical decisions.
any tool capable of reading and writing American In the context of DLs, roles and influence of
standard code for information interchange (ASCII) informal communication channels on the decision-
text. By definition, the only requirement for a Web making process to adopt or reject WS technology
service is to use XML. has not been investigated before. The adoption of
The basic WS platform is composed of XML a new technology, WS, which is its early stages
and a transport protocol. HTTP is the commonly of adoption in the DL environment, may provide
used transport protocol on the Internet (Hickey, a significant opportunity to investigate decision
2003). XML, simple object access protocol factors. The goal of this study is to shed a light on
(SOAP), and Web services description language the decision-making process to adopt or reject a
(WSDL) are tools to create WS. A Web service new technology, WS, in the context of DLs.
provides the framework for creating the next As technologies rapidly change and the in-
generation of distributed systems by which or- formation landscape is transformed, DLs find
ganizations can encapsulate existing business themselves dealing with the issues of technology
processes, publish them as services, search for adoption decisions to exploit this dynamically
and subscribe to other services, and exchange changing technology environment to meet their

1128
Adoption of Web Services

users’ needs and expectations. Therefore, under- factors (e.g., formalization and centralization)
standing the decision-making process regarding as well as other innovation characteristics (e.g.,
adoption of WS technologies in the context of cost, profitability, social approval) influencing
DLs is important. adoption of an innovation, and therefore Rogers’
DOI theory needs to be extended to accommodate
such factors, specifically in organizational set-
Background tings (Daft, 1978; Damanpour, 1991). In addition,
Tornatzky and Klein (1982) found that relative
This study used the diffusion of innovations (DOI) advantage, compatibility, and complexity have the
and communities of practice (CoPs) as theoretical most consistent relationships with the adoption of
frameworks and a case study of WS technologies innovations across a wide range of industries.
in the DL environment to develop an understand- Rogers (1995, p. 23) defines a social system
ing of the decision-making process. as “a set of interrelated units that are engaged in
joint problem-solving to accomplish a common
Diffusion of Innovations and goal” and the members or units of a social system
Communities of Practice as may be composed of individuals, organizations,
Theoretical Frameworks and informal groups. Patterned social relationships
(e.g., hierarchical positions) among the members
The DOI research methodology provides required of a social system define the social structure of
instruments, both quantitative and qualitative, to a system which, in return, can facilitate or de-
assess the rate and pattern of diffusion of an innova- lay the diffusion of an innovation and lays out
tion and identifies various factors that facilitate or a framework for making predictions about the
hinder its adoption and implementation (Fichman, human behavior in a system since such structure
1992). These major factors include properties of provides regularity and stability to human behavior
the innovation, characteristics of adopters, and (Rogers, 1995).
the means leading to adoption. Established behavior patterns called norms are
An innovation can be an idea, behavior, prac- the ruling principles of a social system, which may
tice, or object perceived as new by the adopter (e.g., also influence diffusion (Rogers, 1995). In other
organization, individual). The concept of newness words, norms serve as a guide or a standard for
may be determined by the human reaction to it the members against which they can assess their
as well as the time passed since its discovery or own behavior. Norms may slow the diffusion
first use. If the idea seems new to an individual process when an innovation does not comply with
or organization, it is considered an innovation the norms of a social system even if the adoption
(Daft, 1978; Rogers, 1995). of an innovation offers important benefits for the
DOI researchers study the characteristics of the system (Raghavan & Chand, 1989).
innovation to explain the rate of adoption of an in- The innovation-decision can be made by an
novation. Rogers (1995) classifies characteristics individual member of a system as well as by
of innovations into five general categories: relative the entire system. The decision can be made
advantage, compatibility, triability, observability, collectively by reaching a consensus among the
and complexity. Innovations with greater relative members of a social system or by a relatively few
advantage, compatibility, triability, observability, individuals who possess status, power, or technical
and less complexity are more likely to be adopted expertise (Rogers, 1995). A decision made by an
faster than others that lack these characteristics individual to adopt or reject a new idea indepen-
(Rogers, 1995). However, there are structural dently from other members of a system is called

1129
Adoption of Web Services

an optional-innovation decision (Rogers, 1995). acting regularly (Wenger, McDermott, & Snyder,
An adoption decision may be influenced by the 2002). CoPs provide a learning environment
norms of the system and informal communication through social participation, where participation
channels. In this case, the decision is made by an refers to being active participants in the practice
individual member of the system rather than the and building a sense of identity associated with
entire social system, and the individual member the CoP to which they belong.
is fully responsible for the consequences of the CoPs embody individuals with diverse back-
decision. Collective-innovation decisions are grounds and social structures (e.g., other CoPs,
made by members of a system through a consensus organizations), which in turn, reduce the learn-
to adopt or reject a new idea. All the units within ing curve and rework, and promote innovation
the social system are expected to comply with the by enabling them to share and disseminate both
decision. However, reaching a collective deci- tacit and explicit knowledge (Lesser & Storck,
sion may be a time-consuming process because 2001). Sharing tacit knowledge requires personal
it is made by a consensus among the members. interaction and CoPs provide such an informal
Authority-innovation decisions are made by a learning platform through conversation and ap-
select set of members of a social system who prenticeship, for example. Members of the com-
have authority and higher status in the organiza- munity become aware of their peers’ expertise,
tional chart; in this decision-making process an knowledge, and skills through creating a venue
individual member has little or no influence on for them to interact with each other. Thus, they
the decision. In organizational settings, collective are able to compare, verify, and benchmark their
and authority-innovation decisions are more com- professionally developed expertise in the field
mon than the optional-innovation decisions, and against their colleagues’ knowledge. When these
authority-innovation decisions result in higher rate benefits of CoPs are considered, their contribu-
of adoption than others (Rogers, 1995). tion to DL development efforts is vital in making
Diffusion of an innovation is a social process informed technology, specifically WS, adoption
that is influenced by various factors such as decisions. The literature (e.g., Borgman, 1999;
characteristics of the innovation (e.g., relative Marchionini, 1998) and nature of DL development
advantage) and the decision-making unit (e.g., efforts (e.g., open source) suggest the existence
individual characteristics), depending on the level of informal structures such as CoPs.
of adoption (individual vs. organizational). The As the organizations, specifically commercial
information about the innovation is communicated ones, expand in size, geographical coverage, and
through formal (e.g., mass media) and informal complexity, knowledge has become the key to
in the course of the innovation-decision process. improving organizational performance and the
Rogers (1995) suggests that having some expo- formation of informal social groups like CoPs
sure to mass media and informal communication become a natural part of organizational life
channels such as interpersonal networks increases (Lesser & Storck, 2001; Wenger et al., 2002).
a potential adopter’s chance of knowing about CoPs make knowledge an integral part of their
an innovation earlier than others. This chapter ongoing activities and interactions. Interpersonal
specifically focuses on CoPs which serve as an interactions play an important role, especially in
informal communication channel. sharing tacit knowledge; the learning tools uti-
CoPs are composed of people who share a lized by CoPs such as storytelling, conversation,
concern, common problems, or a passion about the and apprenticeship increase the efficient use of
domain, and who want to gain more knowledge knowledge. CoPs act as a “living repository” for
and expertise pertaining to the domain by inter- collective knowledge through creating a value

1130
Adoption of Web Services

for both the members and the organizations sup- at the local level. However, standards develop-
porting and sponsoring these social structures ment is a complex social process and requires
(Wenger et al., 2002). consensus among stakeholders (Moen, 1997).
The DL conferences, funding agencies, work- The second level encourages individual DLs to
shops, and professional societies (e.g., Association create standards-based services that can be easily
for Computing Machinery) play important roles accessible and used by other DLs.
both in building and cultivating the CoPs in the DL A vision set forth for the DLs by the President’s
field, and such meetings serve as a breeding ground Information Technology Advisory Committee (PI-
for future collaboration in DL development efforts TAC) Panel on Digital Libraries is that of providing
(Borgman, 1999). In addition, the experts in the the means of searching and accessing all human
field reached a consensus that “efforts associated knowledge anytime and anywhere via Internet
with development of digital libraries are primar- for all citizens (Reddy & Wladawsky-Berger,
ily collaborative” in a Delphi study conducted by 2001). One of the key issues in accomplishing
Kochtanek and Hein (1999, p. 253). this vision is improving the ability to store and
retrieve digital content across disparate and in-
Web Services in Digital Libraries dependent systems and collections by improving
interoperability among diverse DL implementa-
In general, DLs enable far broader range of users tions (Reddy & Wladawsky-Berger, 2001). Thus,
than traditional physical and organizational ar- interoperability is an important factor to consider
rangements (e.g., libraries) to access information. in the DL environment when making decisions to
Gathering, organizing, sharing, and maintaining adopt WS technologies.
such information resources require a flexible, scal- Important decisions have been made in the past
able, and interoperable infrastructure (Larsen & as to adopt or reject a new technology for vari-
Watctlar, 2003). Interoperability is an important is- ous reasons including the pursuit of this vision,
sue where various system architectures, operating delivering content in more efficient and advanced
systems, and programming languages are required manner, and social status (e.g., being a pioneer
to communicate with each other. In addition, DL in offering new DL services) (Pasquinelli, 2002).
development efforts are closely related with the Some of the key technologies and standards related
progress in general purpose technologies such as with interoperability that have been adopted in
high-speed networking, security, and interoper- the past in DL environments such as the ANSI/
ability (Marchionini, 1998). However, the size, NISO Z39.50 protocol, open archives initiative
heterogeneity, and complexity of the today’s in- protocol for metadata harvesting (OAI-PMH),
formation resources become critical factors when and open URL.
building DL systems because such factors may Hickey (2003) lists various ways of using
create immense challenges for interoperability, WS technology in DLs from registering different
or the ability to ensure seamless information types of objects and search services to navigating
exchange across multiple DLs and information hierarchies and decomposing objects into simpler
resources (Akscyn & Witten, 1998; Gonçalves et objects. The search/retrieve Web service (SRW)
al., 2002; Marchionini, 1998). Marchionini (1998) is a standardized Web service built on the 20
addresses interoperability in two levels. The first years of experience of the Z39.50 information
level is the efforts to create standards for data retrieval protocol. SRW provides an easy way to
storage and transmission, for query representa- implement the protocol with the power of older
tion, and for vocabulary control; DLs adopt such and more complex Z39.50 (Sanderson, 2004).
standards and modify their content and services Even some libraries are replacing Z39.50 with WS

1131
Adoption of Web Services

technologies as the protocol of choice between Main Thrust of the Chapter


library portals and online electronic resources
(Boss, 2004). WS facilitate access to electronic This chapter attempts to explore and describe fac-
databases, and digital libraries providing access tors, activities, processes, and forces involved in
to such resources benefit from this technology the decision-making process related to adoption
(Boss, 2004). of WS technologies in DLs.
The flexible and extensible digital object
and repository architecture (Fedora) system, Research Problems
designed by the Cornell University Information and Methodology
Science and The University of Virginia Library’s
Digital Library Research and Development Group The research strategy consisted of two compo-
(DLR&D), is a promising open source digital nents: a qualitative methodology and a case study.
library software initiative. Fedora was originally This strategy provided a framework of methods
implemented based on CORBA architecture; and data that would yield answers to the two re-
however, the next release (Fedora 2.0) has ad- search questions: (1) What are the key decision
opted a service-oriented approach (SOA) based factors that lead decision makers to adopt or reject
on WS (“Tutorial 1: Introduction,” 2005). DSpace WS in the DL environment? and (2) What are the
is another open source system, developed by activities and entities that influence the decision
Hewlett-Packard and MIT Libraries, to store the regarding adoption of WS technologies in the DL
digital research and educational material produced environment?
by an organization or institution as a repository. The exploratory and descriptive nature of the
Greenstone is yet another open source digital li- study justified the use of a qualitative research
brary software from New Zealand Digital Library approach that allows discovery and description
Project at the University of Waikato that has a of the social processes involved in decision mak-
focus on publishing (Don, Buchanan, & Witten, ing. Although quantitative methods have been
2005). The DELOS network pays close attention predominant in information technology (IT) adop-
and contributes to the use of WS technologies in tion research (Choudrie & Dwivedi, 2005), this
digital libraries. EBSCO publishing, a provider chapter aims to develop a better understanding
of a broad range of full-text and bibliographic of decision factors influencing adoption of WS
databases, has introduced its WS interface to technologies in the context of DLs.
EBSCOhost, an electronic journal service for In-depth information about this complex social
academic and corporate subscribers, forming a process involving decision makers was acquired
basis of real-time communications among library through semistructured interviews and documen-
systems, portals, and all other systems in the future tary evidence (e.g., meeting minutes and reports).
(Boss, 2004). The interview respondents and academic libraries
The major strength of WS is its reliance on that they are associated with were selected based
XML. Given the characteristics of WS technolo- on characteristics of DL programs identified by
gies and current use in DLs and e-commerce, WS Greenstein and Thorin (2002). These character-
are poised to play an important role as a technol- istics included age of the program, staff size, and
ogy providing interoperable standards-based organization and orientation of the program. Seven
access to DLs. respondents with different responsibilities (ad-
ministrative vs. technical) were interviewed from
five different DL programs in the US. These DL

1132
Adoption of Web Services

programs included big (i.e., staff size) programs carefully carried out to include information-rich
such as California Digital Library and University respondents.
of Texas at Austin and relatively smaller ones such Documentary evidence provided additional
as University of North Texas and University of and clarifying information supplemental to the
Texas at Dallas. data collected through interviews. Further, a
Following Patton’s (2002) guidelines, purpose- comprehensive member check was conducted
ful sampling, specifically maximum variation which allowed to obtain additional information
sampling, was employed when selecting the re- from respondents and to have study findings re-
spondents who had the best knowledge, expertise, viewed by them. This final verification process
and overview about the topic of the research. The allowed respondents to evaluate the researcher’s
maximum variation sampling aimed at “capturing interpretation of findings and analysis of data
and describing the central themes that cut cross from their perspectives (Lincoln & Guba, 1985;
great deal of variation” (Patton, 2002, p. 234). Patton, 2002).
The respondents were from DL programs at
the California Digital Library, University of North Results and Findings
Texas, University of Texas at Dallas, University of
Texas at Austin, and a university in the American Data revealed a number of factors that influenced
Southeast. Some of the participating libraries are and informed the decision-making process in WS
members of various influential professional soci- adoption. These factors are categorized at organi-
eties and organizations in the DL field, including zational, individual, and technical levels.
Digital Library Federation (DLF), Association Characteristics of DL programs that appeared
of Research Libraries (ARL), and Coalition for to influence the decision-making process and cat-
Networked Information (CNI). Seven interviews egorized as organizational level factors included:
were conducted with administrators and technical organizational culture, program’s relationships
personnel who were involved in the decision- with surrounding academic units and external part-
making process at these five academic libraries’ ners, management style and work structure, focus
DL programs. and direction of a program, formalization (e.g.,
Patton (2002) sets no rules for the sample size flexibility in hierarchal order), functional differ-
in qualitative inquiry by arguing that “the valid- entiation in a program, size and age of a program,
ity, meaningfulness, and the insights generated administrative attitude toward change, financial
from qualitative inquiry have more to do with resources, technology readiness (e.g., expertise,
the information richness of the cases selected and technology infrastructure), and program’s expecta-
the observational/analytical capabilities of the tions (e.g., user needs). These organizational level
researcher than with sample size” (p. 245). The factors appeared to play a critical role, especially
researcher stopped conducting interviews when in influencing members’ information-seeking and
data saturation was reached to meet the research communication behaviors. Individual level fac-
goal, that is, to understand and describe decision tors included members’ connectedness with their
factors related to WS adoption. Data saturation colleagues, skill-set (e.g., competence), participa-
is defined as the point in a data collection pro- tion in CoPs, perception of organizational culture
cess where new information becomes redundant and goals, and openness to new ideas. Technical
(Bogdan & Biklen, 1992). Romney, Batchelder, level factors included: interoperability, modular-
and Weller (1986) conclude that samples as small ity, scalability, flexibility, addressability, rapid
as four participants could be enough to meet re- deployment of services, subscription service, and
search objectives where purposeful sampling is open-standards base of WS.

1133
Adoption of Web Services

Some of the organizational level factors in- cal background and skills were closely associated
clude management style, focus and direction of with perception of these characteristics.
the program, size and age of the program, and Technology readiness of the DL program was
organizational culture. Organizational level fac- another organizational factor that appeared to
tors were closely associated with the organization have an influence in the decision-making process.
itself and indirectly impacted by a DL program’s Technology readiness had two aspects: a human
staff, for example. Administrative personnel had aspect (e.g., expertise, staffing), and the techno-
an influence on some of organizational level fac- logical compatibility of WS technologies with
tors (e.g., management style). Other factors could existing technical infrastructure (i.e., hardware,
be regarded as more individual characteristics of software, and standards). In addition, technology
DL staff members in terms of their information- readiness was also closely associated with avail-
seeking and communication behavior. Impact of ability of financial resources in case a hardware or
individual level factors on decision-making vary software upgrade was needed. Respondents noted
from one technology to another depending on that compatibility of WS technologies with their
role of the technology (i.e., mission critical vs. existing technological infrastructure was an im-
non-mission critical) in the DL program. portant factor that informed the decision-making.
Respondents identified financial concerns as Tornatzky and Klein (1982) found compatibility
a critical factor in guiding technology adoption as one of the most addressed innovation attributes.
decisions, and these concerns included: initial Compatibility also refers to consistency of an in-
cost, ongoing cost, payoff, budgetary restrictions, novation with existing values and norms of the
and funding requirements. However, the extent of DL program. Furthermore, technology readiness
influence of these factors on WS adoption deci- was also an important factor for triability pur-
sions appeared to vary from one DL program to poses. Small scale experiments were generally
another depending on the DL program’s expecta- conducted in DL programs prior to making an
tions from the technology and needs, focus, and adoption decision.
direction of the program. These expectations and Individual level adoption decisions could be
needs were closely related with size and age of made especially for the use of WS technologies
the program. Data suggested that as programs in nonmission critical applications. A personal
grew in size over time, so did their collections, positive experience with WS, existing skill-
responsibilities, and user expectations. For ex- set, potential benefits for the work (i.e., Davis’
ample, although respondents formed positive [1985] perceived usefulness), and having easy
opinions regarding open source software, they access to experienced-based knowledge through
were aware that the lack of necessary skills in CoPs appeared to influence an individual’s per-
the program would be an important factor when ception and lowered the individual’s learning
getting a project initiated and providing technical curve. Technology and specific factors such as
support if they chose to use open source software. interoperability, modularity, flexibility, and WS
In addition, acquiring necessary technology skills subscription service were also decision factors in
through hiring new staff members and additional this case. In DL programs where WS had already
training were also factors impacting cost. Lack of been adopted, adoption decisions were made
technical expertise as a decision factor appeared collectively and WS technologies were used for
to reflect the importance of Davis’ (1989) “ease major and mission-critical applications. Both
of use,” Tornatzky and Klein’s (1982) “ease of organizational and technology-specific factors
operation,” and Rogers’ (1995) “complexity” as were taken into consideration.
innovation characteristics since adopters’ techni- Another important decision factor was tech-

1134
Adoption of Web Services

nology-specific benefits offered by WS, including appeared that they sometimes wanted to be the
interoperability, modularity, and open standards. first or early adopters of some technologies in
WS provides an interoperable platform and is built the DL environment to set an example for other
on open standards (e.g., XML) where programs programs. Though setting an example for other
written in different programming languages DL programs or wanting to be an early adopter
and running on different operating systems are was not a key decision factor, it was one of the
able to communicate with each other based on factors occasionally taken into consideration in
open standards and protocols. Data suggested the decision-making processes.
that interoperability was an important factor Data suggested that availability of financial
since WS would not require major changes in resources, focus and direction, size and age, col-
existing technical infrastructure. In addition to lection size, users’ and programs’ expectations,
interoperability, respondents identified additional and technology readiness were important factors
technology-specific factors. For example: influencing decision makers. Especially when
making an optional-innovation decision, a po-
• Modularity and flexibility: Ability to act tential adopter’s existing technical skill-set and
as building blocks to create distributed ap- connectedness with the adopter’s colleagues were
plications through reuse of existing appli- key factors. At the technical level, interoperability,
cations that can be published and accessed modularity, scalability, flexibility, addressability,
over the internet or intranets. rapid deployment of services, subscription ser-
• Rapid deployment of Services: vices, and the open standards base of WS were
Development time for new applications or key decision factors leading decision makers to
services is greatly reduced through use of adopt or reject WS in the DL environment.
standard interfaces (e.g., WSDL) and open Figure 1 presents a conceptual framework for
standards. the study informed by the theoretical frameworks
• Scalability: Ability to handle a growing in light of results and findings discussed earlier.
amount of usage loads (e.g., Web caching, Organizational and technical (i.e., technology
load balancing). specific) level factors have an influence both on
• WS subscription service (UDDI): A reg- adopters’ information seeking behaviors and on the
istry services for WS and allows other WS adoption decision itself depending on the type of
applications to automatically discover ser- innovation decision (i.e., optional, collective, and
vices and use them. authority) made. Individual level factors (i.e., indi-
vidual characteristics) guide adopters’ information
Small DL programs often looked to older and seeking activities (e.g., participation in CoPs) and
bigger DL programs when it came to adopting influence their perception of organizational values
new technologies and standards. Their limited which in turn inform adopters’ contribution to the
financial resources, staff size, and skill set were decision-making process. As shown in Figure 1,
important barriers preventing them from taking potential adopters may acquire information about
initiatives that might be considered risky. These an innovation through formal (e.g., mass media)
initiatives included technologies or standards and informal (e.g., interpersonal) communication
that have not been tried or are in early stages of channels. The information acquired through these
adoption in other DL programs. They often chose channels includes perceived characteristics of an
to rely on experiences of other DL programs so innovation that may play a key role as decision
that they would be less likely to fail and run into factors in the decision-making process. Data sug-
unexpected problems. As for bigger programs, it gested that use of information collected through

1135
Adoption of Web Services

Figure 1. Conceptual framework

these channels in the decision-making process vary DL programs benefited from such connections
due to adopters’ degree of participation in CoPs, not only by accessing their expertise but also by
characteristics of knowledge (explicit vs. tacit) acquiring their content and collections.
acquired, and factors at organizational, technical, DL staff members’ interactions with others
and individual levels. appeared to be maintained informally and the
On the other hand, there were number of activi- organizational structure of DL programs encour-
ties that members of DL programs participated in, aged informal communication. Informality in
entities that provided them with guidance, pro- communicating with others is one of the key
cesses that helped them develop an understanding characteristics of CoPs. Respondents interacted
of WS, motivations that encouraged or discouraged with their colleagues who were part of their own
them towards WS technologies, and forces that DL programs, as well as people from libraries,
informed and guided their information seeking other university units, or external institutions
and communication behaviors. These activities, and organizations to advance and share their
entities, processes, and forces were in play when knowledge and contribute to the field. Informal
making a decision regarding adoption of WS communication was also cited as an important part
technologies in the DL environment. of the technology assessment process.
DL programs had good ties with surrounding Some of the participating academic libraries
academic departments and information services, were members of various professional associations
including other library departments, IT depart- and organizations which have an influence on DL-
ment, library and information school, and faculty. related issues, including use of DL technologies,

1136
Adoption of Web Services

digital preservation, standards, and DL develop- discussions enhanced and improved members’ un-
ment activities. Participation in these organizations derstanding of the technology. In other words, this
(e.g., DLF, CNI, ARL) provided venues for DL mediating process gave rise to cross-fertilization
programs to share their work and connect with of ideas and appeared to improve credibility of
other DL programs. The DLF promotes work the knowledge generated and housed in CoPs.
on DL structures, standards, preservation, and CoPs provided a living repository for the knowl-
use, CNI is interested in various areas critical to edge generated within the community while they
present and future of DLs, and ARL is one of the were also perceived as places where up-to-date
sponsor organizations of CNI and its member and quality information could be acquired. CoPs
institutions are very active in the field. Further, were also used to verify information acquired
these organizations engage in collaborative activi- from different sources.
ties with each other in pursuit of their missions The conceptual framework (see Figure 1)
and goals. Collegial activities that were made helped to structure the data-found analysis in a
possible through these relationships with these format which may help the reader see this very
external entities appeared to play a central role in complex landscape and understand this complex
formation and continuation of informal commu- social process. Further, this chapter provided
nities that can be characterized as CoPs. Entities evidence that Rogers’ DOI model needs to be
in the DL environment generally included CoPs, complemented with organizational level factors
surrounding academic units and external partners, identified by other researchers such as Daft (1978),
funding agencies, and the program itself. Davis (1989), and Tornatzky and Klein (1992) to
In addition, attending national and international understand and describe diffusion of innovations
conferences were the most commonly used venues in organizational settings.
to obtain new information and served as a breeding
ground for building personal contacts with col-
leagues. These collegial activities were regarded Conclusion
as communal activities. Preexisting personal con-
tacts and the connections established in various This research was an exploratory and descrip-
venues with other institutions, organizations, and tive study to shed a light on the decision-making
DL initiatives appeared to be very important for process to adopt or reject a new technology, WS,
information access and sharing purposes. in the context of DLs and the unit of analysis was
CoPs attracted individuals with diverse back- the decision to adopt or reject a new technology.
grounds and skills from all around the world The information landscape is transformed as
regardless of their geographical locations and technologies rapidly change and DLs often find
provided an informal learning platform for their themselves in a critical position to make a decision
members. These CoPs were generally built and whether to adopt or reject emerging technologies
maintained in an online environment and occa- such as WS.
sionally supported with face-to-face interactions. Since the study employed a qualitative case
In addition to distributed virtual CoPs, there were study approach that supported the exploratory
other CoPs, which may be subgroups of a broader and descriptive nature of the research, results
CoP, that were locally networked and physically and findings of the study are not intended to be
located. Participation of members in discussions in statistically generalizable to other technology
CoPs enabled online communities to cultivate and adoption cases. However, detailed description
nurture knowledge acquired thorough experience, in the narrative may assist the reader of this case
print, or other online resources and, in turn, these study research to determine applicability of these

1137
Adoption of Web Services

findings to other technology adoption decisions Acknowledgment


in the reader’s own setting.
This chapter provides evidence that CoPs as The author thanks Corrie Marsh for her valuable
informal communication channels practice play feedback in writing of this chapter and to Dr.
an important role in enabling staff members of William Moen for his guidance and support for
a DL program to access up-to-date and experi- this research.
enced-based knowledge, providing a distributed
problem-solving and learning platform, facilitat-
ing informal communication and collaborative References
activities among DL programs, and informing
the decision-making process. Technical charac- Adams, H., Gisolfi, D., Snell, J., & Varadan, R.
teristics (e.g., interoperability, open standards), (2002). Best practices for Web services: Part
compatibility with existing technical infrastruc- 1 - formation of a semantic framework. IBM
ture, applicability to existing DL projects, total developerWorks. Retrieved March 5, 2004, from
cost of ownership (e.g., licensing, maintenance http://www-106.ibm.com/developerworks/Web-
cost), technical expertise in the DL program (e.g., services/library/ws-best1/
staffing, training, learning curve), and success of Akscyn, R. M., & Witten, I. H. (1998). Report
a pilot project are cited as key decision factors of First Summit on international cooperation on
influencing adoption Web services technologies digital libraries. Retrieved May 3, 2005, from
in the DL environment. This chapter provides http://Web.archive.org/Web/20010529104724/
an adequate foundation for further research on ks.com/idla-wp-oct98/
the impact of organizational, individual, and
technology-specific factors on decision-making Barefoot, D. (2002). Web services primer. Cape-
processes in the DL environment. Science.
The complexity of the decision-making process
Bogdan, R. C., & Biklen, S. R. (1992). Qualita-
and the variety of factors that informed and influ-
tive research for education: An introduction to
enced this process are reflected. A review of the
theory and methods (2nd ed.). Boston: Allyn
relevant literature suggests that this is a complex
and Bacon.
process, and the findings inform and provide de-
tails about the complexity, as presented in Figure Booth, D., Haas, H., McCabe, F., Newcomer, E.,
1. The theoretical frameworks selected for this Champion, M., Ferris, C., et al. (2004, February).
chapter proved useful to achieve the goal of the Web services architecture (W3C Working Group
study. The chapter attempts to provide a complete Note 11). The World Wide Web Consortium
account of decision factors related with adoption (W3C).
of WS technologies in the DL environment. An
Borgman, C. L. (1999). What are digital librar-
outcome of this study suggests that an exploratory
ies? Competing visions. Information Processing
and descriptive study such as this is an important
& Management, 35(3), 227–243. doi:10.1016/
step towards understanding the decision-making
S0306-4573(98)00059-4
process as technologies rapidly change in the DL
environment. Boss, R. W. (2004). Web services. Retrieved No-
vember 1, 2005, from http://www.ala.org/ala/pla/
plapubs/technotes/Webservices.htm

1138
Adoption of Web Services

Choudrie, J., & Dwivedi, Y. K. (2005). Inves- Kochtanek, T. R., & Hein, K. K. (1999). Delphi
tigating the research approaches for examining study of digital libraries. Information Processing
technology adoption issues. Journal of Research & Management, 35(3), 245–254. doi:10.1016/
Practice, 1(1). S0306-4573(98)00060-0
Daft, R. L. (1978). A dual-core model of orga- Larsen, R. L., & Watctlar, H. D. (2003). Knowledge
nizational innovation. Academy of Management lost in information. Report of the NSF Workshop
Journal, 21(2), 193. doi:10.2307/255754 on Research Directions for Digital Libraries.
Damanpour, F. (1991). Organizational inertia Lesser, E. L., & Storck, J. (2001). Communities
and momentum: A dynamic model of strategic of practice and organizational performance. IBM
change. Academy of Management Journal, 34(3), Systems Journal, 40(4), 831–931.
555–591. doi:10.2307/256406
Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic
Davis, F. D. (1989). Perceived usefulness, per- inquiry. Beverly Hills, CA: Sage Publications,
ceived ease of use, and user acceptance of infor- Inc.
mation technology. MIS Quarterly, 13(3), 318.
Marchionini, G. (1998). Research and develop-
doi:10.2307/249008
ment in digital libraries. In A. Kent (Ed.), Ency-
Don, K., Buchanan, G., & Witten, I. H. (2005). clopedia of library and information science (Vol.
Greenstone 3: A modular digital library. Retrieved 63, pp. 259-279). New York: Marcel Dekker.
November 1, 2005, from http://www.sadl.uleth.
Moen, W. E. (1997). The Development of ANSI/
ca/greenstone3/manual.pdf
NISO Z39.50: A Case Study in Standards Evolu-
Fichman, R. G. (1992, December). Informa- tion. Unpublished Dissertation, Syracuse Uni-
tion technology diffusion: A review of empirical versity.
research. Paper presented at the Thirteenth In-
Oguz, F., & Moen, W. E. (2006). Texas library
ternational Conference on Information Systems
directory Web services application: The potential
(ICIS), Dallas.
for Web services to enhance information access to
Gonçalves, M. A., Fox, E. A., Watsom, L. T., legacy data. Paper presented at the International
& Kipp, N. A. (2002). Streams, structures, Conference on Next Generation Web Services
spaces, scenarios, societies (5s): A formal model Practices (NWeSP’06), Korea.
for digital libraries. [TOIS]. ACM Transac-
Pasquinelli, A. (2002). Digital library technology
tions on Information Systems, 22(2), 270–312.
trends. Retrieved December, 9, 2005, from http://
doi:10.1145/984321.984325
www.sun.com/products-n-solutions/edu/whitepa-
Greenstein, D., & Thorin, S. E. (2002). The pers/pdf/digital_library_trends.pdf
digital library: A biography. Digital Library
Patton, M. Q. (2002). Qualitative research and
Federation.
evaluation methods (3rd ed.). Thousand Oaks,
Hickey, T. B. (2003). Web services for digital CA: Sage Publications, Inc.
libraries. Paper presented at the Cross Language
Raghavan, S. A., & Chand, D. R. (1989). Diffusing
Applications and The Web 27th Library Systems
Software-Engineering Methods. IEEE Software,
Seminar, Bern, Switzerland.
6(4), 81–90. doi:10.1109/52.31655

1139
Adoption of Web Services

Reddy, R., & Wladawsky-Berger, I. (2001). Digital Tornatzky, L. G., & Klein, K. J. (1982). Innova-
libraries: Universal access to human knowledge: tion characteristics and innovation adoption-
A report to the President. President’s Information implementation: A meta-analysis of findings.
Technology Advisory Committee (PITAC), Panel IEEE Transactions on Engineering Management,
on Digital Libraries. 29(1), 28–45.
Rogers, E. M. (1995). Diffusion of Innovations Tutorial 1: Introduction. (2005). Retrieved August
(4th ed.). New York: The Free Press. 15, 2005, from http://www.fedora.info/down-
load/2.0/userdocs/tutorials/tutorial1.pdf
Romney, A., Batchelder, W., & Weller, S. (1986).
Culture as consensus: A theory of culture and in- Wenger, E., McDermott, R., & Snyder, W. M.
formant accuracy. American Anthropologist, 88, (2002). A guide to managing knowledge: Culti-
313–338. doi:10.1525/aa.1986.88.2.02a00020 vating communities of practice. Boston: Harvard
Business School Press.
Sanderson, R. (2004). SRW: Search/retrieve Web-
service version 1.1. Retrieved February 2, 2005,
from http://srw.cheshire3.org/SRW-1.1.pdf

This work was previously published in Managing Web Service Quality: Measuring Outcomes and Effectiveness, edited by K.
Khan, pp. 307-320, copyright 2009 by Information Science Reference (an imprint of IGI Global).

1140
1141

Chapter 4.15
A Context-Based
Approach to Web 2.0 and
Language Education
Gary Motteram
University of Manchester, UK

Susan Brown
University of Manchester, UK

Abstract INTRODUCTION

Web 2.0 offers potentially powerful tools for the This chapter explores the way that participants
field of language education. As language teacher on a module run as part of an MA in Educational
tutors exploring Web 2.0 with participants on an Technology and TESOL learn about, make use of
MA in Educational Technology and TESOL at the and evaluate Web 2.0 technologies. This module
University of Manchester, UK, we see that the po- is a new departure for the course and represents
tential of Web 2.0 is intimately linked with teachers’ the ongoing need for the MA to refresh itself and
perceptions of their teaching contexts. This chapter to bring new and developing technologies into its
will describe a “context-based” approach to the domain.
exploration of Web 2.0 on a module focusing on the Web 2.0 has its advocates and its detractors;
potential role of distributed courseware in language however, it has become a de facto part of today’s
education. It will begin by giving an overall picture Internet landscape. The very nature of Web 2.0, its
of where and how the exploration of Web 2.0 tools emphasis on such features as collaboration, inter-
fits into the MA program. It will then describe the activity and user-generated content, seems to make
main aims and aspects of the module and discuss it an obvious choice for a focus of discussion when
in some detail our context-based approach in rela- it comes to looking at current trends in the use of
tion to participants as well as Web 2.0 in existing technology in language education. These trends, as
literature. The chapter will conclude with two case shall be later discussed, reflect a focus on learner
studies concerning how teachers incorporate Web centered, collaborative tasks which, in Second
2.0 technologies in courseware for their contexts. Language Acquisition terms, allow channels for au-
thentic language input and output (Chapelle, 1998).
DOI: 10.4018/978-1-60566-190-2.ch007

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
A Context-Based Approach to Web 2.0 and Language Education

However, it is important to realize that for many would want in order to support learners language
language teachers Web 2.0 may simply appear development: they can potentially distribute the
to be another technological innovation that will learning and enable students to be in regular touch
pass them by along with the many others that they with a world-wide community of learners; they
have seen during their career, despite the slowly appear to enable an easier connection to be made
increasing range of references to the uses and between the classroom and the “real” world; they
benefits of key Web 2.0 technologies (e.g. blogs, might enable learners to take some control over
podcasts and wikis) in language education. their learning making use of tools that excite them
As people and communities in various parts and which they are using in their everyday lives;
of the world increasingly embrace Web 2.0, some they seem to offer engagement in active rather than
educational institutions are inevitably responding passive learning, in process as well as product;
to those societal trends and trying to harness Web learners can also potentially engage in discourses
2.0 in their learning programs. Others, although that take them beyond the classroom.
they are in societies where technology is more On the MA program in Educational Technology
normalized (Bax, 2003a) have, for various reasons, and TESOL at the School of Education, University
not taken those technologies on board. Perhaps of Manchester, it is important to explore Web
now, more than at any other time, language teach- 2.0 technologies in language education and help
ers may need to negotiate these changes as they teachers understand generic functions of Web 2.0
impact, or not, on their institutions, and consider in order to facilitate their evaluation of its potential
the implications of ever greater technology use uses. An evaluation of this potential should not, and
for their language teaching. They may be in- cannot, be divorced from considerations pertain-
spired or effectively obliged to engage with the ing to the “ecology” of the teaching environment
nature of Web 2.0 and analyze its affordances in which teachers work, or have worked in the
for language education. Other teachers, even if past and how that pertains to the wider changes
they are aware that it is being used in the wider in society. The use of the term “ecology” here
world may currently see no application for it in signifies all of the rich, interacting elements that
language education. create the dynamic of teachers’ teaching contexts
Web 2.0 is described as relatively easy to use including top down societal, curricula and insti-
and therefore accessible to anybody with access tutional influences and the bottom up influences
to the Internet. This is in contrast to its Web 1.0 which may stem from teachers’ knowledge of and
predecessor which is seen to require at least some enthusiasm for Web 2.0.
familiarity with HTML as a minimum. Setting up In this regard the MA tutors continue to observe
and contributing to a blog for example may seem that the way teachers “make sense” of Web 2.0
comparatively uncomplicated. However, once a genres, i.e. understand how their various intrinsic
blog has been set up the user may be confronted operations — a process which is intimately bound
with concepts and technicalities that may be up with teachers’ “evaluation” of the potential
more difficult to get to grips with, RSS, by way affordances of such software for their contexts
of example. Teachers struggling to understand — is mediated by the teachers’ perceptions of
the concept behind RSS, and the different tech- the context in which they work, or have worked
nologies that support it, are unlikely to be able in the past.
to stand back and evaluate its uses in language Considerations of context are bound up on the
learning terms. MA program with those of pedagogy, and the “fit”
Such issues notwithstanding Web 2.0 tools do of Web 2.0 genres to pedagogical approaches.
appear to offer a lot of what language teachers We have always worked as teacher-educators on

1142
A Context-Based Approach to Web 2.0 and Language Education

the principle that the use of technology in lan- THE MA PROGRAMME


guage education should be firmly underpinned
by considerations of pedagogy and appropriate Brief Overview of the Program
methodology (Holliday, 1994). The functions of
the technologies explored on the MA program The participants on the Master’s program in Edu-
are therefore considered according to how they cational Technology and TESOL come from dif-
can facilitate and possibly enhance pedagogical ferent parts of the world including South America,
approaches that respond to the specificities of Asia, the Middle East and Western and Eastern
different contexts and the needs of learners in Europe and may be either non-native or native
that context. In other words we adopt a context speakers of English. They all have at least three
and pedagogic driven rather than a technology years teaching experience. This level of experi-
driven model. This central focus on pedagogy as ence is a prerequisite for entry onto the degree as
it is relates to context and the role that technology its whole focus is not on our (the tutors) forming
can play in contexts has led us to evaluate the and shaping of the participants’ thinking about
affordances of Web 2.0 as they might respond to teaching, but on facilitating the reflective process
contextual factors, what we have started to term that will allow the participants, drawing from their
a “niche” approach to evaluating Web 2.0 teaching experience, to shape their own thinking
The main discussion in this chapter will centre about their teaching. The participants’ experiences
on one of eight component modules that the MA of using technology vary and range from no use to
program participants take, entitled “Courseware a significant engagement. This obviously affects
Development for Distributed and Blended Learn- the extent to which they can reflect on their own
ing” (CDDBL) and the context-based approach to practice using technology.
CDDBL introduced above. This module explores The MA was set up in the 1980s (in those
a range of Web 2.0 tools and how they may be days it was an MEd) to meet the needs of teach-
exploited both for blending and distributing ers who were becoming interested in using video
courseware. The chapter describes the module’s and computers as part of their language teaching
explorations of Web 2.0 and tutors’ evolving processes. The course has changed considerably
thinking about the way to best approach Web over time, but still keeps as its main foundational
2.0. The changing nature of participants on the aim a focus on the pedagogical implications of
module since its inception in 2005 is discussed the uses of technology (see Wildner, 1999). The
and two case studies will illustrate ways in which specific modules that are relevant to technology
participants have employed or how they envisage and language learning include: Language Learn-
employing Web 2.0 in their own contexts. ing and Technology, which explores the general
In what follows the chapter gives the reader uses made of technology in language classrooms
an overall view of the MA course and how Web to support language skills development; Multi-
2.0 is included, describing our current approach media in Language Education, which combines
to exploring Web 2.0 in CDDBL and discussing an exploration of second language acquisition
aspects of the literature informing the module. processes with the design and development of
It then focuses on two case studies illustrating language tasks using Web 1.0 technologies;
how former participants on the course evaluated Teaching and Learning Online, in which we ask
Web 2.0 in relation to their own contexts. The the participants to explore and reflect on experi-
chapter concludes by discussing possible future ences of online learning. The fourth technology-
developments in our context approach to Web focused module, Courseware Development for
2.0 on CDDBL. Distributed and Blended Learning (CDDBL), is

1143
A Context-Based Approach to Web 2.0 and Language Education

the one that is described in detail in this chapter. As we designed the first iteration of the module
Other modules that students do reflect a more it occurred to us that unlike Web 1.0 technologies,
general TESOL diet. where the extent of interactivity that a learner
We have both onsite and offsite (distance) can engage in is more likely to be determined
participants. The offsite participants study part by the designer/tutor, with Web 2.0 technologies
time as they are generally practicing teachers the development of courseware need not be the
and via an online virtual learning environment preserve of the tutor designer, but also of learners.
(VLE – currently WebCT). Onsite participants Such differences between Web 1.0 and Web 2.0
are studying for the most part full-time and are technologies are explicitly discussed on CDDBL
therefore removed from their teaching context, as are the ways that the two technology types can
particularly if that context is not in the locality be effectively combined to suit the specificities
of Manchester or is overseas. of teaching contexts.
The current iteration of CDDBL therefore
COURSEWARE FOR DISTRIBUTED explores the following Web 2.0 tools: blogs;
AND BLENDED LEARNING wikis; social bookmarking; e-portfolio software;
and podcasting. It further focuses on two VLE
The nature of the program we offer means that platforms, WebCT and Moodle, an open source
there is a continual refreshment of the modules VLE which, in response to ongoing feedback from
and CDDBL is the latest re-working, the first run designers/tutors using the software, continues to
of this new module taking place in 2005. The have new tools incorporated into it, the majority
aim of the module is to assist in the development of which are Web 2.0 tools such as blogs.
of skills that will enable teachers to review and The assessment procedure for CDDBL requires
create effective blended and distributed learning participants to create sample courseware materials
materials for their context, with all of the at- which combine Web 2.0 technologies and which
tendant considerations that this involves. While address issues related to language learning in their
“Multimedia in Language Education” looks at context. They are currently asked to articulate
materials design at task level, CDDBL considers the thinking behind their courseware through a
the integration of activities at the level of a course 30-minute presentation and short executive sum-
or scheme of work. mary and to discuss the courseware in relation to
Early on in the development of CDDBL we relevant educational literature.
took the decision to focus primarily on Web 2.0
tools and their affordances in distributed course- Perspectives on Web 2.0 in
ware. We had originally intended to focus solely the literature and CDDBL
on Virtual Learning Environments (VLEs) but
realized that in doing so we would be missing There are areas of difference and confluence be-
the opportunity to explore emerging technologies tween perspectives on Web 2.0 in the educational
from the perspective of courseware development, literature and our own perspective on important
technologies that potentially change the way we considerations relating to Web 2.0. in courseware
view that development, both in terms of the greater development. We have said that the way partici-
ease with which courseware might be created by pants on CDDBL evaluate the potential of Web
tutors but also in the degree of control that the 2.0 technologies is intertwined with their percep-
participants themselves have in the materials tions of the contexts in which they teach; this is
design process. having an increasing influence on the ongoing

1144
A Context-Based Approach to Web 2.0 and Language Education

development of the module and on our approach used extensively in the literature on language
to the exploration of Web 2.0 genres. teaching. The tools associated with the term tend
It is fair to say that the relationship between to be subsumed under the umbrella terms of Com-
considerations of context in the educational lit- puter Mediated Communication (CMC) and social
erature and the nature and potential of Web 2.0 networking. Nevertheless the potential of those
has not, as yet, been extensively explored. Much tools, as articulated in relation to social network-
of the current literature on Web 2.0 in education ing and CMC, is increasingly recognized. They
discusses it from a general perspective, e.g. with may offer scope for exposure to, and production
regard to the uptake of Web 2.0 in society and of, authentic language use in real life intracultural
particularly among the digital or net generation, and intercultural Internet contexts (Kern, Ware &
and mainly with regard to tertiary education Warschauer, 2004). They also offer the learner
(see Oblinger, 2005; Bryant, 2006). Little of the the chance to use language as it is used on the
discussion on Web 2.0 is, as yet, localized. This Internet and be exposed to “emerging genres of
is not the case with discussion on Web 1.0 tech- language use” (Thorne & Payne, 2005, p. 372).
nologies, where a number of studies relate to the Such opportunities for authentic language output
specificities of different local contexts (see, for and the concomitant opportunities for “noticing”
example, Zhong & Shen, 2002). In CDDBL we and “negotiation of meaning” sit comfortably with
explore with participants the general themes in notions of how SLA takes place (See Chapelle,
the literature on Web 2.0; we provide a summary 1998). Wikis for example, with their text editing
of some of these below. features, may provide the learner opportunities to
Those in the field of education who write on “correct their linguistic output” and “engage in
Web 2.0 technologies see it as holding significant target language interaction whose structure can be
possibilities for the field. A lot of Web 2.0 dis- modified for negotiation of meaning” (Chapelle,
cussion is subsumed under the epithet of “social 1998, pp. 23-24). With these opportunities for
software” which is perhaps both indicative of greater levels of authentic, autonomous language
the cryptic nature of the term Web 2.0 and of the engagement more emphasis will need to be placed
significance of the term “social” in the educational by teachers on the development of metacognitive
field where it is widely argued, partly based on skills among learners, i.e. the skills that learners
the ideas of socio-cultural theorists such as Vy- need to order and develop their own learning. In
gotsky (1978), that learning takes place through some ways many Web 2.0 genres have inbuilt
mediated social interactions. This potential is dis- features, e.g. the wiki edit facility which can
cussed in relation to the creation of new learning facilitate metacognitive thinking. Web 2.0 there-
communities which may offer the “personalised fore, may offer the most genuine medium yet for
collaborative learning experiences such as those breaking down the barriers between the classroom
that are already emerging in the world outside and the real world as not only can the learner use
the school gates” (Owen et al., 2006, p. 11). Such English in an authentic medium but that medium
communities can expand discussion beyond the also provides the tools which allows learners to
classroom and provide new ways for students to focus in an authentic way on how language is
collaborate within their class and across the world used. However, in the same way that Web 2.0
(Bryant, 2006). Wenger (1998) is regularly cited is an extension of Web 1.0 and can be seen to
when discussions of the building of communities have some of its characteristics, it is not a good
beyond classrooms is proposed. idea to view the uses of Web 2.0 technologies as
As with the discussion in the educational field somehow divorced from what has been occurring
generally, the term Web 2.0 has not, as yet, been for many years in the world of Computer Assisted

1145
A Context-Based Approach to Web 2.0 and Language Education

Language Learning. This field has certainly advo- in socio-cultural theory which views humans as
cated extensive use of a variety of technologies to embedded in learning communities where social
promote language learning and has made use of a activity, collaboration and interaction are prime
wide variety of tools to do this. It has also drawn factors in the learning process, are seen as for-
heavily on popular theory from a range of contexts tuitously consonant with what Web 2.0 appears
to support its practices. What is potentially differ- to have on offer. However, there is perhaps a
ent is the way that uses are more easily managed tendency in the literature to assume that there is
by the learners themselves and materials can be a direct unmediated link between Web 2.0 and
more easily learner generated. socio cultural pedagogical approaches and that
In all the above examples the value of the the introduction of Web 2.0 automatically engen-
learning that can potentially occur through Web ders greater learner participation and interaction.
2.0 is seen in relation to the extent to which it is Web 2.0 tools may be predicated on the user as
allied to, driven by and a part of the social, cultural broadcaster rather than audience, as creator rather
and economic trends that are shaping the world. than recipient (Horizon Report, 2007) but when
There is a prevailing discourse of urgency evident such tools are harnessed in educational contexts,
in some of the literature relating to technology in the way that the teacher designs and scaffolds
education, perhaps most pithily encapsulated in the activities within these tools has a prime affect on
phrase, “You can’t not do it” (Collis & Moonen, the extent of and ways that students participate.
2001). This discourse sees the world changing Web 2.0 tools may offer the teacher a malleable
at speed, where economies will be driven by a medium for moulding learner development but it
technologically savvy population, where academic is the teacher’s understanding of how best to work
institutions will need to gear themselves to offer- and craft that medium which may well determine
ing flexible learning programs through various how it works in a language teaching context. The
technologies and where the “digital/net genera- importance of the tutor as designer is stressed to
tion” is not only at home with digital technolo- CDDBL participants.
gies but will be increasingly mystified as to why The way that teachers choose to harness Web
they are not an integral feature of their education 2.0 will depend in large part on their teaching
(Oblinger, 2005). If the net generation’s thinking context and we are particularly careful on CD-
and expectations are shaped by their experiences DBL that in focusing on the way pedagogy can
as net citizens and participants then they will bring be enhanced by technology we do not neglect
those expectations into the educational context considerations of context. While in the literature
where Web 2.0 which is geared around interac- on language education there has been some dis-
tion, will really count. cussion on the need for a “context approach” i.e.
The literature identifies important caveats “an approach that places context at the heart of the
relating to the uptake of technology, not least the profession” (Bax, 2003b, p. 278) and on an “eco-
need for pedagogy to drive the way technology is logical perspective” which looks at the dynamic
used rather than the contrary. Salaberry, (2001) in and negotiated relationship between the richness
his overview of the uses of technology and their of a teaching context and methodology (Tudor,
impact on language learning during the twentieth 2003), a “context approach” tends to be sidelined
century, makes the point that, “new technologies— when it comes to thinking of the triadic relation-
revolutionary as they may be from a strictly tech- ship between pedagogy, technology and context.
nological point of view—are normally regarded In arguing that there are snug and beneficial fits
as revolutionary from a pedagogical standpoint between a technology and a single pedagogical
as well” (p. 39). Pedagogical approaches rooted approach, e.g. social constructivism, there is a

1146
A Context-Based Approach to Web 2.0 and Language Education

risk of propounding a one-fits-all pedagogy which the use of technology has become “normalized,”
is unresponsive to specificities of context. In in the institution, or as part of the learning pro-
fact what we would contend is that the inherent cess (Bax, 2003a) that is to say where the use of
flexibility of Web 2.0 can allow for a blended technology has become an integral, assumed and
pedagogical approach which can respond to local unnoticed aspect of the learning process and where
educational contexts. There is an expanding strand learners consider it perfectly natural to engage in
in the literature that argues that technology, as it language learning, as facilitated by technology, in
is harnessed in careful instructional design, can the institution as well as in their own time outside
be effectively used in this way (see, for example, of the educational environment.
Alonso et al., 2005).
The impact of context
The participants’ contexts on evaluation of Web
2.0 technologies
The importance of a “context approach” on CD-
DBL is underscored with every new cohort that We have mentioned the complex and interacting
participates on the module. CDDBL participants factors that make up the teaching contexts of
come from a multiplicity of teaching contexts participants on CDDBL. These will be different
around the globe, from South and South East Asia for every participant on the module and therefore
to the Middle East to South and North America are best represented on an individual basis (See
to Eastern and Western Europe. section entitled Two Case Scenarios). The low,
A preponderance of participants on CDDBL mid and high-tech categories however, provide
comes from low and mid-tech contexts. We de- us with a general starting point for analyzing the
scribe low, mid and high-tech contexts here both way participants on the CDDBL evaluate Web
from the perspective of the institution and of the 2.0 technologies. A top down and bottom up per-
learner (see Figure 1). The extent to which we spective on the use of technologies also provide
consider an institution as low, mid or high tech a useful conceptual framework for analyzing this
depends on a number of factors; primary among dynamic. By top down is meant societal, curricula
these is the level of computer resources available and institutional factors that push for the further
to learners and teachers in the institution and integration of technologies. Bottom up means
the level of computer know-how among staff in influences that may derive from teachers who
the institution. From the learner perspective we are enthusiastic technology users and may see its
describe context as low, mid or high-tech osten- potential in language education.
sibly according to the level of access they have Our observations to date have led us to conclude
to computer technology inside or outside of the that such contextual factors and the way partici-
teaching institution and their familiarity with that pants represent these as “context-in-mind,” in other
technology. While in a low or mid-tech context words their perceptions of context as they see it
some teachers do use Web 1.0 technologies such in their mind’s eye, (Brown, in preparation), have
as PowerPoint, it is more uncommon for Web some impact on the way they evaluate the potential
2.0 technologies to be used although a clutch of of Web. 2.0. We have, over recent months vari-
participants on the current run of the module are ously described this representation; we have used
using Web 2.0 largely as resource areas for their a “smorgasbord” versus “empty table” metaphor
learners. As yet we have not had any participants where the smorgasbord represents a participant’s
that we consider to be from high-tech contexts. perception of the major potential of Web 2.0 in
By high-tech contexts we mean contexts where relation to context and where the empty table is

1147
A Context-Based Approach to Web 2.0 and Language Education

Figure 1. High-tech and low-tech context

indicative of a context that a participant sees as as hostile to the use of Web 2.0, and where they
entirely unconducive to the use of such technol- feel they have no power to alter that situation (once
ogy. Generally, the empty table metaphor applies again such a perception of constraints generally,
to the perceptions of participants working in low- but not exclusively, relates to participants working
tech contexts. There have been perhaps three or in low-tech contexts).
four participants on CDDBL out of the fifty or In whichever way we choose to describe
so participants since the inception of the module participant perceptions of Web 2.0 in relation
who have perceived Web 2.0 as a smorgasbord. to context, such perceptions do seem to be the
These participants work in mid-tech contexts and primary factor in how participants relate to and
have generally excellent IT skills. They are all to evaluate Web 2.0. In CDDBL therefore, we are
greater or lesser extents bottom up introducers of increasingly trying to steer our approach so that
Web 2.0 in institutions that are generally receptive participants perceive strong connections between
to their ideas. what they are doing on the module and what they
We also characterize the way participants rep- will be able to do in context. While discussion of
resent their context in terms of “considerations,” the literature relating to the use of such technol-
“challenges,” and “constraints.” We find that a ogy can give the participant a general sense of the
participant talks in terms of considerations when possible value of Web 2.0, it does not seem to lead
they can see ways of using Web 2.0 technologies to those moments of recognition and connection
in their context, but where there are issues they when a participant “visualizes” themselves using
feel they need to take into account, e.g. the level the technology in a way that will beneficially ad-
of the language learner in order to effectively do dress issues they have in their context or come
that. The word “challenge” we relate to when a to a keen understanding of why a particular Web
participant sees obstacles to the use of Web 2.0, 2.0 tool is not useful. A decision that a Web 2.0
obstacles that they see as surmountable and where tool cannot offer useful affordances for a specific
they can envisage themselves playing a role in context should be an informed decision coming
overcoming them. We use the term constraint to from strong critical engagement with, and analysis
signify times when a participant sees their context of, the tool.

1148
A Context-Based Approach to Web 2.0 and Language Education

Figure 2. Private language school case scenario

A CLOSER LOOK AT THE “CONTEXT and the location of the school can prove difficult,
APPROACH” ON CDDBL and to a pervasive testing system where learners
who do not pass the test have to repeat exactly
We are using different strategies to facilitate the the same course book materials.
moments of recognition and connection which Along with the case scenario, we also draw on
allow participants to visualize a role for Web 2.0 actual Web 2.0 courseware examples (see Figure
in their teaching. One of these is a “case study 3 for the Web resource page containing these
approach.” One facet of this approach is the use examples) which correspond to specific contexts.
of a semi-authentic case scenario centering on Where possible we try to obtain accounts from the
a private language school. In this scenario (il- courseware developers themselves about the way
lustrated in three descriptions: see Figure 2 for they have developed the courseware in response
the first of these) a number of issues are flagged. to considerations of context. Increasingly we are
CDDBL participants are asked to think about the using courseware examples developed by previ-
role Web 2.0 genres may play in addressing these. ous participants on CDDBL and hope to create a
The issues relate, for example, to the number of bank of exemplars which address various aspects
face-to-face sessions the learners at the school are of context. Some previous participants on CDDBL
required to attend, which, given their busy lives have begun to use courseware conceived on the

1149
A Context-Based Approach to Web 2.0 and Language Education

Figure 3. Courseware examples

module in their local contexts and we hope that we contexts e.g. a lack of time to develop process
will be able to tap into their experiences of this in writing skills in face-to-face language lessons.
order to see how their evaluation of the affordances Web 2.0 has a broad and expanding set of func-
of Web 2.0 works out in practice. We hope that tions — that expansion of functions stemming
this will lead to the formation of a “community from the loop development of Web 2.0 genres
of practice” (Wenger, 1998) centering on the use in response to the way users employ or wish to
of Web 2.0 for language education. employ those genres. These functions resultantly
Another approach we are currently exploring offer a varied range of affordances for language
to facilitate “moments of recognition and connec- education. We feel that the range of functions of
tion” relates to our “niche” evaluation of Web 2.0. Web 2.0 is well suited to a blended pedagogical
Earlier in this chapter we described Web 2.0 as approach.
a malleable medium that teachers can harness to
suit the needs of their local contexts and it is this An Exploration of Blogs
inherent flexibility which is at the root of the niche
approach. The emphasis in this approach is not on It is perhaps best to illustrate this emerging niche
Web 2.0 tools as killer applications that will change approach to Web 2.0 through a discussion of the
the face of language teaching. Instead, we look at functions and affordances of one Web 2.0 tool.
the intrinsic functions of Web 2.0 genres to get a We have chosen to focus on blogs to do this partly
better sense of how they may address specific and because the educational literature on blogs is
perhaps seemingly minor issues within various currently more substantial than the literature on

1150
A Context-Based Approach to Web 2.0 and Language Education

other Web 2.0 tools. This literature is beginning applications may be juxtaposed with the thinking
to analyze the increasingly varied nature of blog behind the “slow blogging” movement. The slow
types and blog interactions and flesh out peda- blogging movement has its own manifesto reject-
gogical approaches that relate to these. In doing ing “immediate” blogging and the “disintegration
so it usefully corresponds to our own emerging into the one-liners” (Slow blogging manifesto,
thinking around Web 2.0 and we can therefore online) “what I am doing now” type blogging.
discuss our approach in close relation to recent It expounds, in contrast, an unhurried, reflective
literature. “speaking like it matters” approach to blogging
A growing number of articles on blogs in educa- which has its roots in the conception of blogging
tion discuss their role in fostering communicative as a diary space.
and collaborative interactions (Belderrain, 2006; We can see from such discussion in the
Cereijol & Myers, 2006; Efimova & Moor, 2005; literature that blogging may be used in educa-
Owen et al., 2006; Williams & Jacobs, 2004). tional settings for a variety of purposes serving
Owen et al. (2006) define the interactional aspect to promote interactions and conversations of
of Weblogs as those properties that allow “read- various types and reflective thinking that is not
ers to comment on postings, to post links to other predicated on interaction. While blogging may be
blogs and through using pingback and trackback consonant with pedagogical approaches rooted
functions (which essentially constitute referencing in socio-cultural theory and therefore predicated
systems between comments on different blog sites) on social interactions, it might equally support
to keep track of other blogs referencing their posts” approaches that are not necessarily intrinsically
(Owen et al., 2006, p. 41). Efimova and Moore connected with such interaction, approaches, by
discuss the “distributed” generally “spontaneous” way of example, rooted in cognitive theory. On
conversational interactions which blogs can en- CDDBL we discuss how the various uses that
gender (2005, p. 1), conversations that are tightly blog environments can be put to and the types
associated with the functions of blogs, namely the of user behavior they can engender can relate to
“comment” feature, “trackback” function and RSS the specificities of CDDBL participants’ teaching
aggregator. Efimova and Moor’s research into contexts. To this end we are increasingly using
conversational blogging is particularly useful in terms which reference the specific character and
its explorations of how specific functions of blogs variety of blog spaces such as micro blogs, slow
may relate to the types of interactions that take blogs, soap blogs etc.
place through it. We also focus on the setting panel in blogs, an
As blogging becomes increasingly popular new aspect of blogs that has been little explored in the
tailored blogging environments have been created educational literature. The settings area of blogs
that respond to and cater for changes in interac- (the areas that provides customization functions)
tional types. Twitter (http://twitter.com/) and Jaiku allows the blog owner to disable blog comments
(http://www.jaiku.com/) are both mobile blogging and trackback functions which can effectively
environments which support brief and frequent seal a blog off from interaction. Comments can
“What I’m doing now” type interactions. Such be approved or rejected by a blog owner before
mobile software applications with their ability to they appear on the blog and the owner can also
provide for embedded/contextualized interactions determine who views their blog and who has a
may potentially offer “virtual and real-world sup- role as an author.
port for social interactions and collaboration in a Knowing about these blog properties is im-
real-world context” (de Jong et al., 2008, p. 121). portant for teachers as they allow for a nuanced
The thinking behind such interactive software methodology in relation to context. Group blogs

1151
A Context-Based Approach to Web 2.0 and Language Education

may be set up by teachers to allow the full range Both participants have called on the affordance
of interactions that blogs can afford, including of Web 2.0 tools in interesting ways to address
learner permission to edit the blog. Learners can issues within their context.
set up their own blogs and have full control of
the permissions on that blog. However a teacher Vida Zorko: University of
can take a more prominent controlling role of a Ljubljana (Offsite Participant on
group blog space, in order to facilitate scaffolded the First Run of CDDBL, 2005)
interaction. Such setting functions may prove
useful in some contexts. Vida teaches English for Specific Purposes (ESP)
The participants on CDDBL who work in and develops courseware for groups of sociology
South Asian and South East Asian locations talk students studying at the Faculty of Social Sciences
about the “teacher-centered” contexts they work at the University of Ljubljana who receive ESP
in where the teacher is expected to be the “sage tuition as a part of their degree. Vida felt that a
on the stage,” rather than the “ghost in the wings,” move from a “traditional” lecture-based approach
a metaphor for a teacher who plays a hands-off to Problem Based Learning (PBL) (Savin-Baden,
facilitory role (see Mazzolini & Maddison, 2003). 2000) would better serve the students in their
Such participants, while they see major potential learning. She was instrumental in introducing that
for a more learner-centered approach — which approach, an approach for the most part approved
they view particularly as a beneficial means of of in the Sociology Department. This inevitably
developing fluency skills in English — council required a change in ways of working, both on
a carefully scaffolded approach which slowly in- the part of teachers and students, and impacted
troduces learners to greater autonomous peer-peer on the ESP provision. Vida felt that the introduc-
interaction (see the case study on Andrew Prosser). tion of a wiki (in this case a pbwiki), which she
The setting affordances of blog environments can considered well suited to the social constructiv-
support that transition. ist learning that underpins PBL, could play an
instrumental role in facilitating this change. She
TWO CASE STUDIES combined the use of a wiki, in which students
working in small groups solved real-life sociol-
This chapter has discussed the reasons why a ogy related problems, with a blog space, which
context-based approach has been adopted, particu- she used to co-ordinate aspects of the blended
larly as it regards CDDBL participant perceptions online and face-to-face learning approach and to
of their teaching context. It has explored the way offer advice and help when students encountered
the context-based approach aims to help teachers problems. The students could also access Web 1.0
critically evaluate Web 2.0 for their contexts by html pages, which were used for the delivery of
focusing on the flexibility of Web 2.0 and the range language learning resources and activities.
of pedagogical approaches it may be associated Vida felt that the role of the wiki would prove
with. In what follows, ways in which two CD- valuable in:
DBL participants have perceived the potential of
Web 2.0 for their context are presented and their • Promoting peer-to-peer, teacher-teacher
approach to harnessing the affordances of Web and student-teacher interactions neces-
2.0 tools to address aspects of those contexts ex- sary (as Vida sees it) for the successful
plored. The two participants work in two diverse institution of a problem based teaching
contexts. One of the contexts may be character- approach.
ized as verging on high-tech and one as mid-tech.

1152
A Context-Based Approach to Web 2.0 and Language Education

• Increasing motivation by publicly display- an abundance of opportunities offered by Web 2.0


ing group products. for her context. She looked on wikis as a tool that
• Facilitating the sharing of knowledge with careful scaffolding could facilitate the PBL
among students and teachers. approach that she had instituted and support a sea
• Empowering students with the authority to change in ways of learning in the Sociology De-
construct their own knowledge. partment. The PBL approach adopted by Vida and
• Enabling teachers to better assess students’ the Sociology Department is generally perceived,
progress by monitoring the history of the in the literature, as consonant with the nature of
process. Web 2.0. The Sociology Department was, more-
over, amenable to Vida’s ideas. In these respects
Vida felt that these potential collaborative Vida’s context arguably offered fertile ground for
affordances link to the following features of the the introduction of Web 2.0 and may have made it
PBwiki: easier for Vida to evaluate Web 2.0 and envisage
for it a concrete role. Nevertheless, the introduction
• An interface which is easy to modify to of Web 2.0 stemmed largely from Vida’s bottom
make it more transparent for users. up initiatives and efforts to persuade tutors of
• 1-click access to all areas, thus promoting the value of the wiki. She saw this process as an
greater sharing of knowledge, making stu- enjoyable “challenge,” referring back to the three
dent and tutor contributions easily acces- “c” considerations, challenges and constraints
sible and allowing tutors to better monitor framework, rather than as a constraint that would
student progress and to collate reoccurring impede the introduction of Web 2.0.
language problems and deal with them in a It is perhaps possible that the bottom up influ-
face-to-face environment. ence from Vida, and increasingly her fellow tutors,
• The “whose online” and “edit” function along with the top down department approval
that allow tutors to see who is working in will conspire to normalize Web 2.0 in the faculty,
the wiki at a given time, and to respond al- making it the first context we have encountered
most immediately to student contributions. where this is the case. Early indications through
From this perspective Vida characterizes research Vida conducted for her MA dissertation
the wiki as “almost a synchronous space.” show that the wiki has proved valuable in insti-
• A comment area that allows for easy dia- gating greater collaborative learning and is fast
logue between student and student, and becoming a “normalized” tool (Bax, 2003a).
student and tutor.
Andrew Prosser, Private Language
However, Vida was aware that such func- School, Seoul (Offsite Participant
tions, in and of themselves, would not bring on Second Run of CDDBL, 2006)
about the benefits she felt the use of a wiki would
introduce. The student wiki pages all adhere to We might contrast Vida’s context with that of
a certain format (though students can adjust that Andrew Prosser’s. Andrew’s is a mid-tech context
format themselves) that scaffolds the way the and is in some ways amenable to Web 2.0 use
students work with each other, as in the example in that his learners all have access to computers
in Figure 4. and have some familiarity with Web 2.0 as the
In Vida’s case the “smorgasbord” metaphor majority of them enjoy blogging, a popular Web
referred to earlier in this chapter is entirely ap- 2.0 tool in Korea.
propriate as she saw, as a participant on CDDBL,

1153
A Context-Based Approach to Web 2.0 and Language Education

Figure 4. Example page of Vida’s wiki environment

It would, however, be an exaggeration to say Andrew harnessed the popularity of blogging in


that Andrew saw Web 2.0 in terms of a smorgas- his teaching context by creating courseware with
bord of opportunity for his context, particularly blogs and in Moodle that would lead ultimately
where Web 2.0 is associated with highly learner- to the learners creating their own “tourist guide
centered, autonomous learning. Andrew described to Seoul” blogs. He drew, in the initial stages of
his context as essentially teacher-centered, where his courseware, on those affordances of Moodle
the teacher is viewed as the “sage on the stage” and blogs that he considered in keeping with an
(Mazzolini & Maddison, 2003), “transmitting” “associative” instructional design approach to
knowledge to be memorized by learners. He saw courseware design, an instructional design ap-
the value of a more learner-centered approach in proach which is mainly tutor determined and uses
encouraging greater learner autonomy and learner a linear navigational design structure which asks
interactions which would, in turn, have dividends learners to go through a series of tasks in order
in terms of language development. However, he to assist learners in mastering a specific language
did not advocate a total shift to a learner-centered structure or skill. To this end he created controlled
approach and argued that nudging learners towards Web 1.0 practice exercises in Moodle (see Figure
greater autonomy would require a careful structur- 5), and a blog that modeled the way learners may
ing of courseware. He had a cautiously optimistic approach their tourist blog. Through this scaffold-
approach to the value of Web 2.0 in such course- ing process he gradually shifted from a teacher-
ware but once again felt that its value would be centered approach, to the more learner-centered
highly contingent on careful scaffolding.

1154
A Context-Based Approach to Web 2.0 and Language Education

Figure 5. Andrew’s Moodle environment

blog task that tapped into the learner-centered on CDDBL perceive their contexts in these terms
affordances of blogs. and we do not anticipate that this situation will
Andrew’s Web 1.0 and Web 2.0 meld and his change any time soon. This said, if Web 2.0 con-
perceptions of the potential of Web 2.0 for blend- tinues to be integrated into the fabric of societies
ing pedagogies in many ways constitute a “niche” at its current speed then it is likely in the longer
approach to his own context. term that institutions and the teachers within them
will increasingly need to negotiate their use. The
CONCLUDING REMARKS: issue then will be less one of “empty tables” and
THE FUTURE OF CDDBL “constraints” and more one of how Web 2.0 can
best be used. This may well engender the type of
In choosing to focus on Vida and Andrew we have bottom up thinking demonstrated by Vida Zorko
illustrated two contexts in which both teachers and the “niche” thinking of Andrew. Teachers may
have envisaged a key role for Web 2.0. and suc- increasingly also need to negotiate top-down deci-
cessfully incorporated it into courseware. There is sions about the use of technology and the extent
not scope in this chapter to explore case scenarios to which the use of technology should be Web 1.0
where CDDBL participants have perceived the based and Web 2.0 based. We hope that CDDBL
introduction of Web 2.0 in their contexts entirely will help participants see clear ways to play a role
in terms of “empty tables” or “constraints” militat- in, and negotiate these influences.
ing against their incorporation. However, as we In the latest 2007 offsite run of CDDBL a
have discussed earlier a number of participants small proportion of the participants were already

1155
A Context-Based Approach to Web 2.0 and Language Education

enthusiastically using Web 2.0 in their contexts transformative potential. We have also argued
before the module commenced and are indeed that the potential for Web 2.0 goes hand-in-hand
bottom up initiators of its use. They have a strong with the way it is harnessed by tutors/designers
grasp of the functions of the technologies, even if to suit local contexts.
they have not greatly explored their pedagogical It is clear that information/digital technologies
possibilities. By the end of the module all of the are an increasing feature of net migrants’ lives
participants are actively contemplating using Web and that for the generations coming through the
2.0 technologies in their teaching and learning ability to stay socialized via technologies will be
situations. Assignment presentations that we have a significant part of their identities. Of course, this
viewed include: the use of online video to encour- landscape will continue to change and the elements
age better presentation skills for trainee teachers on the table will constantly refresh, however, we
in Japan; the introduction of blogs to encourage believe that our particular approach will enable
more accurate writing skills in Mexico; the use of both ourselves and the module participants to deal
Moodle as a delivery platform to supplement in with these changes in an informed and pedagogi-
class activity; the use of Ning as a tool to increase cally appropriate way.
participation in e-learning; the use of RSS feeds to
support the development of learner autonomy in ACKNOWLEDGMENTS
Japan; the use of blogs as an e-portfolio in primary
schools in Greece; the use of Moodle, wikis and The authors would like to thank Vida Zorko and
instant messaging to introduce a greater language Andrew Prosser for their kind permission to write
element into cultural visits in the UK. about their perceptions of Web 2.0 and their con-
There will continue to be participants who texts and for their invaluable assistance in ensuring
view Web 2.0 with skepticism largely because these were faithfully represented.
they see their context as militating against its
use. However, interest in the module grows and
assignment presentations show the inventiveness REFERENCES
of the module participants, their increased ability
to analyze their contexts, to bring theory to bear Alonso, F., Lopez, G., Manrique, D., & Vines,
and to integrate a variety of Web 2.0 technologies J. (2005). An instructional model for Web-based
into their teaching. We feel that the context-based, E-Learning education with a blended learning
niche approach we are adopting, which we will process approach. British Journal of Educational
continue to develop, will help to increase this Technology, 36(2), 217–235. doi:10.1111/j.1467-
interest and give our module participants op- 8535.2005.00454.x
portunities to use Web 2.0 in ways they feel will Bax, S. (2003a). CALL — past, present and
enhance their learning context however minor or future. System, 36(1), 13–28. doi:10.1016/S0346-
substantial these modifications may be. 251X(02)00071-4
There will continue to be debate and argument
about whether Web 2.0 is somehow different and Bax, S. (2003b). The end of CLT: A context ap-
transformative in its very nature. We have argued proach to language teaching. English Language
here that, given its flexibility, which we see as Teaching Journal, 57(3), 278–287.
conducive to a bended pedagogical approach,
and the possibilities that it offers for breaking
down the barriers between the real world and the
classroom, it does have a different and possibly

1156
A Context-Based Approach to Web 2.0 and Language Education

Beldarrain, Y. (2006). Distance education trends: Holliday, A. (1994). Appropriate methodology


integrating new technologies to foster student and social context. Cambridge: Cambridge Uni-
interaction and collaboration. Distance Educa- versity Press.
tion, 27(2), 139-153. Retrieved November 24,
Kern, R., Ware, P., & Warschauer, M. (2004).
2007, from http://www.informaworld.com/smpp/
Crossing frontiers: New directions in online
content~content=a749174134
pedagogy and research. Annual Review of Applied
Brown, S. (in preparation). Context-in-mind: A Linguistics, 24(1), 243–260.
cultural historical perspective on language teach-
Kumaravadivelu, B. (2001). Towards a post-
ers’ perceptions of social software.
method pedagogy. TESOL Quarterly, 35(4),
Bryant, T. (2006). Social software in academia. 537–559. doi:10.2307/3588427
EDUCAUSE Quarterly, 2, 61-64. Retrieved No-
Mazzolini, M., & Maddison, S. (2003). Sage,
vember 25, 2007, from http://www.educause.edu/
guide or ghost? The effect of instructor interven-
ir/library/pdf/eqm0627.pdf
tion on student participation in online discussion
Cereijo, M., & Myers, C. (2006). Weblogs: New forums. Computers & Education, 40, 237–253.
communication technology uses in resource doi:10.1016/S0360-1315(02)00129-X
limited environments. In Gonzales et al., (Eds),
Oblinger, D. (2005). Educating the Net genera-
Current Developments in Technology Assisted Ed-
tion. EDUCAUSE e-book, Boulder, Colorado.
ucation, 1 (pp.720-726). Retrieved November 24,
Retrieved November 25, 2007, from http://www.
2007, from http://www.formatex.org/micte2006/
educause.edu/LibraryDetailPage/666?ID=PUB7
Downloadable-files/oral/Weblogs.pdf
101
Chapelle, C. (1998). Multimedia call: Lessons
Owen, M., Grant, L., Sayers, S., & Facer, K.
to be learned from research on instructed SLA.
(2006). Social software and learning: An Open-
Language Learning & Technology, 2(1), 22–34.
ing Education report from Futurelab. Retrieved
Collis, B., & Moonen, J. (2001). Flexible learning November 30, 2007, from http://www.futurelab.
in a digital world. London: Kogan Page. org.uk/research/opening_education/social_soft-
ware_01.htm
De Jong, T., Sprecht, M., & Koper, R. (2008). A
Reference model for mobile social software for Report, H. (2007). New Media Consortium. Re-
learning. Int. J. Cont. Engineering, Education trieved January 29, 2007, from http://www.nmc.
and Lifelong Learning, 18(1), 118-138. Retrieved org/pdf/2007_Horizon_Report.pdf
November 25, 2007, from http://dspace.learn-
Salaberry, R. (2001). The use of technology for
ingnetworks.org/bitstream/1820/996/6/08+De+
second language learning and teaching: A ret-
JONG_16.pdf
rospective. Modern Language Journal, 85(1),
Efimova, L., & Moor, A. (2005). An explor- 39–56. doi:10.1111/0026-7902.00096
atory study of conversational blogging practices.
Savin-Baden, M. (2000). Problem-based learn-
Proceedings of the 38th Hawaii International
ing in higher education: untold stories. SRHE &
Conference on Systems Sciences. Retrieved No-
Open University Press, Buckingham.
vember 24, 2007, from http://ieeexplore.ieee.org/
iel5/9518/30166/01385452.pdf?tp=&isnumber=
&arnumber=1385452

1157
A Context-Based Approach to Web 2.0 and Language Education

Thorne, S. L., & Payne, J. S. (2005). Evolution- means that the potential of technologies cannot be
ary trajectories, Internet-mediated expression, evaluated in abstract terms but as it is interlinked
and language education. CALICO Journal, with contextually appropriate practice.
22(3), 371–397. Retrieved November 30, 2007, Ecology (Teaching Environment): The
from http://language.la.psu.edu/~thorne/thorne_ teaching ecology refers to all the rich, interacting
payne_calico2005.pdf elements that create the dynamic of a teacher’s
context. These may include top down societal,
Tudor, I. (2003). Learning to live with complexity:
curricula and institutional elements and bottom
towards an ecological perspective on language
up elements such as learner requests to use more
teaching. System, 31(1), 1–12. doi:10.1016/S0346-
technology in the classroom.
251X(02)00070-2
High-Tech Contexts: One where the use of
Vygotsky, L. S. (1978). Mind in society. Cam- technology is integrated into everyday life, so
bridge, MA: Harvard University Press. you would expect there to be easy access to the
internet, probably these days through wi-fi; for
Wenger, E. (1998). Communities of practice:
the bulk of the population to carry mobile phones
Learning, meaning, and identity. New York:
and for technology to feature strongly in the edu-
Cambridge University Press.
cation system.
Wildner, S. (1999). Technology integration into Low-Tech Contexts: Whilst the middle
preservice foreign language teacher education classes may have access to mobile phones and
programs. CALICO Journal, 17(2), 223–250. access to the internet at home, schools may only
have traditional computer rooms which may not
Williams, J. B., & Jacobs, J. (2004). Exploring
well be networked. Access to the internet for the
the use of blogs as learning spaces in the higher
general population is via internet cafes in urban
education sector. Australasian Journal of Educa-
areas rather than through wi-fi.
tional Technology, 20(2), 232–247.
Pbwiki: One of a burgeoning number of wiki
Zhong, Y., & Shen, H. (2002). Where is the environments. The following page provides a
technology-induced pedagogy? Snap-shots from useful comparative analysis of different wiki
two multimedia EFL classrooms. British Jour- environments: http://www.wikimatrix.org.
nal of Educational Technology, 33(1), 39–52. Trackback/Pingback: Links that allow blog
doi:10.1111/1467-8535.00237 users to reference content on each others’ blogs.
For example, say every learner in a class has
their own blog and one learner embeds a video
file in their blog about a trip they have been on,
KEY TERMS if other learners comment on the video in their
own blogs and use the trackback function, this
Context-Based Approach: An approach that
will automatically show in the blog of the learner
encourages teachers to have the confidence to
who embedded the video. Note that blogger.com
creatively reflect on their teaching practice as it
does not currently offer the trackback and ping-
responds to the particularities of their own teaching
back function
contexts. Kumaradivelu refers to this as a “teacher
generated theory of practice” (2001, p. 541). This

This work was previously published in Handbook of Research on Web 2.0 and Second Language Learning, edited by M. Thomas,
pp. 119-136, copyright 2009 by Information Science Reference (an imprint of IGI Global).

1158
1159

Chapter 4.16
An Adaptive and
Context-Aware Scenario Model
Based on a Web Service
Architecture for Pervasive
Learning Systems
Cuong Pham-Nguyen
TELECOM, France

Serge Garlatti
TELECOM, France

B.-Y.-Simon Lau
Multimedia University, Malaysia

Benjamin Barbry
University of Sciences and Technologies of Lille, France

Thomas Vantroys
University of Sciences and Technologies of Lille, France

ABSTRACT on pervasive TEL systems in a learning situation


at the workplace. We introduce a context-aware
Pervasive learning will become increasingly im- scenario model of corporate learning and work-
portant in technology-enhanced learning (TEL). ing scenarios in e-retail environments such as
In this context, development efforts focus on shops and hypermarkets. This model enables us
features such as context-awareness, adaptation, to integrate contextual information into scenarios
services retrieval and orchestration mechanisms. and to select how to perform activities according
This paper proposes a process to assist the de- to the current situation. Our pervasive learning
velopment of such systems, from conception system is based on a service oriented architec-
through to execution. This paper focuses mainly ture that consists of an infrastructure for service

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

management and execution that is flexible enough This process is reciprocal: the environment also
to reuse learning components and to deal with does it and becomes “intelligent”. In ubiquitous
context changes that are not known in advance and computing, the main goal is to integrate large-
discovered on the fly.. [Article copies are available scale mobility with pervasive computing func-
for purchase from InfoSci-on-Demand.com] tionalities.
In this article, we consider that mobile, per-
vasive and ubiquitous learning systems have the
INTRODUCTION properties of mobile, pervasive and ubiquitous
computing systems respectively. We focus our
Nowadays, technology-enhanced learning (TEL) attention on pervasive learning systems. Mobile
systems must have the capability to reuse learning learning is not just about learning at anytime,
resources and web services from large reposito- at any place and in any form using lightweight
ries, to take into account the context and to allow devices, but learning in context and across con-
dynamic adaptation to different learners based texts. It is best viewed as providing mediating
on substantial advances in pedagogical theories tools in the learning process (Sharples, 2006).
and knowledge models (Balacheff, 2006). This is Many definitions of pervasive learning are given
particularly true of mobile learning, where context in the literature (Bomsdorf, 2005; Hundebol &
is variable. The reuse of learning resources and Helms, 2006; Jones & Jo, 2004; Thomas, 2007).
web services requires interoperability at a seman- One useful definition is that a “pervasive learning
tic level. In other words, it is necessary to have a environment is a context (or state) for mediating
semantic web approach to design TEL systems. learning in a physical environment enriched with
Moreover, knowledge models and pedagogical additional site-specific and situation dependent
theories can be fully represented by means of a elements – be it plain data, graphics, information
semantic web approach. In the mobile learning -, knowledge -, and learning objects, or, ultimately,
area, a number of terms are commonly used; mo- audio-visually enhanced virtual layers” (Hunde-
bile, pervasive and ubiquitous learning systems bol & Helms, 2006). One could consider pervasive
(Brodersen, Christensen, Gronboek, Dindler, & learning as an extension to mobile learning where
Sundararajah, 2005; Hundebol & Helms, 2006; the roles of the intelligent environment and of the
Sharples, 2005; Thomas, 2007). In computer sci- context are emphasized (Laine & Joy, 2008). In
ence, mobile computing is mainly about increasing pervasive learning, computers can obtain informa-
our capability to physically move computing tools tion about the context of learning from the learn-
and services with us. The computer becomes an ing environment where small devices, sensors,
ever-present device that expands our capabilities pads, badges, large LCD screens, people, and so
by reducing the device size and/or by providing on, are embedded and communicate mutually.
access to computing capacity over the network The physical environment is directly related to
(Lyytinen & Yoo, 2002). In mobile computing, an learning goals and activities. The learning system
important limitation is that the computing model is dynamically adapted to the learning context.
does not change while we move. This is because Consequently, a pervasive learning system needs
the device cannot obtain information about the to have an appropriate software architecture to
context in which the computing takes place and support these features.
adjust it accordingly. In pervasive computing, the In the workplace context, learning can oc-
computer has the capability to inquire, detect and cur in purposeful situations in which there is
explore its environment to obtain information an explicit goal to learn as well as in incidental
and to dynamically build environment models. situations in which there is no explicit learning

1160
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

goal or interest. Working involves an activity or tivities through shops and hypermarkets). In such
a related set of activities that require effort and a framework, we are interested in the following
are aimed at achieving one or more objectives. learning issues: the combination of formal learn-
Learning emphasizes what a learner knows or is ing (formal classroom at the workplace, etc.) and
able to do, while, in contrast, working is related work-learning integration, integration of mobile
to performance improvement (Michael-Spector & devices in broader lifelong learning and working
Wang, 2002). In other words, when performing scenarios, learning in context, seamless learning
a work task, it often happens that learning also across different contexts and context-as-construct
occurs. The performance and quality of work may (Balacheff, 2006; Sharples, 2006; Vavoula &
also be enhanced following learning experiences. Sharples, 2008). In such a framework, we focus
Working activities are mainly about solving prob- on a scenario-based approach for TEL system
lems, and in knowledge-intensive organizations design. Scenarios are used to describe the learn-
this implies continuous learning. Carrying out the ing, working and tutoring activities to acquire
particular working task is the priority; learning some domain knowledge and know-how, solve
is just a means (Farmer, Lindstaedt, Droschl, & a particular problem or support working activi-
Luttenberger, 2004). The distinction between ties. Scenario analysis reveals that learning and
learning and working activity is blurring, working working situations can be modeled by an explicit
being a way of learning, and vice versa. Simon hierarchical task model because working and
(2007) asserts that traditional methodologies such learning activities are well structured and stable. In
as formal classroom teaching and even Internet pervasive learning systems, activities, represented
based, content oriented courses and programs have by tasks, can be achieved in different ways accord-
their place at the worksite. Nevertheless, these ap- ing to the current situation. Methods associated
proaches are generally inflexible to the demands with tasks enable us to provide different ways to
of contextualised, learner centred, performance carrying out those tasks. Activities need to have
related challenges (Simon, 2007). Thus, learning access to supporting resources or web services.
processes need to be embedded in organizations, Thus, a context-aware and adaptive mechanism
so that learning becomes pervasive and a natural is necessary to select relevant methods associated
part of work. A particular architecture is required with a task and their corresponding resources
to facilitate the redefinition of learning to mean and web services. For a particular couple (Task,
a work activity and to provide an infrastructure Method), resources and web services may also be
for seamless work-learning integration (Simon, selected according to the current situation.
2007). In such framework, situated learning can In pervasive computing, the computing device
be used, where the location, time, environment and has to seamlessly and flexibly obtain information
tasks, etc. are taken into account. It provides the about its environment in which the computing
right learning support at the right time according takes place and to adjust itself accordingly. From
to the situation parameters and to the goals in the a software architecture viewpoint, a pervasive
working context. Situated learning increases the learning system has to be flexible enough to
quality of learning and is attractive for learning at reuse learning components (learning resources
the workplace and for work-learning integration or learning web services) which are not known
(Oppermann & Specht, 2006). in advance and discovered on the fly. A service
In the p-LearNet project (p-LearNet, 2006), oriented architecture (SOA) approach facilitates
a pervasive learning system aims to integrate the deployment of an adaptive learning environ-
context-aware corporate learning and working ment based on the aggregation and orchestration
activities within the e-retail framework (retail ac- of the services needed by an organization. This

1161
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

approach can be effective for pervasive learning explore its environment to obtain information
systems if one provides for continuous adapta- and to dynamically build environment models.
tion based on the available services and other As web services are closely related to learning
contextual information. needs by means of the scenario model, we can
The main contribution of this article is an provide the right learning support according to
adaptive and context-aware model of scenarios for the current situation and deal with pervasive
a pervasive learning system supporting working computing issues.
and learning activities. The pervasive learning In the next section we begin by presenting
system architecture is based on a service ori- the p-LearNet project and e-retail system goals.
ented architecture to meet pervasive computing Secondly, we present our pervasive learning
requirements. Web services are retrieved and system architecture based on SOA that serves
orchestrated, and can be used for different working as a platform of service management in e-retail
and learning activities. Thus, the scenario model systems. Thirdly, our context management model
can invoke web services to undertake activities. (organization, features) is detailed and linked
The scenario model and the web service retrieval to a situation. Fourthly, we present our context-
and orchestration are based on a semantic web aware scenario model and its relationships with
approach which enables us to represent the ex- the context management model. Fifthly, the web
plicit common knowledge of the communities of service requirements and specifications are pre-
practice involved in the p-LearNet project. The sented and also the semantic metadata schema
scenario model is based on a hierarchical task describing the web services. Sixthly, the context
model having the task/method paradigm. An management process, i.e. the detection and cre-
activity, represented by a task, may have several ation of new situations are explained. Seventhly,
associated methods. A method represents a way the adaptation process which links the context
of performing a task in a particular situation. model, the scenario model, the metadata schema
The context-aware and adaptive mechanism can and the context management is presented. Finally,
be viewed as the selection of the relevant content a conclusion highlights the main results of this
(methods or web services) for a given task ac- study and some perspectives on the results.
cording to the current working and/or learning
situation. This mechanism is based on matching
content description to the current situation for P-learnet project AND
filtering, annotation and ranking. Content and e-retail systems
situation need to have corresponding features
for adaptation purposes. Methods are described The 3-year p-LearNet project is an exploratory
by contextual features while web services are work on adaptive services and usages for human
described by metadata. Situations are described learning in the context of pervasive communica-
according to a context model. For managing web tion. One aim is concerned with providing learning
services, we also define the service requirement to professionals during their work activities. This
specification for web service retrieval. Moreover, project therefore addresses concepts and method-
a pervasive learning system architecture is pro- ologies to facilitate this type of learning. In such
posed to facilitate its design and execution in a a framework, the main issues of the p-LearNet
workplace environment, based on Open Services project are: work-integrated learning and customer
Gateway initiatives (OSGi) and Universal Plug learning support, continuous professional learning
and Play (UPnP). SOA enables us to design an at the workplace, professional learning whatever
architecture that is able to inquire, detect and the place, the time, the organisational and tech-

1162
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

nological contexts of the individual or collective RFID, etc.) surrounding them (Derycke, Chevrin,
learning and working processes, context-as- & Vantroys, 2007).
construct and seamless learning. The design and In the e-retail framework, the context is con-
engineering of pervasive learning systems must tinuously evolving during the selling process: a
be considered as an interdisciplinary problem seller can communicate with customers while
requiring the integration of different scientific revising his/her knowledge, checking the inven-
approaches from computer science, education, tory or contacting the supplier about products, etc.
commerce, social sciences, etc. Learning focuses Sometimes learning and working are interwoven
mainly on how to support individual and group in a pervasive environment. A substantial part of
learning processes through pedagogical guidance learning does not happen during training but dur-
and how to enhance the learner’s knowledge and ing working activities. Learning and professional
know-how. In this project, one of our corporate activity support must therefore be integrated. A
partners is an international retail companies hav- learning system must overcome three main ob-
ing chains of shops and hypermarkets wishing stacles: time pressure, inadequate learning support
to explore sales staff learning at the workplace. in the working context and cognitive and structural
Corporate partners identify the problems and disconnection between work, knowledge, and
requirements about quality and efficiency of in- learning (Farmer, Lindstaedt, Droschl, & Lut-
formation and services to increase market share tenberger, 2004). For example, a seller equipped
and the corresponding learning goals. Several with a portable device, for example a PDA or a
innovative scenarios have been set up according UMPC, close to shelves (without a customer) can
to two main learning and/or working situations for revise thier knowledge about products and selling
both a seller and a customer as learners: i) Seller techniques or can continue thier previous learn-
or customer, outside the shop counters: seller ing activities to improve their knowledge, or they
in the back office or storage areas, customer at can verify product labeling or access to product
home or elsewhere; ii) Seller in his department, information. During the selling process or the
alone or with a customer having resources from selection / decision phase, the seller can use his/
the Smart Spaces (large LCD screen, printers, her mobile device as a coach to help them, etc.

Figure 1. Learning requirements and needs at the workplace

 What Need
Need toto model
model
activities activities
activities
can I do in (scenario)
(scenario)
 How can I
get support or the selling
-Need
-Need to
to situation?
help?
integrate
integrate learning
learning
and
and working
working activities
activities  How can I
-Support
-Support continuous
continuous realize an activity
professional
professional learning
learning when the
at
at the
the workplace
workplace or or working situation
elsewhere
elsewhere  What and changes?
how can I learn
during working
activities? Seller Need
Need toto achieve
achieve
activities
activities in
in
different
different contexts,
contexts,
across
across different
different
contexts
contexts

1163
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

In such a framework, the following needs are which is composed here of a set of devices offering
required (cf. Figure 1). services. These devices can communicate with
One of the key problems is the requirement each other and they can be reachable by anyone,
for continuously following the user into a wide within the constraints of access rules (e.g. sellers
variety of contexts, and our efforts address this can access more services than customers, etc.).
issue inside the intermediation infrastructure, Thus, the selling space becomes a part of the
between the service universe and the communica- process and can collect information (Derycke,
tion universe, by adding “intelligence” to provide Vantroys, Barbry, & Laporte, 2008). Our pervasive
seamless services regardless of channel. The next learning system can access services provided by
section will present the pervasive learning system devices, external services and semantic composi-
architecture we developed to deal with our require- tion services. External services may be provided
ments. We then explain in more detail the context by the company information system. Semantic
and the context-aware scenario model. composition services are mainly composed of
four modules (scenario, metadata, context and
domain) corresponding to the models mentioned
Pervasive learning in the next sections. Each module is itself a set of
system Architecture services and managed by the Context Manager
System component (cf. Figure 2). These services
We developed a generic and flexible software in- can be used by the devices and by anyone in the
frastructure (cf. Figure 2) to support learning and sales department (sellers, customers, etc.). Activi-
working activities at workplace. As described in ties are another source for context.
the introduction, learning in context must support Secondly, our architecture allows us to ag-
certain functionalities, as listed below. gregate and to coordinate all available services,
Firstly, we have to recover information or data both from software (learning system, database
from the context, such as activities (current and access, information system, etc.) and from the
available), location, and devices available in the environment (televisions, webcams, printers,
environment. We identified two main sources for etc.). It is what we call extensibility and modu-
context recovery. One is the user’s environment, larity. Extensibility means that we can integrate

Figure 2. An overview of pervasive learning system architecture

1164
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

and use any new services (e-learning modules) tecture is possible thanks to OSGi (OSGi, 2007).
or devices (printers, televisions, etc.) We have OSGi is a software layer over Java. It reuses the
to support live detection of service abilities. For dynamic class loading capabilities of the Java
example when a device is turned on or off, the language. It enables a development model where
system knows if its services are available or not. applications are (dynamically) composed of
Then, we need semantic discovery of all available many different (reusable) components (bundles).
services. Modularity means addition or removal It acts as a shell over the Java Virtual Machine:
of some interactive elements, for adaptation to the it manages the component life cycle, so we can
needs of a particular retail company. The services dynamically add or remove components. Each
provided have to be modular too: the system can component hides their implementation from other
provide services with different implementations components while communicating through ser-
to match the location, time, users, devices, etc. vices, which are objects that are specifically shared
And, finally, our system has to be compliant with between components. A similar implementation
existing services. Our system can use services example is the Eclipse IDE, which is also built
which have already been developed by a particular on an OSGi framework. Each Eclipse plug-in is
company. For practical purposes, we do not want an OSGi bundle, which is why we can easily add
to re-develop services already used for e-learning or remove plug-ins.
activities. To communicate easily with external services,
Thirdly, in order to present all services to the such as our context management system, we
user in a relevant way, we need adapt available used web services access (using a SOAP-based
services for user applications (see the section “Ad- implementation). To discover devices present in
aptation” below). Our system uses multi-channel the selling space, we use the UPnP (UPnP, 2007)
and multimodal intermediations between a mobile protocol. UPnP is a communication protocol al-
personal user device and a collection of e-services lowing the creation of spontaneous networks of
(Chevrin, Sockeel, & Derycke, 2006). devices (TV sets, HVAC, light control etc.) and
To satisfy these requirements we have chosen control points (PDA, Smartphone, touch panels,
the following technologies. The modular archi- etc.). UPnP enables live detection of devices and

Figure 3. Middleware architecture

1165
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

their use via the SOAP protocol, so the services ties embedded in a particular physical world (or
provided by devices are also seen as web ser- environment) are key issues to give us intention
vices. and meaning according to different situations and
As a consequence, we have to allow com- finally to determine the relevant features describ-
munication between different services from dif- ing the different situations. Many definitions of
ferent protocols, each with their own formalism. context are given in the literature, including:
Therefore we have implemented a dedicated “learning context is used to describe the current
communication middleware based on OSGi tech- situation of a person related to a learning activity;
nology, as shown in Figure 3. Our system is built in addition to attributes relying on the physical
by assembling services in an OSGi container. world model (Derntl & Hummel, 2005); “infor-
Each service can communicate with external mation and content in use to support a specific
components using web services. For example, the activity (being individual or collaborative) in a
context recovery service calls our context man- particular physical environment” (Kurti, Milrad,
ager system as a web service. Moreover, thanks Alserin, & Gustafsson, 2006).
to different OSGi bridges, physical devices can In our framework, the context model has to
be seen as UPnP devices. The gateway allows us deal with learning in context, seamless learning
to plug in any kind of device as long as a suitable across different contexts and context-as-construct.
OSGi driver is available. Each device company In other words, we have to manage the context
can provide such drivers, OSGi and UPnP being to provide the right learning support. According
open protocols. to these features, a context management model
must have the following properties: dynamic
and “context-as-construct”. Therefore, context
context modeling settings could be “unpredictable” because it is
dynamically created and modified through con-
In mobile learning, context may be viewed as tinual interactions with users and the environment.
“context-as-construct”, i.e. “context should be Context and activities are central to determine
reconceived as a construct that is continually learning goals and to choose how achieve them,
created by the interaction of learners, teach- more particularly in a working environment.
ers, physical settings, and social environments” According to these requirements and issues,
(Sharples, 2006) and “learning not only occurs our context management model is composed of
in the context, it also creates context through a context model, a context metadata schema,
continual interaction” (Balacheff, 2006). Dourish several context views and a set of situations. It is
(2004) outlines certain modeling characteristics organized as follows (cf. Figure 4).
of the context that need to be tailored for structur- There is a context model, defining a set of
ing contextual information: i) Contextuality is a relevant context dimensions and their features.
relational property that holds between objects or The context dimensions are divided into two
activities; ii) The context is not defined in advance. categories: abstract dimensions and atomic dimen-
On the contrary, the scope of contextual features is sions. An abstract dimension can be recursively
defined dynamically; iii) Context is an occasioned broken down into sub-dimensions which are
property, relevant to particular settings, particular either abstract or atomic. An atomic dimension
instances of activities; iv) Context arises from consists only of a set of features. Potentially, our
the activities. Context cannot be separated from context dimensions are the scenario (a hierarchal
activities. It is actively produced, maintained and task model having a task/method paradigm), the
enacted during activities. We claim that activi- user (an employee, a sales person, an expert,

1166
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

Figure 4. Context Management Model


Context
Management
Model

is composed of is composed of
Is composed of is composed of is composed of

is described by Context
Context Context
Context View contains utilizes
Dimension Medatdata Management

is managed by
is-a is-a utilizes
is-a
is-a is-a
is-a
Pedagogical Professional is-a
View View is-a is-a is-a
is-a
User Time Location Network
is-a is-a
is-a

Organization
Scenario Device Pedagogical Storage
unit Resource
Tool

Creation Change Modification

etc.) with sub-dimensions: the role, previous a given content category (methods, resources, or
knowledge, know-how, preferences, loyalty card, web services) and a given activity type (learning
purchase intentions, intention of use), the device, or working) for adaptation. An adaptation process
the location (office, shelf, stockroom, cash desk, does not manage the same features for different
etc.), the time, the learning tools, the network, content categories. In the adaptation process,
the physical environment and the resources (e.g. the current situation, filtered by a view and the
learning object, services, media resource, system corresponding content description are compared.
resource, etc.). The context metadata schema is Potentially, a view can also be created for a sub-
described by a set of properties (some of these category of content category or activity type (for
are shown in Table 1). Dimensions and features example, a sub-category of web services or learn-
are described by context metadata to deal with ing activities). The predicate contextViewFeatures
“unpredictable” context settings. The main idea described below in F-logic specifies all context
is to specify in advance for each type of context features for a view identified by a content category
property how to manage it for creation, change, (AC) and an activity type (AT). Context views are
and adaptation. In Table 1, the four displayed used to define different viewpoints for adaptation.
features are used to manage adaptation. Thus, Consequently, different adaptation categories are
we apply a specific adaptation policy according specified accordingly (see Box 1).
to the values of these features. A set of situations is organized in historical
There is also a set of views. A view consists of dependencies. A situation is a partial instantiation
a subset of context features which are relevant to of the context model consisting of the obtained

Table 1. A subset of context metadata properties


Some context metadata fea-
Description
tures
State {Mandatory, Transitory}.
ContentType {methods, resources, web services}
ActivityType {learning, professional, etc.}

1167
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

features describing the current learning or working learning across contexts. As a scenario describes
situation and its physical environment. It defines user activities, an author/designer can manage a
a complete context state associating all interac- global activity consistently. In this article, we focus
tions between the user and the learning system on the work-learning integration context given in
at the workplace in a given time interval. A user Figure 6 (described in detail later).
activity can be influenced by his previous work In pervasive learning systems, activities cannot
and learning activities. As soon as a new situation be achieved in the same way in different situations.
occurs, the next stage can be chosen according to It is necessary to have a context-aware and adaptive
the historical dependencies, for instance, to ensure mechanism to decide how to perform an activity
seamless learning. So far, we do not have enough according to a given situation. Derntl and Hummel
details from our experiments to define relevant address these needs by introducing a UML-based
strategies to manage seamless learning across modeling extension for including relationships
contexts. Future experiments will gives us more between context and learning activities in the
detail about this issue and also about the real set learning design models (Derntl & Hummel, 2005).
of features describing the context management Several research studies in artificial intelligence
model. At present, it is mainly a kind of generic focus on the hierarchical task model using the task/
context management model having an accurate method paradigm (Trichet & Tchounikine, 1999;
structure and associated principles to manage Wielinga, Velde, Schreiber, & Akkermans, 1992;
learning in context and seamless learning across Willamowski, Chevenet, & François, 1994). In a
contexts in our pervasive learning system, in other learning environment, hierarchical task models
words adaptation. Consequently, our context- were also used for designing, for instance, author-
aware scenario model is defined according to this ing tools (Ikeda, Seta, & Mizoguchi, 1997) and
context management model. learning systems (Betbeder & Tchounikine, 2003;
Choquet, Danna, Tchounikine, & Trichet, 1998;
Ullrich, 2005). The mechanism of hierarchical and
Context-Aware SCENARIO recursive decomposition of a problem into sub-
MODEL problems is one of the basic characteristics of the
hierarchical task model (Trichet & Tchounikine,
In our framework, the goal of scenarios is to de- 1999; Wielinga et al., 1992; Willamowski et al.,
scribe the learning, working and tutoring activities 1994). In the MODALES project, an adaptive and
to acquire some knowledge domain and know-how context-aware model of scenarios has been suc-
to solve a particular problem or to support working cessfully proposed and implemented. It is based
activities. The main role of a scenario model is on an interdisciplinary approach (didactics, phys-
to integrate mobile devices into broader learning ics and its epistemology, computer science and
and working scenarios, to combine formal (formal education) (Tetchueng, Garlatti, & Laubé, 2008).
classroom at the workplace) and work-learning This hierarchical task model has been reused and
integration and to enable us to manage seamless modified according to our new requirements.

Box 1.

FORALL AT, CC, ViewFeature contextViewFeatures(AT, CC, ViewFeature)


<- EXISTS cm cm:ContextMetadata AND cm[ActivityType->>AT; ContentType-
>>CC; ForContextFeature->>ViewFeature].

1168
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

The Task/Method Concept Learning Scenario for Sellers at the


Workplace
Within the framework of the Task/Method para-
digm, tasks represent activities and sub-activities In our framework, all activities (working and
managed by a knowledge-based system (cf. Figure learning) are integrated into a single scenario
5). A method describes how a particular task that is modeled by a hierarchical structure of
can be achieved. There are two types of tasks: tasks/methods (cf. Figure 6). The selection of
abstract tasks and atomic tasks. An atomic task the relevant method to achieve a given task ac-
is not composed of sub-tasks. It can be achieved cording to the current situation, restricted to the
by a simple procedure defined inside a method. relevant context view and the context descriptor,
An abstract task represents a high level activity are compared.
composed of sub-activities. A method defines Figure 6 shows a part of the e-retail scenario,
how an abstract task is composed of sub-tasks which represents the decomposition of the task
which can be either abstract or atomic. A method “S.3 – Sale assistance in situation” by a method
associated with an abstract task defines a control “M13” in a sequence order of sub-tasks (“S.3_T.1”
structure that allows the recursive decomposition and “S.3_T.2”). The task “S.3_T.1” can be achieved
of tasks into sub-tasks and the sub-task order at in two different situations: one without a customer
runtime by means of operators. At present, three and one with a customer. Thus, the decomposi-
different operators are used: sequence, alterna- tion of the task “S.3_T.1” can be made by two
tive and parallel. A method associated with an different methods (“M131”, “M132”). Each one is
atomic task can have: i) a resource specification associated to a set of contextual features (context
for resource retrieval; ii) a service specification descriptor), which specifies the relevant situations
for web service retrieval; iii) a procedure/function for which the method could be selected. Because
specification for a simple procedure or function. the task “S.3_T.1” is an abstract task, its methods
For a given task, several methods can be used to decompose it into sub-tasks by means of opera-
achieve it. Methods are described by contextual tors determining the sub-task execution order.
features for selection and adaptation. The method “M131” aims to provide to the seller

Figure 5. Context-aware task model description


Methods described by
contextual features (Web) Services/ resources
described by metadata
realized by Method1
Resource
Resource
Method1 requirement
requirement

Retrieval-
Retrieval-Matching
Retrieval-Matching
(Web)
Task
Task (Web)
Service
Method2
Method2 Service
requirement
requirement

function
function
Method3
Method3 specification
specification

Control
Control
SubTask1
SubTask1
Method4
Method4 structure
structure

Define a decomposition of a task into subtasks SubTask2


SubTask2

1169
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

Figure 6. A small part of the e-retail scenario


S.3
Legend: Sale assistance in situation
Context descriptor:
Task Method -Role = {Salesman}
M13 -Location = {Shop}
{SEQ(S.3_T.1, S.3_T.2)}
Context descriptor: (a seller without
customer) S.3_T.1 S.3_T.2
-Role = {Salesman} Preparing for the sale Finalizing the sale
-DeviceType = {PTA, Smartphone}
-NetworkType = {Wifi, Bluetooth, GMS} Context descriptor: (a seller with
Location = {Shelve}} customers)
-Role = {Salesman}
M131 M132 -CollaboratorRole = {Customer}
-DeviceType ={PTA, Smartphone}
-NetworkType = {Wifi, Bluetooth}
{PAR(S.3_T.1.1, S.3_T.1.2)}
{SEQ(S.3_T.1.3, PAR(S.3_T.1.4, S.3_T.1.2, S.3_T.1.6,
OPT(S.3_T.1.5)))}

S.3_T.1.1 S.3_T.1.3 S.3_T.1.4 S.3_T.1.5


S.3_T.1.2 S.3_T.1.6
Verify the Identify the Search and Get help
Revise the Consult to
information client deliver form
knowledge FAQ
about the information of colleagues or
labeling of of products products to the experts
products M1321 client M1324
M1323
M1312 M1322
M1311

with both activities available (“S.3_T.1.1 - verify systems. A review of current literature reveals
information about the labeling of products” and numerous ongoing efforts aimed at exploiting
“S.3_T.1.2 - revise the product knowledge”) when semantic web technologies for web service re-
s/he approaches shelves without customers (Loca- trieval and adaptation (Janssen, Lins, Schlegel,
tion = {Shelves}). The method “M132” achieves Kühner, & Wanner, 2004; Keidl & Kemper,
the task “S.3_T.1” for the seller with a customer 2004; Pathak, Koul, Caragea, & Honavar, 2005;
(CollaboratorRole = {Customer}). It is divided Sheshagiri, Sadeh, & Gandon, 2004). Semantic
into four sub-tasks that are carried out as fol- modeling of web services is a pre-requisite for
lows: (SEQ (S.3_T.1.3, PAR (S.3_T.1.4, S.3_T.1.5, successful service retrieval. Service descriptions
S.3_T.1.6))) means that we start with the sub-task should specify what the service offers, how the
“S.3_T.1.3” then follow with one of three sub-tasks service works and how to access the service (Yang,
(“S.3_T.1.4”, “S.3_T.1.5” and “S.3_T.1.6”) that will Lan & Chung, 2005). Work in this area utilises
be executed in parallel. WSDL+DAML-OIL (Sivashanmugam, Verma,
We shall use this part of scenario as an ex- Sheth, & Miller, 2003), WSDL-S (Akkiraju et al.,
ample. A task can be achieved by using web 2005) and OWL-S (Martin et al., 2004) among
services, so in the next section we will show how other technologies. Pathak et al. (2005) incor-
services are described and annotated for retrieval, porate OWL-S descriptions to describe service
orchestration and adaptation in the scope of the requests and providers that can be processed by a
task requirements. matchmaking engine that is aware of the relevant
domain ontologies. The framework supports
selecting services based on the user’s functional
SERVICE SPECIFICATIONS and and non-functional requirements, which are then
requirementS ranked based on user-specified criteria. Dolog et
al. (2004) propose a service-based architecture
In pervasive computing, service oriented archi- for establishing personalized e-Learning provided
tectures are increasingly used to design learning

1170
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

by various web-services, including personaliza- a selected product or a product type required


tion services and support services. Their work by the Service Requirement Specification
encompasses annotation schema and ontologies of the method “M132” is first carried out
for learning resources which may include web according to the current situation.
services, though these are not mentioned explicitly • Phase 2: All relevant web services of all
(Dolog, Henze, Nejdl, & Sintek, 2004). the suppliers are organized, invoked and
In our learning and working scenarios, an executed for searching for product informa-
activity may be realized by either a simple in- tion and resources.
ternal service or a set of external services. The • Phase 3: The found resources are assembled
problem to solve is to search for a service or a and adapted according to the current situa-
number of services that may be composed in a tion. They are filtered and sorted according
certain order to fulfill the requirements of the to their degree of relevance.
current task / method pair. Towards this goal, • Phase 4: The relevant resources are delivered
services have to be semantically annotated. In to the target devices. All services are man-
systems without context adaptation, web services aged and executed by our pervasive learning
are statically bound to a method at design time. system architecture previously described.
In this regard, the problem is to bind, invoke and
execute these web services in a “known” order.
Take for instance Task “S.3_T.1.4” in Figure 6. Web Service requirement
service A can be bound to this task (through its wspecification
method “M132”) at design time for searching for
product information and resources (images, sound The characteristics and functionalities of a service
messages, documents, etc.). At runtime, logic is required for a relevant task or activity have to be
built-in to web service A to search for relevant specified semantically to facilitate accurate and
resources annotated with description metadata efficient discovery and matching of the relevant
from repositories or from the database and deliver services. The primary goal of a service require-
them to the client making the request. ment is the description of how a service is to be
The proposed solution is semi-dynamic service “desired”. It is a request issued by the system
searching and matching. The relevant services wishing to interact with a service provider in
in this regard can be a single matched service or order that a task should be performed on behalf
a set of matched services. The latter is aimed at of the learner in the current situation. By our
searching for all the relevant services provided definition, a service satisfies a service requirement
by service providers according to the current by providing a set of desired output parameters
situation and the service requirement identified for a desired goal with a set of pre-existing input
in the selected method. This is done based on an parameters and situational context features. Our
adaptation process applied to services (Please refer proposed service requirement is summarized as
to the Context Management, Service Requirement follows:
Specification and Adaptation Process sections).
Taking Task “S.3_T.1.4” above for example, the • Functional requirements: describe the
realization solution consists of four phases. Each capabilities of web services desired by a
phase is considered as a single service or a set of user. It is characterized by input parameters,
matched services: expected output parameters, pre-conditions
and expected post-conditions;
• Phase 1: a search of web services (in a service • Non functional requirements: include
repository) which provide information about the identity of a service (e.g. name, owner,

1171
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

type etc.) as well as performance related concepts that covers a sub-domain (ontology of
characteristics, such as Quality of Service products) for query refinement. These concepts
(QoS), security etc; are used to “compare” with the DomainContent
• Content requirements: specify a list of category of a service (see Table 3) through the
domain concepts or a query identifying the domain ontology.
content requirements of services (e.g. Mark,
Price, ProductModel etc.). Service Description

Table 2 shows an example of a service require- Service requests and provision are modeled
ment specification “SRS_M1322” for the method with a common service description metadata
“M1322” which realizes the task “S.3_T.1.4 – model to formally specify the functional and
Search and deliver information on products to non-functional requirements of services. The
clients”. This specification consists of three main fundamental consideration in describing a web
parts. In the functional requirement part, the input service to support accurate and efficient service
is composed of three parameters (ProductType ?pt, discovery and matching is to fulfill a three-part
ProductInfoType ?pi and DeliveredResourceType ontology (Milanovic & Malek, 2004): function,
?rt) describing desired capabilities of the service behavior and interface. Interface dictates how the
for retrieving the desired resource (e.g. document, service can be invoked and what resources are to
media, voice file, pages, database, etc.). The pre- be assembled to provide the desired functionalities
condition is a predicate that verifies, for example a of the service. It is syntactic in nature. We propose
mandatory presence of a ProductType. The output to follow the WSDL binding standard for message
describes a list of relevant resources to be delivered format and protocol details for interoperability.
to the client. The non-functional requirements Hence, the problem of searching and matching the
portion represents QoS related context features right service provision to a client service require-
that can be used for ranking purposes. Finally, ment in a web service architecture is basically
the content requirement presents a list of domain reduced to a problem of matching the service

Table 2. An example illustrating the service requirement specification for the method “M132”

Service Request SRS_M1322

Functional
Requirement (
(ProductType ?pt)
input (ProductInfoType ?pi)
(DeliveredResourceType ?rt)
output (ProductResource ?pr)
(isNotNull ?pt)
pre-condition

p o s t- c o n d i-
tion (isNotNull ?rt)
)

1172
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

function and behavior descriptions to the service changes and demands of situated learning at run-
requirement specification. In this article, we pro- time. To this end, it detects the situation changes
pose a service description ontology to describe and then generates a new situation or updates
web services with a service description feature the current situation for maintaining activity
set. Table 3 enlists features which are generally relevance, seamlessly. This process is divided
applicable to most web services. into five main stages (see Figure 7):
These features can be extended to include
features which describe more specific function- • Stage 1: Context change detection and aggre-
alities of web services such as pedagogy. The gation: this stage determines context changes
features on a feature list can either be mandatory and checks whether these changes lead to
or optional. They serve to index a service for the creation of a new situation or an update
searching, filtering and ranking purposes. As of the current one. The changes can come
soon as a web service has been chosen according from collaborators (colleagues, customers,
to the service specification and requirements, it is tutors or learners), user interactions, loca-
achieved and managed by our pervasive learning tion, network, device, time, scenario, etc.;
system architecture. • Stage 2: Gathering of the initial context
information to create the new situation, this
is done by a partial or complete copy of the
CONTEXT MANAGEMENT last situation and/or by querying the context
PROCESS ontology;
• Stage 3: Selection of the current task: the last
The context management process integrates all task status can be restored in the new situa-
parts and models of the system in a permanent tion to ensure seamless learning and work-
way to deal with complex and dynamic context ing based on the historical dependencies of

Table 3. Service global feature set


Sub categories Feature set
General {name, description, language, owner, type (name, taxonomy, value), entityType}
{metadataCreator, metadataValidator, creationDate, validationDate, language, for-
Meta-metadata
mat}
{creator, dateCreated, version, status, contributor, publisher, dateUpdated, extentOfVa-
Life-cycle
lidity}
Right {IP, accessRight, signature, provenance, dateCreated, dateUpdated}
{URI, resource, resourceURI, resourceFormat, replacedBy, realisation, modeOfIn-
Technical
teraction}
{listInput(name, type, value, ontologyURI), listOutput(name, type, value, ontolo-
ServiceRequirement
gyURI), expectedEffect}
DomainContent {listDomainConcepts}
{r oleMo d el s , lo c a t io n (c o o r d i n a t e , s p a t i a l L o c a t io n , lo c a t io n Rel a t iv-
it y), physica l(del ive r y C h a n nel, del ive r ySyst e m , dev iceMo del, t o ol),
Context
informaticResource(hardware, software), temporal(temporalCoverage, frequencyRe-
quirement)}

1173
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

situations. When the last task is finished, the ADAPTATION


next one is selected according to the current
scenario and the historical dependencies of The fundamental issue in a pervasive learning
situations. When the last task is not finished, environment is how to provide learners with the
it is necessary to continue it; right activities, the right learning content at the
• Stage 4: Searching for the relevant couples right time and in the right way. Thus, adaptation
(Task, Method): According to the current is mandatory to all types of learning activities
situation and the task, the adaptation process in pervasive learning environments (Bomsdorf,
has to select the relevant methods. The main 2005). At present, we focus on adaptation mecha-
role of the adaptation process is to maintain nisms dedicated to two content categories: method
consistency between the learning system, and web service adaptation. At the scenario level,
the physical environment and the current adaptation is aimed at accomplishing an activity
learning or working situation and also to according to the current situation, or in other
ensure seamless learning and working; words, how to select the relevant methods for a
• Stage 5: Discovery and execution of relevant given task. The learning system has to dynamically
web services to serve the current task: Re- select the relevant way to perform the different
alization of a task requires the discovery, tasks of a scenario. At the service level, adapta-
orchestration, invocation and execution of tion has to refine the service retrieval process
relevant web services. Service requirement for perform an atomic task. Thus, web services
specification which specifies the required are only available in atomic tasks. The learning
service functionalities and characteristics is system needs to select the relevant web services
defined in the selected method of the selected according to the current couple (task, method)
Task/Method pair in Stage 4. The adaptation and the current situation. As context is dynamic,
process is hence aimed at searching for the it is not possible to know in advance how the
relevant web services to realize the current next situation will be structured. In other words,
task according to this specification and the it is not possible to anticipate the set of features
current situation. The specification and composing the different situations. For managing
management of web services is described adaptation, it is necessary to trigger rules using
in the next section. the contextual features of methods and the service

Figure 7. The context management process


System
Learning context Context management process
components

User,
(3)
Collaborator
restoring
Middlewares or other

Location (5)
Capture modules

(2) Gathering the current


searching
Applications /

Network (1) Detection initial context task


for the
systems

Device
of context information relevant
changes for creating a services/
Time new situation (4) searching for resources
Tool
the relevant
couples (task,
method)

1174
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

description of web services which are defined a Adaptation Process


priori and the set of situation features which can
be unexpected – not known in advance. Conse- The adaptation process is specified for a content
quently, it is not possible to define rules for each type (methods and web services) and an adapta-
possible configuration of features in a situation. tion category. The input content of the adapta-
Like the Mobilearn European project (Lons- tion process can be achieved in different ways
dale & Beale, 2004), we associate specific meta- depending on the content type: input methods
data with situation properties and/or dimensions are specified directly by the current task while
for managing adaptation. Situation features are input services are selected by a search process
divided into two categories: permanent and tran- based on a query which “compares” the service
sitory. Features describing scenarios and users requirement of the selected couple (task, method)
which are available in all situations are perma- with the service description. The three stages of
nent features in a situation. Thus, it is possible to the adaptation process are presented as follows
manage them as usual because they are known (cf. Figure 8):
a priori. Other features are transitory. For them,
it is necessary to analyse how each property • Evaluation/Classification: input content is
contributes to content adaptation – methods and classified according to the current situation
services selection - differently according to its in several equivalence classes: two classes
role in the adaptation process. Some of them are {“Good”, “Bad”} for each transitory feature
used to filter the content while others are used to and up to five equivalence classes {“Very-
rank or annotate it where filtering, ranking and Bad”, “Bad”, “ToConsider”, “Good”,
annotation are the adaptation techniques. For “VeryGood”} for all permanent situation
instance, learning and working methods can be features, together. The content belongs to an
filtered according to learning places or used de- equivalent class if it satisfies its membership
vices while it is annotated according to the user’s rules.
knowledge or preferences. As soon as features • Filtering: all content belonging to “Bad”
are used to filter content (methods or services), it classes are filtered out. In other words, these
is easy to manage them. In other words, content content are discarded. For example, with
will or will not be filtered out by a present transi- the network dimension, the class “Good”
tory feature. At present, all transitory features of is considered as relevant while the class
content are used to filter. Nevertheless, we shall “Bad” is not. This means that the system
have to investigate this issue in depth according hence will eliminate all content that belongs
to more detailed scenarios in future. to the class “Bad”.

Figure 8. Adaptation principle


Metadata/ indexing
(Context descriptor/
Service description)

Classified content Relevant content


content:
(Methods Evaluation
Filtering Adaptive
or Classification navigation
services)

{« Good »- « Bad »}
situation situation
(with transitory features) (with permanent features)

1175
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

• Adaptive navigation: permanent situation class “Good”, can be provided to the user. The
features are used to evaluate and classify the user can choose one of the relevant methods to
remaining content. An adaptive technique carry out the current task.
can be chosen by the system or by the user Table 4 shows an example of two membership
according to an author decision. All content rules described in a reduced form using F-Logic
belonging to the same equivalence class are syntax. The first rule EvaluateBadMethod verifies
treated in the same way. Annotation and if a method belongs to the class “Bad” according
sorting are processed according to the total to a situation. It is done based on the mismatching
order of equivalence classes. For hiding, only between the method’s context descriptor and con-
contents belonging to the class “Good” and textual features of a situation. The mismatching of
“VeryGood” are maintained. each contextual feature is verified by each sub rule.
The second rule EvaluateBadLocationForMethod
Adaptation of Methods is a sub rule that verifies the mismatching of the
current situation’s location feature and the corre-
Method adaptation is specified as follows: 1) All sponding the supported location list of a method.
context dimensions for this adaptation category The predicate contextViewFeatures verifies if
possess the same set of equivalent classes {“Good”, the location feature is one context view feature
“Bad”}; 2) All methods for which the current for methods. For the filtering, all methods which
context and the user features match up to the do not belong to class “Bad” are considered as
corresponding method’s contextual features (or relevant methods (see the rule MethodFiltering
“belong to” for multiple-valued features) belong in Table 4).
to the class “Good” and others belong to the class
“Bad”; 3) If the class “Good” is empty, it is con- Adaptation of Services
sidered as a problematic situation and required a
designer action to remediate or to provide a new Firstly, a service retrieval process is carried out
method and context adapted to the user and the to search relevant services in the repository based
task. Otherwise, all methods, belonging to the on the Service description to fulfill the Service

Table 4. Illustration of a membership rule of the equivalent class “Bad” for methods

RULE EvaluateBadMethod:
FORALL aMethod, aSituation belongingToBad (aMethod, aSituation)
<- EXISTS aCxtDescriptor (aMethod[has_Context_Descriptor->> aCxtDescriptor] AND (
IsMismatchedLocation(aSituation->>hasLocationPlace, aCxtDescriptor->>Location) OR
IsMismatchedNetwork(aSituation->>hasCurrentNetwork, aCxtDescriptor->>hasNextworkTypes) OR
IsMismatchedDevice(aSituation->>hasUsedDevice, aCxtDescriptor->>hasDeviceType) OR
IsMismatchedCollaborator(aSituation->>hasCollaborator, aCxtDescriptor->>hasCollaboratorRole) OR
IsMismatchedRole(aSituation->>hasCurrentUser->>hasRole, aCxtDescriptor->>hasRole) OR
IsMismatchedKnowHow(aSituation->>hasCurrentUser, aCxtDescriptor->>hasPrerequisiteKnowHow))).
RULE EvaluateBadLocationForMethod:
FORALL aLocation1, aListLocation2 IsMismatchedLocation (aLocation1, aListLocation2)
<- NOT inList(aListLocation2, aLocation1) OR NOT contextViewFeatures(“Method”, “Learning”, getCorrespondingCon
textFeature(aLocation1)).
RULE MethodFiltering:
FORALL aMethod, aSituation isRelevant(aMethods, aSituation) <- aMethod:Methods AND NOT
belongingToBad(aMethod, aSituation).

1176
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

request for a task. This is done by querying the CONCLUSION


Service descriptor metadata repository (match-
ing particularly the ServiceRequirement and We have proposed an adaptive and context-aware
DomainContent of Services with the functional, model of a scenario based on a hierarchical task
non-functional and content requirements of a model having the task/method paradigm - with
request). Secondly, the relevant Services serve as methods defining how to achieve a task - for a
the input for the adaptation phase which is aimed pervasive learning system supporting working
at refining the relevant services according to the and learning situations. This model enables us
current situation and user’s interest. Adaptation to include contextual information in learning
of services is specified as follows: 1) all transi- scenarios at the design stage and to choose how
tory context features in the Context category are to achieve activities according to the current
used to classify input services into two equivalent situation at runtime. In other words, the relevant
classes {“Good”, “Bad”}; 2) all services belong to methods are selected dynamically according to
class “Bad” will be eliminated; 3) some context the current situation for performing activities. We
features from the Context, Quality, Financial, etc. have also integrated web services described at
categories are taken into account for the service the semantic level for indexing services. From a
ranking process. This process is based on user scenario perspective, the system can dynamically
interest features. For example, users can choose select and adapt learning components (resources
qualityRating of services as with high priority and services) that are not known in advance for
while others may be more interested in perfor- achieving an atomic task. Our context model and
mance. Moreover, ranking requirements also adaptation process deal with dynamic “context-as-
depend on the type of services. With services for construct” by means of transitory and permanent
voice communications, the performance is very situation features managed in different ways. A
important while with services for financial trans- service oriented architecture approach is suitable
actions, security is a high priority. Therefore, the for pervasive learning systems to deal with such
user can build an “interested” feature priority list dynamic learning content and environments. At
by annotating each feature with an “interest” level present, because our scenarios are limited, issues
(from very low to very high). This information of seamless learning and working across contexts
will be stored in the user’s profile for later reuse are managed in a limited way. The integration of
and be part of the current situation information for scenarios and learning resources is not actually
this user. Based on the list of “interest” features, taken into account. As an area for future work,
the system can rank all matched services. we will enrich our scenarios and from that, we
can study in greater depth the transitory and per-

Figure 9. Service retrieval and adaptation

Metadata/ Metadata/
Adaptation
Service description Service description Classified
services Relevant
Matched services
services
service
requirement Matching Evaluation/
Classification Filtering Ranking
specification

Service Invocation,
Orchestration, Choreography
Domain ontology current situation and Execution Subsystem
current situation
(interested features)

1177
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

manent situation features and adaptation policies international conference on World Wide Web.
needed to manage different adaptation categories
Chevrin, V., Sockeel, S., & Derycke, A. (2006,
and seamless learning. Our model will lead to
June 26-29). An Intermediation Middleware for
experiments in the scope of the p-LearNet project
supporting Ubiquitous Interaction in Mobile
with our industrial partners to be evaluated in
Commerce. Paper presented at the ICPS’06, IEEE
real situations.
International Conference on Pervasive Services,
Lyon.
ACKNOWLEDGMENT Choquet, C., Danna, F., Tchounikine, P., & F.
Trichet. (1998). Modelling Knowledge-Based
This work is supported by the p-LearNet project Components of a Learning Environment within
funded by the ANR “Agence Nationale de la the Task/Method Paradigm. Paper presented at
Recherche” in France. It receives the support of the Intelligent tutoring Systems, ITS’98, San
the PICOM “Pôle de Compétitivité des Industries Antonio (USA).
du Commerce”.
Derntl, M., & Hummel, K. A. (2005). Modeling
Context-Aware E-learning Scenarios. Paper
presented at the Pervasive Computing and Com-
REFERENCES
munications Workshops. 3rd IEEE PerCom 2005
Workshops.
Akkiraju, R., Farrel, J., Miller, J., Nagarajan, M.,
Marc-Thomas, Schmidt, et al. (2005). Web Service Derycke, A., Chevrin, V., & Vantroys, T. (2007,
Semantics - WSDL-S. Retrieved 15 November, 13th May). P-Learning and e-retail: a case and
2008, from http://www.w3.org/Submission/2005/ a flexible Sotware Architecture. Paper presented
SUBM-WSDL-S-20051107/ at the Pervasive Learning 2007. An International
Workshop on Pervasive Learning, Toronta, On-
Balacheff, N. (2006). 10 issues to think about the
tario, Canada.
future of research on TEL. Les Cahiers Leibniz,
Kaleidoscope Research Report (147). Derycke, A., Vantroys, T., Barbry, B., & Laporte,
P. (2008). E-retail: Interaction of Intelligent Sell-
Betbeder, M.-L., & Tchounikine, P. (2003).
ing Space with Personal Selling Assistant. Paper
Structuring collective activities with tasks and
presented at the ICEIS 2008, an International
plans. Paper presented at the IEEE International
Conference on Enterprise Information System,
Conference on Advanced Learning Technologies
Barcelona, Spain.
(ICALT’2003).
Dolog, P., Henze, N., Nejdl, W., & Sintek, M.
Bomsdorf, B. (2005). Adaptation of Learning
(2004). Personalization in distributed e-learning
Spaces: Supporting Ubiquitous Learning in
environments. Paper presented at the Proceedings
Higher Distance Education. Paper presented at
of the 13th international World Wide Web confer-
the Mobile Computing and Ambient Intelligence:
ence on Alternate track papers & posters.
The Challenge of Multimedia, Dagstuhl Seminar
Proceedings. Dourish, P. (2004). What we talk about when we
talk about context. Personal Ubiquitous Comput.,
Brodersen, C., Christensen, B. G., Gronboek, K.,
8(1), 19-30.
Dindler, C., & Sundararajah, B. (2005). eBag: a
ubiquitous Web infrastructure for nomadic learn- Farmer, J., Lindstaedt, S. N., Droschl, G., & Lut-
ing. Paper presented at the Proceedings of the 14th tenberger, P. (2004, April 2-3). AD-HOC - Work-

1178
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

integrated Technology-supported Teaching and Issues and challenges in ubiquitous computing,


Learning. Paper presented at the 5th conference 45(12).
on Organisational Knowledge, Learning and
Martin, D., Burstein, M., Hobbs, J., Lassila, O.,
Capabilities, Innsbruck, Austria.
McDermott, D., McIlraith, S., et al. (2004). OWL-
Hundebol, J., & Helms, N. H. (2006). Pervasive S: Semantic Markup for Web Services. Retrieved
e-Learning - In Situ Leaning in Changing Con- 15 November, 2008, from http://www.w3.org/
texts. Submission/OWL-S
Ikeda, M., Seta, K., & Mizoguchi, R. (1997). Task Michael-Spector, J., & Wang, X. (2002). Inte-
Ontology Makes It Easier To Use Authoring Tools. grating Technology into Learning and Working:
Paper presented at the IJCAI. Promising Opportunities and Problematic Issues.
Educational Technology & Society, 5(1).
Janssen, D., Lins, A., Schlegel, T., Kühner, M., &
Wanner, G. (2004). A Framework for Semantic Milanovic, N., & Malek, M. (2004). Current Solu-
Web Service Retrieval. Fraunhofer Institute for tions for Web Service Composition. Berlin.
Industrial Engineering and University of Applied
Oppermann, R., & Specht, M. (2006). Situated
Sciences, Schellingstr.
Learning in the Process of Work. In S. Nether-
Jones, V., & Jo, J. H. (2004). Ubiquitous Learn- lands (Ed.), Engaged Learning with Emerging
ing Environment: an Adaptive Teaching System Technologies (pp. 69-89).
using Ubiquitous Technology. Paper presented at
OSGi. (2007). OSGi Alliance. from http://www.
the ASCILITE, Perth, Australia.
osgi.org
Keidl, M., & Kemper, A. (2004). Towards Context-
p-LearNet. (2006). p-LearNet project. Retrieved
Aware Adaptable Web Services. Paper presented
from http://p-learnet.univ-lille1.fr/web/guest/
at the In Proceedings of the 13th International
home.
World Wide Web Conference (WWW’04).
Pathak, J., Koul, N., Caragea, D., & Honavar, V.
Kurti, A., Milrad, M., Alserin, F., & Gustafsson,
G. (2005). A framework for semantic web serv-
J. (2006, October 4-6). Designing and implement-
ices discovery. Paper presented at the Workshop
ing ubuquitous learning activities supported
On Web Information And Data Management.
by mobile and positioning technologies. Paper
Proceedings of the 7th annual ACM international
presented at the the Ninth IASTED International
workshop on Web information and data manage-
Conference computers and Advanced Technology
ment, Bremen, Germany
in Education, Lima, Peru.
Sharples, M. (2005). Learning As Conversation:
Laine, T. H., & Joy, M. (2008). Survey on Context-
Transforming Education in the Mobile Age. In
aware Pervasive Learning Environments. Paper
Proceedings of Conference on Seeing, Under-
presented at the Mlearn 2008, the bridge from
standing, Learning in the Mobile Age (pp. 147-152),
text to Context, Wolverhampton.
Budapest, Hungary.
Lonsdale, P., & Beale, R. (2004). Towards a dy-
Sharples, M. (2006). Big Issues in Mobile Learn-
namic process model of context. Proceedings of
ing: Report of a workshop by the Kaleidoscope
Ubicomp 2004 workshop on Advanced Context
Network of Excellence Mobile Learning Initiative.
Modeling, Reasoning and Manage.
LSRI, University of Nottingham.
Lyytinen, K., & Yoo, Y. (2002). Issues and chal-
lenges in ubiquitous computing. SPECIAL ISSUE:

1179
An Adaptive and Context-Aware Scenario Model Based on a Web Service Architecture

Sheshagiri, M., Sadeh, N. M., & Gandon, F. (2004). Tasks and Methods. International Journal of
Using Semantic Web Services for Context-Aware Expert Systems With Applications, Elsevier Sci-
Mobile Applications. Paper presented at the Mo- ence, 16, 105-120.
biSys 2004 Workshop on Context Awareness.
Ullrich, C. (2005). Course generation based on
Simon, A. (2007). Blurring the lines between HTN planning. In J. A. & B. B. (Eds.), Proceed-
learning and working. Yellow Edge, The Perfor- ings of 13th annual Workshop of the SIG Adap-
mance Architect. Retrieved One, from http://www. tivity and User Modeling in Interactive Systems
yellowedge.com.au/Downloads/Issue%201%20 (pp. 75-79).
-%20Blurring%20the%20lines%20between%20
UPnP. (2007). UPnP, Official web site. from http://
learning%20and%20work.pdf
www.upnp.org
Thomas, S. (2007, May 13). Pervasive Scale:
Vavoula, G. N., & Sharples, M. (2008). Challenges
A model of pervasive, ubiquitous, and ambient
in Evaluating Mobile Learning. Paper presented
learning. Paper presented at the An International
at the mLearn 2008, the international conference
Workshop on Pervasive Learning, in conjunction
on mobile learning, University of Wolverhampton,
with Pervasive 2007, Toronto, Ontario, Canada.
School of Computing and IT, UK.
Sivashanmugam, K., Verma, K., Sheth, A. P.,
Wielinga, B., Velde, W. V. d., Schreiber, G., &
& Miller, J. A. (2003). Adding semantics to web
Akkermans, H. (1992). The KADS Knowledge
services standards. Paper presented at the Pro-
Modelling Approach. In R. Mizoguchi & H.
ceedings of the International Conference on Web
Motoda (Eds.), Proceedings of the 2nd Japanese
Services, ICWS ‘03, CSREA Press.
Knowledge Acquisition for Knowledge-Based
Tetchueng, J. L., Garlatti, S., & Laubé, S. (2008). A Systems Workshop (pp. 23-42). Hitachi, Ad-
Context-Aware Learning System based on generic vanced Research Laboratory, Hatoyama, Saitama,
scenarios and the theory in didactic anthropology Japan.
of knowledge. International Journal of Compu-
Willamowski, J., Chevenet, F., & François, J.
ter Science and Applications, 5(Special Issue on
M. (1994). A development shell for cooperative
New Trends on AI Techniques for Educational
problem-solving environments. Mathematics and
Technologies), 71-87.
computers in simulation, 36(4-6), 361-379.
Trichet, F., & Tchounikine, P. (1999). DSTM:
Yang, S, Lan, B., & Chung, J. (2005, 29 March-1
a Framework to Operationalize and Refine a
April 2005). A new approach for context aware
Problem-Solving Method modeled in terms of
SOA. Paper presented at the The 2005 IEEE
International Conference on e-Technology,
e-Commerce and e-Service, 2005. EEE ‘05.
Proceedings.

This work was previously published in International Journal of Mobile and Blended Learning, Vol. 1, Issue 3, edited by D.
Parsons, pp. 41-69 , copyright 2009 by IGI Publishing (an imprint of IGI Global).

1180
1181

Chapter 4.17
Exploring the Effects of
Web-Enabled Self-Regulated
Learning and Online Class
Frequency on Students’
Computing Skills in Blended
Learning Courses
Pei-Di Shen
Ming Chuan University, Taiwan

Chia-Wen Tsai
Ming Chuan University, Taiwan

ABSTRACT (SRL) to enhance students’ skills of deploying


database management system (DBMS). Four class
Web-based courses have shown to be successful sections with a total of 172 second-year students
in providing quality distance education. However, were taken as four distinct groups. The results
due to a national education policy, pure online showed that students in the SRL and BL groups
courses are not permitted in Taiwan. In addi- with 5 online classes had the highest grades for
tion, there exists a lack of appropriate design using DBMS among the four groups. Students who
and delivery of blended learning courses. In this received the treatments of web-enabled SRL also
study, the authors conducted a quasi-experiment outperformed a control group that did not have the
to examine the effects in applying blended learn- benefit of instruction in SRL. The implications
ing (BL) with web-enabled self-regulated learning of this study are also discussed.

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

INTRODUCTION Internet, it is essential to investigate and develop


an appropriate design and arrangement of BL
The goals of vocational schools concentrate on courses for schools and teachers. For example,
developing a highly skilled workforce (Lee & what frequency of online classes in a BL course
Huang, 1996). Professionals with a vocational is more appropriate to the students, particularly
degree represent a major portion of the work force for those with low self-regulatory skills? The
in Taiwan (Shen, Lee, Tsai, 2007a). However, authors conducted an experiment to explore the
vocational education in Taiwan is highly com- appropriate online class frequency that supports
petitive in that it must attract sufficient student student learning.
enrollments in the face of a continually decreas- Through the Internet, learners are free to
ing birth rate and rapidly increasing number of access new information without restrictions (Li,
schools. Schools, facing the high pressure of Tsai & Tsai, 2008); however, this may also be
market competition, often emphasize the propor- one of its greatest dangers. There is a continuing
tion of students awarded certificates before they debate about the effectiveness of online learning
graduate. That is, teaching in this sector usually environment designs (Azevedo, 2005; Jacobson,
focuses on helping vocational students to pass 2005). Online learning differs from didactic pre-
the certification examinations (Shen, Lee & Tsai, sentation, where the student has few opportunities
2007). The grades on students’ certificates and to deviate from the teacher’s presentation of the
the numbers achieved are the main criteria to material (Greene & Azevedo, 2007). Moreover,
evaluate teachers’ teaching and students’ learn- it is indicated that vocational students are more
ing. In this regard, how to help students enhance Internet-addicted than students in general (Yang
their professional skills and pass the certificate & Tung, 2007). Many vocational students are ad-
examinations is a major concern to many teachers dicted to shopping websites, online games, and on-
in vocational schools in Taiwan. line messengers, and prefer this rather than getting
Web-assisted instruction has been advocated involved in courses, particularly online courses
by contemporary educators and researchers (Liu (Shen, Lee, Tsai & Ting, 2008). This addiction to
& Tsai, 2008). Asynchronous, web-based edu- the Internet and the lack of on-the-spot teacher
cational programs have been shown to be quite monitoring in web-based instruction makes it
successful in providing quality distance education even more difficult for students to concentrate
(Overbaugh & Casiello, 2008). However, the policy on online learning (Shen, Lee & Tsai, 2008). To
of e-learning in Taiwan is relatively conservative respond to this challenge, the authors adopt self-
in contrast with that in the U.S. For example, regulated learning (SRL) that can help students
earning an academic degree entirely through better regulate and improve their learning.
online courses is still not allowed at present. That As more and more institutions of higher educa-
is, teachers in some nations with conservative tion provide online courses, the question arises
institutions and implementations of e-learning, whether they can be as effective as those offered in
have to adopt a mode of blended learning (BL) the traditional classroom format (Shelley, Swartz
rather than pure online learning when imple- & Cole, 2007). However, few studies have dis-
menting e-learning. The effectiveness of BL has cussed effective online instructional methods for
already been demonstrated (Liu, Chiang & Huang, vocational students (Shen, Lee, and Tsai, 2007a).
2007; Pereira, Pleguezuelos, Merí, Molina-Ros, Furthermore, we expect that innovative teaching
Molina-Tomás & Masdeu, 2007; Shen, Lee & Tsai, methods and technologies could improve students
2007b), nevertheless, due to limited research on learning in BL courses. Specifically, this study
how BL can be conducted effectively using the explores the potential effects of web-based SRL

1182
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

with variations in online class frequency on the Moreover, Nota, Soresi and Zimmerman (2004)
development of vocational students’ computing indicate that the cognitive self-regulation strategy
skills. Based on suggestions from earlier research, of organizing and transforming proves to be a
we have re-designed the course and conducted a significant predictor of the students’ course grades
series of quasi-experiments to examine the effects in mathematics and technical subjects in high
of web-enabled SRL, varying with frequency of school, their subsequent average course grades
online classes on vocational students’ comput- and examinations passed at university.
ing skills. Previous studies have established that self-
regulation skills can help foster learning from
any instructional method (see Ertmer, Newby &
SELF-REGULATED LEARNING MacDougall, 1996; Weinstein, 1989; Zimmerman,
1990). In addition, many educators and teach-
Zimmerman and Schunk (1989) define SRL in ers recognize the importance of SRL in online
terms of self-generated thoughts, feelings, and learning environments. In Lynch and Dembo’s
actions, which are systematically oriented toward (2004) study that investigates the relationship
attainment of the students’ own goals. SRL is an between self-regulation and online learning in a
active, constructive process whereby learners BL context, it is indicated that verbal ability and
set goals for their learning and then attempt to self-efficacy related significantly to performance,
monitor, regulate, and control their cognition, together explaining 12 percent of the variance in
motivation, and behavior in the service of those course grades. Moreover, it is indicated that suc-
goals (Winne, 2001; Winne & Hadwin, 1998; cessful students in an online course generally used
Zimmerman & Schunk, 2001). Characteristics SRL strategies and the effect of self-regulation
attributed to self-regulated persons coincide on students’ success was statistically significant
with those attributed to high-performance, high- (Yukselturk & Bulut, 2007). Therefore, in this
capacity students, as opposed to those with low study, SRL is applied in this BL course to enhance
performance, who show a deficit in metacognitive, students’ computing skills.
motivational, and behavioral variables (Montalvo
& Torres, 2004; Reyero & Tourón, 2003; Roces &
González Torres, 1998; Zimmerman, 1998). BLENDED LEARNING
In an SRL environment, students take charge of
their own learning by choosing and setting goals, Blended learning (BL) is a form of technology-
using individual strategies in order to monitor, reg- mediated learning that improves learning out-
ulate and control the different aspects influencing comes through an alternation of face-to-face
the learning process and evaluating their actions. courses and Internet courses (Lai, Lee, Yeh &
Eventually, they become less dependent on others Ho, 2005). Marino (2000) discovered that some
and on the contextual features in a learning situa- students experienced difficulty adjusting to the
tion (Järvelä, Näykki, Laru & Luokkanen, 2007). structure of online courses, managing their
Bielaczyc, Pirolli and Brown (1995) incorporated time in such environments, and maintaining
self-explanation and self-regulation strategies in self-motivation. Students may feel frustration in
the attainment of the cognitive skill of computer fully online courses, particularly those who are
programming. They found that their treatment dependent learners, are less self-regulated, and
group, which incorporated self-regulation strate- need frequent direction and reinforcement from
gies, outperformed a control group that did not a visible instructor. These frustrations could be
have the benefit of instruction in these strategies. eased when the online course is combined with

1183
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

periodic opportunities for face-to-face interac- pre-test confirmed that all participants had aver-
tions (Rovai & Jordan, 2004). age or little knowledge of the course content. In
With regard to the effects of BL in previous addition, there was no student who had taken a
research, Yushau (2006) shows the positive ef- web-based course before. We evenly and randomly
fect of blended e-learning on students’ attitude divided the students into the four experimental
toward computers and mathematics. It is found groups.
that performance, as measured by the final mark
of the course under a hybrid teaching method that Course Setting
incorporated both traditional face-to-face lec-
tures and electronic delivery and communication The course is a semester-long, 2 credit-hour
methods, is higher than that of using a traditional class, targeting second-year college students
teaching method alone (Dowling, Godfrey & from different major fields of study. The major
Gyles, 2003). In Castelijn and Janssen’s (2006) focus of this course is to develop students’ skills
study, their statistical results also indicated that in applying the functions of a database manage-
BL students had higher exam scores in a financial ment system (DBMS), which is a powerful tool
management course. for creating and managing large amounts of data
As for the effects of BL on student success efficiently, robustly and safely, over long periods
rates in learning to program, Boyle, Bradley, of time (Матросова, 2007). Students received a
Chalk, Jones, and Pickard’s (2003) research results study task dealing with the subject of Microsoft
indicate a generally positive evaluation of the Access, one of the popular DBMS applications.
main elements of the blend, and widespread use Further, according to the school’s policy, the course
of the new online features. Moreover, students in targets helping students to earn a certificate in
the BL group attained significantly higher aver- database applications. That is, students have to
age scores than those in the traditional teaching take a certification test in Access at the end of
group. Similarly, Pereira et al. (2007) concluded the semester.
that BL was more effective than traditional teach-
ing. Therefore, BL is applied in this study to help Experimental Design and
students learn and develop their skills in using Procedure
application software.
The experimental design is a factorial pretest -
post test design. Four classes were selected from
METHODS two successive years for this quasi-experiment,
with three classes chosen for BL classes and the
Subjects last one used as a control group. In the first week,
the lecturer declared in the three BL classes that
The subjects in this study were 172 vocational the class section would be partially provided with
students from two consecutive semesters taking innovative instructional methods mediated on the
a compulsory course entitled ‘Database Manage- web as an intervention. Students had the freedom
ment Systems’. Students at this university are to drop this class section and take another teacher’s
expected to spend much more time and effort in class section, if preferred. After this declaration,
mastering a variety of technological skills when 172 students continued in the three BL classes.
compared to those in comprehensive universities In the first week, students were pre-tested and
in Taiwan. None of students in this study majors the results showed that the differences of students’
in information or computer technology, and the computing skills among the four groups were

1184
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

not statistically significant. That is, students in two-hour lecture was delivered discussing how
the four groups had similar levels of computing students could manage study time and regulate
skills before they received the interventions. Then, their learning. The content of this SRL course
participants were randomly assigned to one of the was composed of the four processes addressed
four experimental conditions. The ‘SRL and BL by Zimmerman, Bonner & Kovach (1996), that
with 10 online classes’ group (C1, n=44), ‘SRL is, self-evaluation and monitoring, goal-setting
and BL with 5 online classes’ group (C2, n=41), and strategy planning, strategy implementation
‘non-SRL and BL with 5 online classes’ group and monitoring, and monitoring of the outcome
(C3, n=42) are experimental groups, while ‘non- of strategy. Students were taught how to imple-
SRL and face-to-face’ group (C4, n=45) is the ment these four processes to become more self-
control group. C2 and C3 were conducted in the regulated learners.
first semester, while C1 and C4 were conducted In addition to the two-hour lecture, students in
in the second. SRL groups were required to regularly preview
This experiment was implemented in the and review the textbook, and practice the skills in
‘Database Management Systems’ course. Stu- using DBMS. The teacher assigned course work to
dents needed to pass the examination to earn students in the Assignments and Exercises section
the Microsoft Access certificate. The certificate of the course website and students had to com-
examinations were held immediately after the plete the assignments by the required deadline.
completion of teaching the course (16th week They were also required to record their learning
of the semester). The detailed schedule of the behavior on the course website every week. The
experiment is depicted in Figure 1. button for submission became unavailable when
the time was up. Activities and results of students’
Intervention Concerning SRL learning were captured and recorded on the course
website instead of in their notebooks, to prevent
Students in C1 and C2 received instruction in falsification of records.
an after-school class teaching SRL strategies.
The students were gathered in a classroom and a

Figure 1. The schedule of the course and skill tests in one semester

Database Management Systems Course (16 weeks)

Week 1: Week 2: Week 6: Week 9: Week 13: Week 16:


Students in C1 Teacher checked The mid-term The quiz was
are divided and C2 students’ adaptation exam was administered. exam was
into 4 groups received SRL to the web-based administered. administered.
and learning and
pretested. progress.

1185
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

Intervention Concerning BL Differences in students’ skills of using DBMS at


the beginning stage among the four classes were
Networked multimedia technologies and software not statistically significant. Thus, the researchers
were applied in the BL classes (C1, C2 and C3). ruled out initial differences as a plausible alterna-
A course website was provided for BL students. tive explanation for the differences detected after
The teacher lectured about how to solve simu- treatments (Gribbons & Herman, 1997).
lated computing problems through the Internet The examinations for the certificate in Access
or in the classroom. The teacher recorded every were conducted immediately after the course
lecture session whether in the classroom or via concluded. There were two problems, each consist-
the Internet and later on translated lectures into ing of 5 to 7 sub-problems in the examinations.
HTML files with flash, video, and voice. These Before testing, students were assigned random
HTML files were then loaded onto the course seats. All students were tested at the same time.
website. Students could then preview and review A student’s grade came from their correctness
the course sessions on this course website. They and completeness in problem solving. A student
could also download the examples and upload could get professional certification using DBMS
their homework to the site. if his/her grade was higher than 70. Finally, we
From the third week, some course work to be tested the differences between students’ skills in
delivered online, that is, five or ten non-contiguous using Access under different conditions.
weeks in total, was moved onto the website. At
the beginning of the semester, the teacher urged Design of the Course Website
students to adjust their learning gradually and
smoothly. The remaining weeks’ teaching was An open-source Learning Management System
still conducted in the traditional classroom. In (LMS), Moodle, was adopted as the platform for
the set of face-to-face classes, the teacher gave the course website in this study. Teachers who use
lectures and students could ask questions. The this LMS can access an array of powerful tools
mid-term examination, quiz, and certificate such as assignments, forums, journals, quizzes,
examination were all administered in the face- surveys, chat rooms, and workshops (Cole, 2005).
to-face classes. The course website mainly consisted of five
sections: Course Information, Course Content,
Evaluation and Qualification Course Discussion, Student System, and Assign-
ments and Exercises. Course Information provides
A detailed evaluation of the project was conducted. course description, syllabus, assignments, grading
The authors explored the potential effects of and course-related information. Course Content
web-based SRL with variations in online class includes lectures delivered, conversations that
frequency on students’ skills in using Microsoft happened in the classroom and the students’ ex-
Access. To examine levels of change manipulated ercise files. Students can download the files and
by variations in experimental conditions, we first listen to audio recordings to review or complete
measured students’ Access skills as they entered exercises, repeatedly. Teachers may ask ques-
the class. In the first week, students completed tions in the Course Discussion board in order to
two database files in Access as a pretest. The promote discussion and interaction. Students’
pretest grades showed that the computer skills of personal information and their logs are recorded
almost all the students were low. This confirmed in the Student System. They could also write their
that all participants in the four classes had little learning journals as blogs in this section. Finally,
knowledge or skill using this software package. teachers can assign course work to students in

1186
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

Assignments and Exercises section and students group (C1) and ‘SRL and BL with 5 online classes’
have to complete and upload the assignments group (C2). As shown in Table 2, students’ aver-
according to deadlines. age grade for DBMS in C2 (93.83) is significantly
higher than that in C1 (84.85) group. This is, a
blended course with 5 online classes contributed
RESULTS to better learning effects for vocational students
than that with 10 online classes.
A ‘one-way ANOVA’ was used to compare stu- Results from Table 3 show that the students’
dents’ computing skills in using DBMS under average grade for DBMS in C2 (93.83) is higher
different conditions. As shown in Table 1, students than that in the C3 group (86.40). Thus, the ef-
from the control group (C4) received the lowest fects of web-based SRL on students’ skills in
grades among the four groups, and differences using DBMS are positive, and higher than those
in grades among them are significant. However, without SRL intervention.
the difference among the three BL groups (C1,
C2 and C3) in students’ computing skills is not
statistically significant. Further analysis is needed DISCUSSION AND
to explore whether the online class frequency IMPLICATIONS
influenced students’ learning.
The independent samples t-test is used to com- Many researchers and educators have highlighted
pare the difference of students’ computing skills the importance of using technologies to help
between ‘SRL and BL with 10 online classes’ students learning (Chen, Kinshuk, Wei, Chen &

Table 1. One-way ANOVA: Students’ Grades for Using DBMS

Mean
( I ) ( J ) Std. 95% Confidence
Grades Difference Sig.
Groups Groups Error Interval
(I-J)
Lower Upper
Bound Bound
1 2 -8.981 3.816 .141 -19.76 1.79
3 -1.556 3.792 .982 -12.27 9.15
4 19.256(*) 3.727 .000 8.73 29.78
2 1 8.981 3.816 .141 -1.79 19.76
3 7.425 3.859 .299 -3.47 18.32
4 28.237(*) 3.795 .000 17.52 38.95
3 1 1.556 3.792 .982 -9.15 12.27
2 -7.425 3.859 .299 -18.32 3.47
4 20.812(*) 3.771 .000 10.16 31.46
4 1 -19.256(*) 3.727 .000 -29.78 -8.73
2 -28.237(*) 3.795 .000 -38.95 -17.52
3 -20.812(*) 3.771 .000 -31.46 -10.16

* The mean difference is significant at the .05 level

1187
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

Table 2. Independent samples t-test: Scores of C1 and C2

Group n Mean S. D. F t-value df p


C1 44 84.85 17.350 1.274 -2.493 83 .015**
C2 41 93.83 15.749

Note. **p < 0.05

Table 3. Independent samples t-test: Scores of C2 and C3

Group n Mean S. D. F t-value df p

C2 41 93.83 15.749 .857 2.507 81 .014**


C3 42 86.40 10.834

Note. **p < 0.05

Wang, 2007; Connolly, MacArthur, Stansfield Firstly, our research contributed to the existing
& McLellan, 2007; Liu & Tsai, 2008; Shen, Lee literature by demonstrating that lower online
& Tsai, 2007b). There are several advantages class frequency in a BL course is more helpful
in applying a BL approach (Boyle et al., 2003). to students learning. Second, this study speci-
It is also found that many students studying in fies how teachers help students to regulate their
undergraduate and part-time graduate programs learning by applying web-enabled SRL in a BL
indicated their preferences for retaining some form course, and further contribute to their learning.
of face-to-face teaching while at the same time Finally, this empirical study demonstrated that a
taking advantage of e-learning (Lee and Chan, BL course with SRL strategies was more effective
2007). Meanwhile, the policy of e-learning in in developing students’ computing skills than a
Taiwan is relatively conservative in contrast with course in the traditional classroom.
that in the U.S. Teachers in this context have to
adopt BL if they want to implement e-learning Effects of Online Class
in their courses. Moreover, many vocational stu- Frequency
dents are addicted to shopping websites, online
games, and online messengers (Shen, Lee, Tsai As the results showed in Table 1, the effects of
& Ting, 2008). In this regard, it is necessary to BL on students’ scores for computing skills is sig-
develop an appropriate design and arrangement nificantly higher than those who learned through
for BL courses. traditional teaching. Students’ computing skills
To improve our understanding of this issue, in C1, C2 and C3 were significantly higher than
the authors brought in and then tested rigorously those in C4. The result in this study is similar to
a set of hypotheses among four conditions. Ac- Castelijn and Janssen’s (2006), Shen, Lee, and
cording to the findings of this study, we believe Tsai’s (2007b), and Yushau’s (2006) studies that
that our research has made some contributions show the positive effect of blended e-learning on
to e-learning theory in three different ways. students’ learning and attitude toward computers
and mathematics.

1188
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

Moreover, this study attempted to further


understand and develop the appropriate design The Effects of Web-Enabled SRL
and arrangement of BL courses for schools and
teachers. It is shown in Table 2 that a BL course With respect to the effects of web-enabled SRL,
with 5 online classes could result in significantly the results shown in Table 3 indicate that the
better learning effects than that with 10 online difference of students’ computing skills between
classes (p = 0.015). It is mentioned that students C2 and C3 groups is statistically significant (p
in the vocational system tend to have lower levels = 0.014). The success in online courses often
of academic achievement (Lee, 2003), have low depends on students’ abilities to successfully
confidence and motivation in learning (Su, 2005), direct their own learning efforts (Cennamo, Ross
have low interest and negative attitude toward their & Rogers, 2002). E-learning should be treated as
learning (Chen & Tien, 2005), do not adequately self-directed learning because the learner attends
get involved in their schoolwork, and do not care lectures only to register time, place, subject, and
so much about their grades (Shen, Lee & Tsai, to alter the order of attending lectures (Lee &
2007a). In this specific context, teachers could Lee, 2008). Through the intervention of SRL, the
adopt technologies and teaching websites to help teacher assigned course work to students in the
students achieve better learning performance. For course website and students had to complete and
example, the BL course and audio-recorded con- submit the assignments at the required time. The
tent provide the flexibility and opportunities for web-enabled SRL helped students regulate their
students to attend class, review the course content, learning behaviors, and further contributed to their
and practice what they learn at their convenience, learning effects. This result is similar to Chang’s
particularly before the certificate examinations. (2005), and Yukselturk and Bulut’s (2007) stud-
However, the online class frequency in a blended ies indicating that self-regulation helped students
course is also one of the critical factors that in- become more responsible for their learning and
fluences students learning. In the online classes, contribute to further success.
the physical absence of the instructor and the Based on our findings, we provide suggestions
increased responsibility demanded of learners to for teachers who teach application software, par-
effectively engage in learning tasks may present ticularly for those emphasize earning certification.
difficulties for learners, particularly those with Teachers who wish to stick to traditional methods
low self-regulatory skills (Dabbagh & Kitsantas, of teaching, without applying networked multi-
2005). Students retreating to the isolation of their media, may no longer be employing a fruitful
computers may avoid school activities and course approach. Students from the control group (C4)
involvement (Treuer & Belote, 1997). Too many received the lowest grades among four groups
online classes in a BL course may even damage (see Table 1). In this traditional learning environ-
low-achieving students’ learning. In this regard, ment, students have neither chance nor channel
teachers should arrange the appropriate mix of to review or practice for the tests, and usually
blended classes for students. For example, one ignored the problems of their inadequate skills
traditional class session accompanied the last and knowledge. Therefore, it is suggested that
online class for students to ask questions or for teachers should redesign their courses, design
the teacher to check students’ progress. With appropriate arrangement of BL courses, and adopt
appropriate design, BL could really contribute new instructional methods and technologies to
to students learning. fully exploit the benefits of web-based learning
environments.

1189
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

In conclusion, this study explores that the Chen, C.H., & Tien, C.J. (2005). Market Segmenta-
effects of web-based SRL with variations in fre- tion Analysis for Taking Skill Test by Students in an
quency of online classes on the development of Institute of Technology. Retrieved July 15, 2008,
vocational students’ computing skills. This study from http://www.voced.edu.au/td/tnc_85.574
highlights the necessity of applying innovative
Chen, N.S., Kinshuk, Wei, C.W., Chen, Y.R., &
teaching methods and technologies, and appropri-
Wang, Y.C. (2007). Classroom climate and learn-
ate arrangement of BL courses to help students
ing effectiveness comparison for physical and
learn and pass the certificate examinations. Fur-
cyber F2F interaction in holistic-blended learn-
thermore, this research may provide reference
ing environment. Proceedings of the 7th IEEE
about the intervention of web-enabled SRL and
International Conference on Advanced Learning
arrangement of online classes in BL courses for
Technologies (pp. 313-317).
schools, scholars and teachers preparing for or
presently engaged in implementing e-learning. Chang, M.M. (2005). Applying self-regulated
learning strategies in a web-based instruction - An
investigation of motivation perception. Computer
REFERENCES Assisted Language Learning, 18(3), 217-230.
Cole, J. (2005). Using Moodle: Teaching with
Azevedo, R. (2005). Using hypermedia as a meta-
the Popular Open Source Course Management
cognitive tool for enhancing student learning?
System. O’Reilly Media, CA.
The role of self-regulated learning. Educational
Psychologist, 40(4), 199-209. Connolly, T.M., MacArthur, E., Stansfield, M.H.,
& McLellan, E. (2007). A quasi-experimental
Bielaczyc, K, Pirolli, P., & Brown, A. (1995).
study of three online learning courses in comput-
Training in self-explanation and self-regulation
ing. Computers & Education, 49(2), 345-359.
strategies: investigating the effects of knowledge
acquisition activities on problem solving. Cogni- Dabbagh, N., & Kitsantas, K. (2005). Using
tion and Instruction, 13(2), 221-252. web-based pedagogical tools as scaffolds for
self-regulated learning. Instructional Science,
Boyle. T., Bradley, C., Chalk, P., Jones, R., &
33(5–6), 513-540.
Pickard, P. (2003). Using blended learning to
improve student success rates in learning to pro- Dowling, C., Godfrey, J.M., & Gyles, N. (2003).
gram. Learning, Media and Technology, 28(2 & Do hybrid flexible delivery teaching methods
3), 165-178. improve accounting student learning outcomes?
Accounting Education: An International Journal,
Castelijn, P., & Janssen, B. (2006). Effectiveness of
12(4), 373-391.
Blended Learning in a Distance Education Setting.
Retrieved July 15, 2008, from http://www.ou.nl/ Ertmer, P.A., Newby, T.J., & MacDougall, M.
Docs/Faculteiten/MW/MW%20Working%20 (1996). Students’ approaches to learning from
Papers/gr%2006%2006%20castelijn.pdf case-based instruction: The role of reflective
self-regulation. American Educational Research
Cennamo, K.S., Ross, J. D., & Rogers, C.S. (2002).
Journal, 33(3), 719-752.
Evolution of a web-enhanced course: Incorpo-
rating strategies for self-regulation. Educause Gribbons, B. & Herman, J. (1997). True and
Quarterly, 25(1), 28-33. quasi-experimental designs. Practical Assess-
ment, Research & Evaluation, 5(14). retrieved

1190
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

July 15, 2008, from http://PAREonline.net/getvn. eracy. Ph. D. dissertation. Austin, The University
asp?v=5&n=14 of Texas.
Greene, J.A., & Azevedo, R. (2007). Adolescents’ Li, K.C., Tsai, Y.T., & Tsai, C.K. (2008). Toward
use of self-regulatory processes and their relation development of distance learning environment
to qualitative mental model shifts while using in the grid. International Journal of Distance
hypermedia. Journal of Educational Computing Education Technologies, 6(3), 45-57.
Research, 36(2), 125-148.
Liu, C.C., & Tsai, C.C. (2008). An analysis of
Jacobson, M.J. (2005). From non-adaptive to adap- peer interaction patterns as discoursed by on-line
tive educational hypermedia: Theory, research, small group problem-solving activity. Computers
and design issues. In S. Chen & G. Magalas (Eds.), and Education, 50(3), 627-639.
Advances in Web-based education: Personalized
Liu, C.H., Chiang, T.C., & Huang, Y.M. (2007).
learning environments (pp.302-330). Hershey,
Assessment of effectiveness of web-based training
PA: Idea Group.
on demand. Interactive Learning Environments,
Järvelä, S., Näykki, P., Laru, J., & Luokkanen, T. 15(3), 217-235.
(2007). Structuring and regulating collaborative
Lynch, R., & Dembo, M. (2004). The relationship
learning in higher education with wireless net-
between self-regulation and online learning in a
works and mobile tools. Educational Technology
blended learning context. International Review
& Society, 10(4), 71-79.
of Research in Open and Distance Learning, 5(2).
Lai, S.Q., Lee, C.L., Yeh, Y.J., & Ho, C.T. (2005). A Retrieved July 15, 2008, from http://www.irrodl.
study of satisfaction in blended learning for small org/index.php/irrodl/article/view/189/799
and medium enterprises. International Journal of
Marino, T.A. (2000). Learning Online: A view
Innovation and Learning, 2(3), 319-334.
from both sides. The National Teaching & Learn-
Lee, P.W.R., & Chan, F.T. (2007). Blended learn- ing Forum, 9(4), 4-6.
ing: Experiences of adult learners in Hong Kong.
Матросова, C.T.A. (2007). Computer World.
In J. Fong & F. L. Wang, (Eds.), Blended Learning
Retrieved July 15, 2008, from http://venec.ulstu.
(pp. 79-87). Retrieved July 15, 2008, from http://
ru/lib/go.php?id=1566
www.cs.cityu.edu.hk/~wbl2007/WBL2007_Pro-
ceedings_HTML/WBL2007_PP079-087_Lee. Montalvo, F., & Torres, R. (2004). Self-regulated
pdf learning: Current and future directions. Electronic
Journal of Research in Educational Psychology,
Lee, J.K., & Lee, W.K. (2008). The relationship
2(1), 1-34.
of e-Learner’s self-regulatory efficacy and per-
ception of e-Learning environmental quality. Nota, L., Soresi, S., & Zimmerman, B.J. (2004).
Computers in Human Behavior, 24(1), 32-47. Self-regulation and academic achievement and re-
silience: a longitudinal study. International Jour-
Lee, L.S., & Huang, J.J. (1996, May 27-29). Tech-
nal of Educational Research, 41(3), 198–251.
nological and vocational teacher education in
Taiwan, R. O. C. Paper presented at the Taiwan- Overbaugh, R.C., & Casiello, A.R. (2008). Distrib-
Australia Conference on Vocational Education uted collaborative problem-based graduate-level
and Training, Melbourne, Australia. learning: Students’ perspectives on communica-
tion tool selection and efficacy. Computers in
Lee, Y.M. (2003). An Investigation of Taiwanese
Human Behavior, 24(2), 497-515.
Graduate Students’ Level of Civic Scientific Lit-

1191
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

Pereira, J.A., Pleguezuelos, E., Merí, A., Molina- Journal of Distance Education Technologies,
Ros, A., Molina-Tomás, M.C., & Masdeu, C. 6(3), 69-84.
(2007). Effectiveness of using blended learn-
Shen, P.D., Lee, T.H., Tsai, C.W., & Ting, C.J.
ing strategies for teaching and learning human
(2008). Exploring the Effects of Web-Enabled
anatomy. Medical Education, 41(2), 189-195.
Problem-Based Learning and Self-Regulated
Reyero, M., & Tourón, J. (2003). The development Learning on Vocational Students’ Involvement
of talent: acceleration as an educational strategy. in Learning. European Journal of Open, Dis-
A Coruña, Netbiblo. tance and E-Learning, 2008(1). Retrieved July
15, 2008, from http://www.eurodl.org/materials/
Roces, C., & González Torres, M.C. (1998). Ability
contrib/2008/Shen_Lee_Tsai_Ting.htm
to self-regulate learning. In J.A. González Pienda,
& J.C. Núñez (Eds.), Dificultadesde aprendizaje Su, M.H.M. (2005). A study of EFL technological
escolar. Madrid, Pirámide/Psicología. and vocational college students’ language learning
strategies and their self-perceived English profi-
Rovai, A.P., & Jordan, H.M. (2004). Blended
ciency. Electronic Journal of Foreign Language
learning and sense of community: A comparative
Teaching, 2(1), 44-56.
analysis with traditional and fully online gradu-
ate courses. International Review of Research in Treuer, P., & Belote, L. (1997). Current and emerg-
Open and Distance Learning, 5(2). Retrieved July ing applications of technology to promote student
15, 2008, from http://www.irrodl.org/index.php/ involvement and learning. New Directions for
irrodl/article/view/192/274 Student Services, 78, 17-30.
Shelley, D.J., Swartz, L.B., & Cole, M.T. (2007). Weinstein, C. (1989). Teacher education students’
A comparative analysis of online and traditional preconceptions of teaching. Journal of Teacher
undergraduate business law classes. International Education, 40(2), 53-60.
Journal of Information and Communication Tech-
Winne, P.H. (2001). Self-regulated learning
nology Education, 3(1), 10-21.
viewed from models of information processing.
Shen, P.D., Lee, T.H., & Tsai, C.W. (2007a). Ap- In B. J. Zimmerman & D. Schunk (Eds.), Self-
plying Web-Enabled Problem-Based Learning and regulated learning and academic achievement:
Self-Regulated Learning to Enhance Computing Theoretical perspectives (pp. 153-189). Mahwah,
Skills of Taiwan’s Vocational Students: A Quasi- NJ: Erlbaum.
Experimental Study of a Short-Term Module.
Winne, P.H., & Hadwin, A. (1998). Studying as
Electronic Journal of e-Learning, 5(2), 147-156.
self-regulated learning. In D. J. Hacker, J. Dun-
Shen, P.D., Lee, T.H., & Tsai, C.W. (2007b, losky, & A. Graesser (Eds.), Metacognition in
September 15-17). Facilitating Students to Pass educational theory and practice (pp. 277-304).
Certificate Tests via Blended E-Learning with Hillsdale, NJ: Erlbaum.
Self-Regulated Learning: A Quasi-experimental
Yang, S.C., & Tung, C.J. (2007). Comparison of
Approach. WSEAS Proceedings on Multimedia,
Internet Addicts and Non-addicts in Taiwanese
Internet & Video Technologies. Beijing, China
High School. Computers in Human Behavior,
Shen, P.D., Lee, T.H., & Tsai, C.W. (2008). 23(1), 79-96.
Enhancing Skills of Application Software via
Yukselturk, E., & Bulut, S. (2007). Predictors for
Web-Enabled Problem-Based Learning and
student success in an online course. Educational
Self-Regulated Learning: An Exploratory Study.
Technology & Society, 10(2), 71-83.

1192
Exploring the Effects of Web-Enabled Self-Regulated Learning and Online Class Frequency

Yushau, B. (2006). The effects of blended e- Zimmerman, B.J., & Schunk, D.H. (1989). Self-
learning on mathematics and computer attitudes in Regulated Learning and Academic Achievement:
pre-calculus algebra. The Montana Mathematics Theory, Research, and Practice. New York:
Enthusiast, 3(2), 176-183. Springer-Verlag.
Zimmerman, B.J. (1990). Self-regulated learn- Zimmerman, B.J., & Schunk, D.H. (2001). Self-
ing and academic achievement: An overview. Regulated Learning and Academic Achieve-
Educational Psychologist, 25(1), 3-17. ment: Theoretical Perspectives. Mahwah, NJ:
Erlbaum.
Zimmerman, B.J. (1998). Developing self-regula-
tion cycles of academic regulation: An analysis of Zimmerman, B.J., Bonner, S., & Kovach, R.
exemplary instructional model. In D. H. Schunk & (1996). Developing self-regulated learners: Be-
B. J. Zimmerman (Eds.), Self-regulated learning: yond achievement to self-efficacy. Washington,
From teaching to self-reflective practice. New DC: American Psychological Association.
York: Guilford.

This work was previously published in International Journal of Mobile and Blended Learning, Vol. 1, Issue 3, edited by D.
Parsons, pp. 1-16, copyright 2009 by IGI Publishing (an imprint of IGI Global).

1193
Section V
Organizational and
Social Implications

This section includes a wide range of research pertaining to the social and organizational impact of Web
technologies around the world. Chapters included in this section analyze social marketing, e-government,
Web vendors, and Web tourism. The inquiries and methods presented in this section offer insight into the
implications of Web technologies at both a personal and organizational level, while also emphasizing
potential areas of study within the discipline.
1195

Chapter 5.1
Building Trust in
E-Commerce through Web
Interface
Muneesh Kumar
University of Delhi South Campus, India, & ESC-PAU, France

Mamta Sareen
University of Delhi, India

Abstract copies are available for purchase from InfoSci-


on-Demand.com]
The emergence of Internet has revolutionalized
the way businesses are conducted. The impact
of e-commerce is pervasive, both on companies Introduction
and society as a whole. It has the potential to
impact the pace of economic development and Convergence of Information Technology and
in turn influence the process of human develop- telecommunication technology has resulted in
ment at the global level. However, the growth emergence of a new economy wherein the buy-
in e-commerce is being impaired by the issue ing and selling process is being executed through
of trust in the buyer-seller relationship which is Internet and other computer networks. This is
arising due to the virtual nature of e-commerce being termed as e-commerce. The proliferation of
environment. The online trading environment is Internet technologies into business has fundamen-
constrained by a number of factors including web tally changed the relationship between suppliers
interface that in turn influences user experience. and consumers. It has provided faster access and
This article identifies various dimensions of web better knowledge of commodities and prices. The
interface that have the potential to influence trust ability to exchange information in both directions
in e-commerce. The empirical evidence presented between producer and consumer has created a
in the article is based on a survey of the web relationship not previously possible. E-commerce
interfaces of 65 Indian e-Marketplaces. [Article is changing the way business is being conducted

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Building Trust in E-Commerce through Web Interface

and eventually all companies will have to make Trust in E-commerce


the transition to remain competitive because soon
all customers will expect this level of service and Trust is defined as “the willingness of a party
it will inevitably become the standard for customer to be vulnerable to the actions of another party
satisfaction (Rust & Kannan, 2003) based on the expectations that the other party
The impact of e-commerce is pervasive, both will perform a particular action important to
on companies and society as a whole. It is the first the trustor, irrespective of the ability to monitor
mass application of information and communica- or control that other party” (Mayer,Davis and
tion technologies in the movement towards digital Schoorman 1995). In the context of e-commerce,
economy. It has broken all man-made boundaries trust may be regarded as a judgment made by the
and provided an opportunity for both buyers and user, based on general experience learned from
sellers to interact among themselves regardless being a customer/seller and from the perception
of difference in language, society, culture and of a particular merchant. In other words, trust is
tradition. also seen as a generalized expectancy that the
The rapid growth of e-commerce is now word, promise, or written statement of another
being related to economic development and is party can be relied on (Rotter, 1980).
often been cited as a driver of economic growth. To date, research on understanding online trust
E-commerce is also been touted as a powerful and e-commerce is limited (Grabner-Kräuter and
medium through which less developed economies Kaluscha 2003; Yoon 2002; Corritore et al. 2003;
can exploit the potential of global markets. It, thus, Kolsaker and Payne 2002). In their critical reviews
has the potential to impact the pace of economic of website and/or ecommerce trust, Corritore et
development and in turn influence the process of al. (2003) and Grabner- Kräuter and Kaluscha
human development at the global level. However, (2003) argued that there is a lack a conceptual
the growth in e-commerce is being impaired by understanding of online trust and theoretical
the issue of trust which is arising due to virtual support for its role in online transactions and re-
nature of e-commerce environment. This virtual lationships. Without trust, businesses are unable
nature of e-commerce environment imposes cer- to function (Reichheld et al. 2000). Jian, Bisantz,
tain constraints on buyer-seller relationship that and Drury (2000) and Bailey et al. (2003) claim
did not exist in traditional face-to-face transac- that trust not only plays a strong role in human-
tions. The on-line trading environment basically to-human interactions, but also plays a critical
thrives on ‘virtuality’ (Handy, 1995) and ‘user role in human-to-computer interactions.
experience’ (Marsh, 2000). Since, the transac-
tions in this virtual environment are conducted
through the ‘veil’ of web interface, trust becomes Literature Review
an important issue. The web interface acts like the
only ‘contact point’ among the buyers and sell- A rich web interface may have a positive impact on
ers. Hence, there is a need for the web interface trust in the faceless environment of e-commerce.
to induce trust in online environment. The focus Several studies like Fogg et al 2001; Lee and
of the present article is to identify various trust Kim & Moon, 2000; Neilsen, 1999, 2005 etc.
inducing web dimensions that may enhance the reported evaluations of a list of design features
effectiveness of web interface and there by help in that could potentially appear on the web interface
inducing trust among the e-commerce players. to impact trust. Ang and Lee (2000) stated that
if the web site does not lead the buyer to believe
that the seller is trustworthy, no business can be

1196
Building Trust in E-Commerce through Web Interface

conducted. In other words, one key consideration the information is presented to the user. The web
in fostering online trust in e-commerce is to build interface is not just how it looks; it is how easy
a trust inducing web interface. Lohse and Spiller it is to learn, how well it recedes into the sub
(1998) identified four interface design features consciousness of users, and how well it supports
that affect the effectiveness of the web interface. users’ tasks. Different authors have suggested
Their results indicated that features like effective various features for making the web interface more
navigation, detailed product descriptions, links, effective (Egger, 2003; Neilson, 1999; Wang et al
etc. affect the trust in e-commerce activity. Xiling 2007 and others). These features can be classified
Zhou asserted that poor quality of web interface, into three broad categories i.e. appeal, content
lack of proper content in the web sites, unintui- and usability.
tive navigation, etc. can diminish the trust in the Appeal: It refers to ‘attitude’ component and
concerned company in e-commerce activities. the first impression a user gets when accessing
Bailey et al stressed that visual aesthetics and a site for the first time Lindgaard (1999) stated
navigation quality of a web site help to assess its that an immediate negative impression may well
trustworthiness in e-commerce activities. Jarv- determine the subsequent perception of the site’s
enpa et al, 1999 stated that a web site with trust quality and usability, whereas one may inherently
inducing features functions as a skillful sales judge a site by its first impression. Literature from
person for that company and therefore moderates psychology also stresses the important role of a
the disadvantages of an impersonal web site. It is party’s first impression, as someone’s confirma-
believed that online buyers in e-commerce look tion bias would entail that all user actions will
for the presence of positive cues about a site’s unconsciously seek to confirm the first impres-
general trustworthiness, as well as for the absence sion rather than falsify it (Kahneman & Tversy,
of negative cues. Hence, the e-commerce players 1973; Good, 1988). Fogg et al. (2002) reported,
by carefully designing their site to set and meet in their large study about how people evaluate
user expectations can influence the trustworthi- the credibility of websites that, almost 50% of
ness of other players. all comments made by participants referred to
Arion et al. (1994) asserted that user interface graphic design. They therefore argue that, in the
is the point where trust is generated. They stated context of online credibility and trust, findings
that trust is a dynamic process, initially based indicate that looking good is often interpreted
on faith due to the lack of evidence that seeks as being good, credible and trustworthy. Hence,
to reach a certain level of confidence, i.e., where appeal has largely to do with the site’s graphics
there is conclusive evidence in favor of trusting design and layout. In addition, Demonstrating
behavior. In their consideration of computer- important clients or providing links towards com-
supported cooperative work (CSCW) systems, pany’s various policies also instills trust among
Arion et al. (1994) argued that development of the users (Doney and Canon, 1997).
trust in the human-computer interaction need to Usability: Usability is the measure of the qual-
be supported by the infrastructure/system. ity of a user’s experience when interacting with the
A web interface provides total “user experi- web site Sweden Canada Link (2001) stated that
ence”. Hence, an effective web interface can make usability is about making the visit to the website
this contact point between the trading partners as effective as possible for the users. The focus of
more meaningful and help in building up trust. usability is on enabling users, whatever their inter-
The effectiveness of the web interface may also ests and needs, by removing barriers and making
be determined by factors such as the aesthetic the system as easy to use as possible. According
appearance of the site, the content and the way to usability expert Jakob Nielsen, usability is a

1197
Building Trust in E-Commerce through Web Interface

necessary condition for survival on the web. If may also be seen as a sign of honesty and compe-
a web site is difficult to use, people leave. If the tence (Egger, 2002). In addition, the provision of
homepage fails to clearly state what a company related content, if relevant, can also be interpreted
offers and what users can do on the site, people as the company truly understanding its custom-
leave. Indeed, visual design is presented to the ers’ needs. The credibility of the information has
user passively, while the user actively needs to also been observed to be very important about a
navigate the website in order to access relevant company’s ethical standards.
information. Usability is all the more important Various dimensions of the web interface ad-
in the context of online shopping as it is known to dress the appeal, usability and content features
be an important condition for the acceptance and of an effective web interface. Various authors
adoption of new technologies. The Technology have identified a number of such dimensions
Acceptance Model (TAM), as defined by Davis (Kim 1998; Egger 2003; Wang et al 2006; etc.).
(1989), holds that usefulness and ease of use are Some of these dimensions are categorized as trust
both strong predictors of trust. This model has inducing dimensions. Kim (1998) had identified
also been explicitly used to relate trust and e- four interface design features which contribute
commerce by researchers like Gefen and Straub to trustworthiness of the web site. Wang et al.
(2000), Pavlou (2001), amongst others. (2007) also stressed on certain web dimensions
Content: Websites contain information and to enhance the richness of web interface. He
serve as a medium that predominantly is used for proposed four broad dimensions of web interface
the transfer of information (either technical or not). namely, graphic design dimension, Structure
This plays an important role in the effectiveness design dimension, Content design dimension and
of the site. Product information has historically Social-cue design dimension that may influence
been regarded as a critical element of the content trust. However, the various studies do not include
of web interface. A number of researchers have features that are commonly found in any modern
investigated the relationship between web inter- e-commerce web site and the e-commerce play-
face and information structure (e.g., Gay et al., ers are interested in. These include reference to
1991: Radha and Murphy, 1992; Mohageg, 1992; the kind of security policy, privacy policy being
Utting and Yankelovich, 1989), concluding that followed by the company for online transactions,
information structure is an essential element of an the technology related policies and procedures
effective web interface. However, a web site may followed by the company to address the security,
also contain other information such as detailed privacy etc. issues, the statements from well
and relevant information about the company, its known customers, media excerpts, etc. Thus,
complete offline address, seals of approvals from there is a need for a more comprehensive model
various trusted third parties, etc. Green (1998) relating trust and web interface.
stated that in the e-commerce environment,
information plays an important role as business
audiences seek more information about products/ Web Interface and Trust
services and the company. To be able to convey the Model
information effectively, it is necessary to structure
it properly. Correct and detailed descriptions of In order to incorporate the various gaps in the
the products and services offered by the company earlier model relating trust with web interface,
helps the users to make informed decisions about the present article proposes five trust inducing
their transactions. Features that reduce user costs, dimensions of an effective web interface. These
such as comparisons with competitive products, dimensions are: a) User Interface Dimension; b)

1198
Building Trust in E-Commerce through Web Interface

Figure 1. Web interface and trust model

User Interface
Appeal

T Social-ability
R Usability
U Demonstrability

S
T Content Information
Structure

Information

Information Structure Dimension; c) Informa- trust levels among the users include: a)
tion Content Dimension; d) Demonstrability Home Page; b) Graphics Interface; c) Links;
Dimension; and e) Social-ability Dimension. d) Professionalism; and e) Loading Time.
The relationships between these dimensions and • Information Structure Dimension: The
three features identified earlier are exhibited in structure dimension defines the overall
the Figure 1. organization and accessibility of displayed
information on the web site. Ease of naviga-
• User Interface Dimension: User Interface tion has frequently been mentioned as a key
indicates the appearance and the tools avail- to promote online trust (e.g., Cheskin/Sapi-
able for accessing the information contained ent Report, 1999; Neilsen, 1998). In other
in the web site/portal. Constaine (1995) words, users must be able to easily locate
pointed out that interface is important aspect the information they seek on the web site.
as the more intuitive the user interface is, This ease-of- use reflects two characteristics
the easier it is to use and trust. The website of a trustworthy web site: simplicity and
must be recognizable as from the organi- consistency. Buyers appreciate simplicity
zation. That is, it must be obvious that the or a clear design of e-commerce web sites
look of the site – colors, logos, layout, etc. because it reduces the perceived risks of
is consistent with other collateral from the deception, frustration, and wasting time.
organization. The various features offered When the structure and design of the web
by the web site that normally giving the first site are consistent, users feel more confident
impression about the company lays the initial using the site because they can transfer
foundation of trust building process. Kim their learning from one sub-site to the next
and Moon, (1999) reported that the overall rather than having to learn everything over
color layout and graphical interface influ- again for each new page and trust is build
ences the trustworthiness of the web site. (Neilsen, 1998,). For example, broken links,
This dimension aids in addressing the appeal meaningless images, and similar ‘‘hygiene
aspect of an effective web interface. Various factors’’ may relate to users dissatisfaction
features that may be useful in enhancing with a web site (Zhang et al., 1999). Key

1199
Building Trust in E-Commerce through Web Interface

features of structure dimension of a web accreditations earned help in building trust


site may include: a) Navigation; b) Acces- levels.
sibility; c) Functionality; d) Consistency; • Social-ability Dimension: This dimension
and e) Learnibility. relates to embedding social cues, such as
• Information Content Dimension: This face-to-face interaction and social presence,
dimension refers to the informational com- into web site interfaces via different commu-
ponents that can be included on the web nication media, because a lack of the ‘‘human
site, either textual or graphical. A logically touch’’ or presence may constitute a barrier
structured web site providing comprehen- for at least some consumers to trust online
sive, correct, and current product informa- merchants (e.g., Riegelsberger & Sasse,
tion instills trust among the users (Egger, 2002). The effectiveness of a personalization
2001; Neilsen, 1999). If the information system improves in the long run. Weiner and
regarding the products is precise, factual A. Mehrabian (1968) stated that the choice
and contains links of details required, if of language can help create a sense of psy-
necessary, then trust is build (Bhattacharya, chological closeness and warmth. Where as
2001). Recent market surveys include that Nass and Steuer (1993) stressed that the use
some of the companies are using their web of natural and informal language can impact
sites as a part of integrated communication perceived social presence. Yoon (2002) also
strategy to create trust and action (Shee- showed that web site trust is influenced by
han and Doherty, 2001). In such cases the consumer familiarity. Every time a customer
website interface plays an important role interacts with the web site, the personaliza-
in e-business transactions. The contents tion mechanism collects new data about the
should be displayed as being less complex user’s preferences, so that a more and more
and more users friendly that enriches the satisfactory service can be offered.
visitor’s experience and motivates him/her
to visit the site again (G.Chakraborty et al
2003). key features of structure dimension Methodology and
of a web site may include: a) Navigation; b) Results
Accessibility; c) Functionality; d) Consis-
tency; and e) Learn-ability. A list of 100 B2B web e-marketplaces operating
• Demonstrability Dimension: Several in India was prepared through the use of various
researchers stress the importance of ‘‘de- search engines like Google, Yahoo, MSN etc. The
monstrability’’ in e-commerce, which is to criterion for the selection of these e-marketplaces
promote the brand reputation of a company was random. On visiting these web sites, it was
online. It is often seen that various features found that some of the web portals were merely
like offline address of the company, details a directory of sellers and buyers and not actually
about real people behind etc play the same carrying out e-commerce transactions. Such web
role of offering certain clues about the cred- portals were excluded from the sample. Hence,
ibility of the company as the physical clues in a total of 65 B2B web sites/e-marketplaces were
the brick and mortar business transactions. selected for the purpose of the survey. Based on
These clues help the potential trading part- the number of elements present, the web site was
ners to assess the credibility of the company. to be rated on a 5 point Likert scale for each of the
Especially displaying the seals of approval trust inducing dimension. Finally, an overall trust
from various trusted third parties and the rating of the web portal was also to be obtained for

1200
Building Trust in E-Commerce through Web Interface

each web interface of the select e-marketplace. having significant contribution towards building
Initially, 25 participants were requested to trust in web interface. This could be because of the
rate six e-marketplaces. The participants were IT fact that in India, complete virtual B2B transac-
savvy in the age range of 30-45 years and came tions do not occur very often and they are aided
from a variety of background like public sector, with various offline channels like phone, fax, etc.
business sector, private sector, banking sector and However, the result assures that all these dimen-
academic sector. The participants were asked to sions of an effective web interface as proposed
provide their own ratings of each dimension and in the model have a potential to enhance trust
also the overall rating of the web site based on a among the users. Further, socio-ability dimension,
questionnaire consisting of 20 questions. Hence, though considered important in B2C e-commerce
a total of 120 responses were received from each activities, was not found to be contributing much
of the 25 participants. Cronbach’s Alpha Scale towards trust levels in e-marketplaces.
Reliability test was used to test the reliability of In order to identify any bias in the evaluation
the questionnaire and it was found to be 0.8751, of the web interfaces of the six e-marketplaces,
which is fairly good degree of reliability. the author also independently evaluated all the
In order to study the significance of these select e-marketplaces. The results of the evalu-
trust inducing dimensions for the level of trust, ations were compared with the trust equation
linear regression model was used (the limitations obtained of these select e-marketplaces. It was
of linear regression model in this context may be observed that the difference between the two
recognized). The model used the level of trust as results was not significant. This would imply
dependent variable and each of the trust inducing that the evaluations of the e-marketplaces done
dimensions as independent variable. The purpose by 25 participants and the author held nominal
was to find out any redundant dimension included bias. Therefore, evaluation of the remaining 59
in the dimension. The regression analysis was e-marketplaces was carried out independently by
carried out on the data collected from these 25 the author, in the same manner as it was done by
participants for each of the six e-marketplaces. The the 25 participants.
adjusted R 2 ranged from 86% to 92%. This would The regression model used earlier was used
imply that more than 85% of the variation in the on the data so collected from the evaluations of
trust ratings could be collectively explained by web interfaces of the selected e-marketplaces.
the five trust inducing dimensions of an effective The results of the linear regression analysis are
web interface. The results of regression analysis presented in the Table 2. As may be observed
for the ratings given for one of the e-marketplaces from Table 2, all the five trust inducing web di-
are presented in Table 1. mensions were found to be contributing towards
During the systematic elimination process, all trust in the web site.
the five trust inducing dimensions were found to All the dimensions of effective web interface
be significant. The results showed Demonstrability as identified in the model were found to be good
and Information Content dimension as significant predictors of level of trust. The results were
trust inducing dimensions of web interface. This fairly comparable with the results obtained from
further strengthens the reasons of the present ar- the linear regression analysis carried out on the
ticle for the inclusion of demonstrability dimension data collected through 25 participants with re-
as a trust inducing dimension for web interface to spect to six select e-marketplaces. Thus, it may
the model proposed by Wang et al. The informa- be concluded that the five trust inducing web
tion structure dimension and the User Interface dimensions namely Information Content dimen-
dimension of web interface also were found to be sion, Demonstrability dimension, Information

1201
Building Trust in E-Commerce through Web Interface

Table 1. Trust ratings of an e-marketplace and trust inducing dimensions. Partial results of
regression
Unstandardized
Model Standardized Coefficients
Coefficients
B Std. error Beta t Sig
(Constant) -0.312 0.460 - -0.677 0.506
User Interface 0.243 0.115 0.242 2.912 0.003
Info. Structure 0.225 0.116 0.265 2.093 0.009
Info. Content 0.327 0.217 0.327 3.209 0.001
Demonstrability 0.338 0.165 0.264 3.049 0.002
Socio-ability 0. 033 0.097 0.003 1.634 0.020

Table 2. Trust ratings of an e-marketplace and trust inducing dimensions. Partial results of
regression
Unstandardized Standardized
Model
Coefficients Coefficients
B Std. Error Beta t Sig
(Constant) -.633 0.232 - -2.732 0.008
User Interface 0.178 0.114 0.242 2.816 0.008
Info. Structure 0.269 0.102 0.211 2.334 0.011
Info. Content 0.421 0.094 0.382 3.486 0.001
Demonstrability 0.296 0.105 0.261 2.836 0.006
Socio-ability 0.067 0.114 0.060 0.590 0.055

Structure dimension, User Interface dimension as many trust inducing dimensions as possible. It
and the Social-ability dimension (in that order) may, however, be possible that the article might
positively contribute towards trust building pro- have overlooked certain issues that might influ-
cesses in e-commerce. ence the web interface of the e-commerce sites.
However, user experience is a very complex
issue related to human-computer interface and
Limitations and Scope for may vary from individual to individual. Ideally,
Future Research different individuals must have rated all the e-
marketplaces. However, it was not possible for
The main limitation of this article could be the the present scope.
coverage of only e-marketplaces in validating
trust in e-commerce. Since, e-commerce involves
various other activities, it would be better, if the Conclusion
sample data would contain web interfaces from
other segments of e-commerce also. Further, the E-commerce has the potential to provide a flip to
model proposed in this article has tried to include the pace of economic development and provide

1202
Building Trust in E-Commerce through Web Interface

a unique opportunity to organizations in less Arion, M., Numan, J. H., & Pitariu, H. (1994).
developed countries to operate in global mar- Placing Trust in Human-Computer Interactions.
kets. This could also have implications on the Proceedings of 7th European Cognitive Ergonom-
socio-economic conditions of the less developed ics Conference (pp. 352-365).
countries. Trust has been a major hurdle imped-
Ba, S., & Pavlou, P. A. (2002). Evidence of the
ing the growth of e-commerce and the need for
Effect of Trust Building Technology in Electronic
enhancing trust cannot be over-emphasized. The
Markets: Price Premiums and Buyer Behavior.
article, through validation, identifies various
MIS Quarterly, 26(3)
trust inducing dimensions for enriching the web
interface and there by inducing trust. Focus on Bailey, J. P., & Faraj, S. (2000).The Role of In-
these dimensions, which demonstrate ‘correct termediaries in the Development of Trust on the
and concise’ information, ‘relevant structure’ WWW: The Use and Prominence of Trusted Third
of information and ‘usability’, would help in Parties and Privacy Statements. JCMC, 5(3).
inducing trust in the faceless environment of e-
Bailey. A. (2005). Consumer Awareness and Use
commerce. This in turn would improve the user
of Product Review Websites. Journal of Interac-
experience and the buyer-seller relationship in
tive Advertising, 6(1), 90-108. 
online trading environment. This supports the
earlier findings of Arion et al. (1994) wherein Bhattacherjee, A. (2002). Individual Trust in On-
trust in human-computer interactions was stated line Firms: Scale Development and Initial Test.
to be supported by the system. It may, however, Journal of Management Information Systems,
be noted that the trust in buyer-seller relationship 19(1), 211-241.
is also influenced by a number of factors and the
Chakraborty, G., Lala, V., & Warren, D. (2002).
enhancement of features of web interface would
An Empirical Investigation of Antecedents of B2B
need to be viewed as an integral part of an overall
Websites Effectiveness. Journal of Interactive
trust building strategy of an organization. Thus,
Marketing, 16(4), 51-72 .
by effectively incorporating various features of the
trust inducing dimensions on the web interface, Cheskin/Sapient Research and Studio Archetype/
the merchant is able to provide a trust worthy Sapient. (1999). E-Commerce Trust Study, Re-
platform for the customer across global boundar- trieved June 10, 2008 from http://www.sapient.
ies to transact among each other. This has helped com/cheskin.
in the development of a virtual society which is
Corritore,C. l., Marble, R. P., Wiedenbeck, S.,
driven by trust in technology and enhances the
Kracher, B., & Chandran, A.(2005) Measuring
growth of e-commerce. Professional bodies and
Online Trust of Websites: Credibility, Perceived
business organizations need to play an important
Ease of Use, and Risk. Proceedings of the Eleventh
role in this regard. Development and adherence
Americas Conference on Information Systems,
to globally accepted standards in this regard can
Omaha, USA.
go a long way in enriching the ‘user experience’
in e-commerce. Davis, J., Mayer, R., & Shoorman, F. (1995). An
integrated model of organizational trust, the acad-
emy of management review, 20(3), 705-734.
References
Davis F. D. (1989). Perceived usefulness, perceived
ease of use and user acceptance of information
ACM Special Interest group on Computer-Human
technology. MIS Quaterly, 13(3), 319-340.
Interaction Group (1992). ACM SIGCHI, Technical
Report. New York: ACM.

1203
Building Trust in E-Commerce through Web Interface

Doney, P. M., & Cannon, J. P. (1997). An Ex- Validation. Journal of Computer Mediated Com-
amination of the Nature of Trust in Buyer-Seller munication, 5(2).
Relationships. Journal of Marketing, 61, 35-51.
Jian, J-Y., Bisantz, A. M., & Drury, C. G. (1998).
Egger F. N. (2003). From interactions to trans- Towards an empirically determined scale of trust
actions: Designing the Trust Experience for in computerized systems: Distinguishing concepts
Business-to-Consumer Electronic Commerce. and types of trust. In Proceedings of the Human
Factors & Ergonomics Society 42nd Annual
Egger F. N. (2001). Affective Design of E-Com-
Meeting, (pp. 501-505).
merce User Interfaces: How to maximize perceived
trustworthiness. Proceedings of The International Keen, P. G. W. (1997). Are you ready for ‘Trust’
Conference on Affective Human Factors Design. Economy. Computer World, 31(16), 80.
London: Asean Academic Press.
Kolsaker, A., & Payne, C. (2002). Engendering
Egger, F. N. (June 27-29, 2001). Affective Design trust in e-commerce: a study of gender based
of E-Commerce User Interfaces: How to maxi- concerns. Marketing Intelligence & Planning,
mize perceived trustworthiness. Conference on 20(4), 206-214.
Affective Human Factors Design. Singapore,
Lohse, G. L., & Spiller, P. (1998). Electronic
(pp. 317-324).
Shopping: Quantifying the Effect of Customer
Fogg B. J., & Nass C. (1997). Effects of comput- Interfaces on Traffic and Sales. Communications
ers that flatter. International Journal of Human- of the ACM, 41(7)
Computer Studies, 46, 551-561.
Lee, J., Kim J., & Moon J. Y. (April 2000).What
Grabner-Kräuter & Kaluscha. (2003). Empirical Makes Internet Users Visit Cyber Stores Again?
research in on-line trust: a review and critical Key Design Factors for Customer Loyalty. Pro-
assessment. International Journal of Human- ceedings of the Computer-Human Interaction
Computer Studies, 58(6), 783–812.   Conference on Human Factors in Computing Sys-
tems, The Hague, Netherlands, (pp. 305-312).
Gefen, D., & Straub, D. W. (2000). The Relative
Importance of Perceived Ease-of-Use in IS Adop- Lee, J., & Huang, Y. (2005). The Design Factors of
tion: A Study of E-Commerce Adoption. Journal the Initial Trust in the Internet environment. Pro-
of the Association for Information Systems, 1(8), ceedings of World Conference on Educational
1-28. Multimedia, Hypermedia and Telecommunica-
tions, (pp. 993-997).
Gray, S. H. (1990). Using Protocol Analyses and
Drawings to Study Mental Model Construction Lingaard G. (1999). Does emotional appeal deter-
during Hypertext Navigation. International mine perceived usability of web sites? Retrieved
Journal of Human-Computer Interaction, 2(4), June 8, 2008 from http://www.cyberg.com
359-377
Marsh, T., & Wright, P. (2000). Maintaining the
Green, M. C. (2002). Perceptions of trust in In- illusion of interacting within a 3D virtual space.
ternet relationships. Computer-supported social Presence 2000: 3rd International Workshop on
interaction conference, Oxford, OH Presence, Delft, The Netherlands.
Handy, C.  (1995). Trust and the Virtual Organiza- Mayer, R. C., Davis, J. H., & Schoorman, F. D.
tion.  Harvard Business Review, 73(3), 40-50. (1995). An integrative model of organizational
trust. Academy of Management Review, 20(3),
Jarvenpaa, S. L., & Tractinsky, N. (1999). Con-
709-734.
sumer Trust in an Internet Store: A Cross-Cultural

1204
Building Trust in E-Commerce through Web Interface

McMahon, K. (2005). An exploration of the Resnick, P., Zeckhauser, R., Friedman, E., &
importance of website usability from a business Kuwabara, K. (2000). Reputation Systems. Com-
perspective. Retrieved July 1, 2008 from http:// munications of the ACM, 43(12), 45-48.
www.flowtheory.com/KTMDissertation.pdf
Riegelsberger, J., Sasse, M. A., & McCarthy, J.
Moon, J.-W., & Kim, Y.-G. (2001). Extending the D. (2003). Trust at First Sight? A Test of Users’
TAM for a World-Wide-Web Context. Information Ability to Identify Trustworthy e-Commerce
& Management, 38, 217-230. Sites. Proceedings of HCI2003, 8-12 243-260.
Nass, C., Steuer, J., Tauber, E., & Reeder, H. Rotter J. B. (1980). Impersonal trust, Trustwor-
(1993). Anthropomorphism, Agency, & Etho- thiness and gullibility. American Pschologist,
poeia: Computers as Social Actors. Conference 35(1), 1-7.
of the ACM/SIGCHI & the IFIP Amsterdam, the
Rust, R. T., & Kannan, P. K. (2003). E-Service:
Netherlands (pp. 24-29).
A new paradigm for business in the electronic
Nielsen, J. (1999). Trust or Bust: Communicating environment. Communications of the ACM,
Trustworthiness in Web Design. Jacob 46(6), 36-43.
Nielsen’s Alertbox. Retrieved April 10, 2008 from Utting, K., & N. Yankelovich (1989). Context
http://www.useit.com/alertbox/990307.html. and Orientation in Hypermedia Networks. ACM
Transactions on Information Systems, 7(1) 58-
Nielsen, J. (May 16, 1999). Who Commits the ‘Top
84.
Ten Mistakes’ of Web Design? Jacob Nielsen’s
Alertbox. Retrieved August 2, 2008 from http:// Wang, E., Barrett, S., Caldwell & Gavriel, S.
www.useit.com/alertbox/990516.html. (2003). Usability comparison: similarity and dif-
ferences between e-commerce and world wide
Neilsen Normen group (2000). Trust: Design
web. Journal of the Chinese Institute of Industrial
guidelines for e-commerce user experience.
Engineers, 20( 3), 258-266.
Retrieved September 6, 2008 from http://www.
nngroup.com Weiner, M., & Mehrabian, A. (1968). Language
within Language: Immediacy, a Channel in Verbal
Sheehan, K. B., & Doherty, C. (2001). Re-weaving
Communication. New York: Appleton-Century-
the Web: Integrating Print and Online Com-
Crofts.
munications. Journal of Interactive Marketing,
15(2), 47-59. Xiaoni, Z. (2003). Factors Contributing to Pur-
chase Intentions. Internet Journal of Internet
Radha, R., & Murphy, C. (1992). Searching Versus
Commerce, 2(1).
Browsing in Hypertext. Hypermedia. 4(1), 31.
Yoon, S. J. (2002). The antecedents and conse-
Reichheld, F. F., & Sasser, W. E. (1990). Zero
quences of trust in online purchase decisions.
defections: Quality comes to services. Harvard
Journal of Interactive Marketing, 16(2), 47-63.
Business Review, 68, 105-111.

This work was previously published in International Journal of Information Communication Technologies and Human Develop-
ment, Vol. 1, Issue 1, edited by S. Chhabra, pp. 64-74, copyright 2009 by IGI Publishing (an imprint of IGI Global).

1205
1206

Chapter 5.2
Swift Trust in Web Vendors:
The Role of Appearance and
Functionality1

Xin Li
University of North Carolina at Pembroke, USA

Guang Rong
Clemson University, USA

Jason B. Thatcher
Clemson University, USA

Abstract dents, we found appearance and functionality


features explained 61% of the variance in swift
With the growth of product search engines such trust. The article concludes with a discussion of
as pricegrabber.com, Web vendors have many findings and implications.
more casual visitors. This research examines how
Web vendors may foster “swift trust” as a means INTRODUCTION
to convert casual visitors to paying customers.
We examine whether perceptions of Web sites’ Trust is a key enabler of e-commerce (Lee &
appearance features (normality, social presence Turban, 2001; Kracher, Corritore, & Wiedenbeck,
and third-party links) and functionality features 2005). Extensive research has found that trust
(security, privacy, effort expectancy and perfor- leads to online end users reporting higher levels
mance expectancy) positively relate to swift trust of purchasing intention, expressing greater loyalty
in a Web vendor. Using a quasi-experimental to a Web vendor, and engaging in actual purchas-
research design, we empirically test the proposed ing behavior (Gefen, McKnight, Choudhury, &
relationships. Based on an analysis of 224 respon- Kacmar, 2002; Gefen, 2002b; Karahanna, &

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Swift Trust in Web Vendors

Straub, 2003a). Given the extensive support in the havior, this study examines the following research
literature for ties from online end users trust to question:
their behavior, it is important to direct attention to How do the appearance and functionality of a
identifying levers that foster trust and encourage Web site influence online consumers’ swift trust
online end users to complete transactions. in a Web vendor?
Fostering trust has proven problematic among The article unfolds as follows. First, we develop
fast-moving online end users. Although search a general model tying appearance and function-
engines direct over 85% of consumers to vendor ality to consumers’ swift trust and purchasing
Web sites, 57% of consumers abandon their shop- intention. In terms of appearance, we examine
ping carts prior to checkout (DoubleClick, 2004). normality, social presence, and links to third-party
Perhaps due to the ease of identifying alternatives assurance providers. In terms of functionality,
through engines such as pricegrabber.com, online we examine the roles of security, privacy, effort
consumers frequently visit Web sites and fail to expectancy and performance expectancy. Next,
complete transactions. Thus, Web vendors are we present the results of a quasi-experimental
under the pressure to quickly win trust, retain study that tests the proposed model. Then, we
first time visitors and convert them to paying discuss our findings. The article concludes with
customers. implications for research and practice.
In this article, we examine how Web vendors
may foster swift trust. Swift trust refers to trust
formed quickly in a new relationship (Meyerson, LITERATURE REVIEW AND MODEL
Weick, & Kramer, 1996). Unlike other forms DEVELOPMENT
of trust which rest on experience or familiarity
with a vendor, swift trust develops quickly dur- Trust is a complex and abstract concept. Although
ing a consumer’s first exposure to an unfamiliar defined differently in many literatures, trust most
Web vendor. Although first time visitors may commonly refers to one’s willingness to depend
lack experience or knowledge of a Web vendor, on another based on the expectation that the other
their perceptions of the “working conditions” of has the attributes to be trusted (Mayer, Davis, &
e-commerce, including appearance and function- Schoorman, 1995). Trust is driven by: 1) risk and
ality of a Web site (Zhang, von Dran, Small, & uncertainty in relationships, 2) the trusting party’s
Barcellos, 1999), influence their understanding vulnerability, and 3) his/her expectations of the
of a vendor’s ability to complete transactions trusted party. When these drivers are present, an
and its trustworthiness. For example, Gefen et al. individual must extend trust to another in order
(2003a) suggested that a fair and open Web site for a relationship to exist.
that clearly states due process, policies that handle E-commerce researchers extended the tra-
the relationship, and provides clear explanations, ditional trust definition and drivers to online
engenders consumer trust. Hence, designers em- shopping and found that they are important in
phasize a Web site’s appearance and functionality studying online end user behavior ( McKnight et
as means to promote swift trust. al., 2002; Gefen et al., 2003a) (see Table 1). For
Although substantial research has examined example, high uncertainty characterizes complex
the implications of trust in Web vendors, research and anonymous online transaction processes.
has left unexplored the influence of appearance While traditional customers can pay at the counter
and functionality on swift trust in a Web vendor. and take the products home immediately, online
To gain deeper insight into online consumer be- end users do not directly observe vendors nor

1207
Swift Trust in Web Vendors

Table 1. Extending trust definition and drivers to the domain of e-commerce


Traditional E-Commerce
The willingness of a party to be vulnerable to the actions of another party based on
the expectation that the other will perform a particular action important to the trustor,
Trust Definition
irrespective of the ability to monitor or control that other party (Mayer et al.., 1995 p.
712).
In the context of online shopping, the
Two parties—a trusting party
Prerequisites: relationship is predicated on the exchange
(trustor) and a party to be trusted
Trustor and Trustee in a of money for goods or services. Online con-
(trustee)—in a social relationship
relationship sumer is the trustor. Web vendor is the trustee
(Wang & Emurian, 2005)
(Wang & Emurian, 2005).
Trust is necessary in risky situations Because of high complexity and anonymity,
(Kee & Knox, 1970). When the trus- the e-commerce environment is more risky
tor has no control or even no means and uncertain than the traditional market
to monitor the trustee’s behaviors, (Wang & Emurian, 2005). Unobservability
Drivers of Trust: the trustee would have chances to causes even more information asymmetries
Risk and Uncertainty act opportunistically. The higher the than those in traditional markets (Tan &
risks and uncertainty exist, the more Thoen, 2002).
trust is needed to enable behaviors
in the context (Ruyter, Moorman, &
Lemmink, 2001).
Vulnerability refers to the extent to E-commerce customers are exposed to both
which a trustor would lose to the financial loss (e.g., credit card theft, defective
trustee’s opportunistic behaviors. or below-expectation product, delivery delay
The more a trustor will lose, the and damage, etc.) and privacy loss (e.g., iden-
Drivers of Trust:
more vulnerable s/he is in spe- tity theft, spam mails, etc.) (Friedman, Kahn,
Vulnerability
cific circumstances, and the higher & Howe, 2000; Gefen, 2002a).
level of trust is needed to justify the
participation in such circumstances
(Hosmer, 1995; Mishra, 1996).
Reasonable expectation refers to Online customers form expectations or
trustor’s belief that the trustee can beliefs of an e-vendor in terms of competence
be relied on. Usually a trustor will (ability of the vendor to do what the customer
Drivers of Trust: form a set of expectations about the needs), benevolence (the vendor’s caring and
Expectations trustee, including its competence, motivation to act in the customer’s interests)
benevolence and integrity (Mayer et and integrity (the vendor’s honesty and prom-
al., 1995). Based on the expecta- ise keeping) ( McKnight et al., 2002; Gefen
tions, the trustor grants trust. et al., 2003a).
Trust leads to risk-taking behavior. Consumers’ trust in an e-vendor leads to
The form of the action depends on their risk-taking behaviors, such as following
Outcomes of Trust the situation (Mayer et al., 1995). advice, providing information, and making
Trusting Behavior and Trust also leads to long-term purchases (McKnight et al., 2002). It also
Relationship relationships and anticipated future plays an important role in developing long-
interaction ( Mayer et al., 1995; term relationships, like enhancing customer
Doney & Cannon, 1997). loyalty.

directly view the actual products they purchase. make online purchases. For example, when pur-
Given the salience of risk and uncertainty in chasing a good, a consumer accepts exposure
online environments (Wang & Emurian, 2005), to the chance that an unfamiliar vendor could
understanding how to foster trust becomes es- misuse the credit card information. Consumers
sential for understanding how to encourage end who accept cookies also open themselves up to
users to purchase goods online. having their activities tracked and information
In light of risk and uncertainty, e-consumers stolen. When consumers perceive higher levels
accept some degree of vulnerability when they of vulnerability (McKnight et al., 2002), they

1208
Swift Trust in Web Vendors

must extend trust to justify entering a business limited interaction between a trustor and trustee,
relationship with a Web vendor. and rests on cues derived from the environment
Consumers extend trust based on their ex- (Meyerson et al., 1996; McKnight, Cummings,
pectations of a Web vendor. Expectations refer & Chervany, 1998) (see Table 2 for a comparison
to perceptions of a vendor’s trusting attributes of forms of trust). However, unlike initial trust,
(McKnight et al., 2002). Typically, online trust is swift trust is developed under time constraints due
conceived as a set of beliefs and expectations that to environmental pressures or consumers’ short
about a vendor’s: a) capability to operate a business attention span (Galleta, 2006). Jarvenpaa and
and complete transactions (i.e., competence); b) Leidner (1999) argue that swift trust is character-
desire to positively interact with consumers (i.e., ized by short time frames, which limit trustors’
benevolence); and c) honesty and adherence to ability to develop expectations of trustee. Once
widely accepted principles (i.e., integrity) (Mayer individuals form swift trust, they are inclined to
et al., 1995; McKnight et al., 2002; Gefen et al., pursue future interactions with a trustor.
2003a). When dealing with a new Web vendor, Swift trust is particularly germane to un-
consumers may not have enough direct knowledge derstanding online consumer behavior. Product
or experience. They have to build such expecta- search engines usually organize results based on
tions based on other sources. price. Frequently, low-cost vendors use low prices
to overcome a lack of name recognition. In this
Swift Trust case, online consumers do not have past experi-
ences or direct interactions with an unfamiliar
To retain quick moving consumers, vendors trustee—the Web vendor—suggested by a search
must foster swift trust. Swift trust refers to trust engine. Due to their numerous shopping choices,
formed quickly in new or transitory relationships the online consumers may expend limited time and
(Meyerson et al., 1996). It forms without the ben- effort exploring an unfamiliar Web site. To “close
efit of familiarity, past experience, and fulfilled the deal,” Web vendors must foster swift trust, or
promises. Swift trust is similar to initial trust in lose the new customers to competitors.
that it develops early in a relationship, reflects

Table 2. Developed trust, initial trust and swift trust


Definition Temporal Studies Trust Foundation
Trust built on shared In an established (Ganesan, 1994; Familiarity, Past
Developed experience and direct relationship Gefen et al., Experiences, Direct
trust interactions between the 2003a) Interactions
two parties.
Trust in an unfamiliar In a new rela- (McKnight et al., Disposition to Trust*,
trustee; a trust relationship tionship 2002; Koufaris & Institutional-Based
in which the actors do not Hampton-Sosa, Trust including Situ-
Initial trust yet have credible, meaning- 2004) ational Normality and
ful information about, or Structural Assurances
affective bonds with, each
other.
Trust formed quickly in a In a new rela- (Meyerson et al., Disposition to Trust*,
new or temporary system, tionship, with 1996; Jarvenpaa & Imported Trust and
without traditional sources time constraints Leidner, 1999) Category-Driven Trust
Swift trust
of trust such as familiar- based on quick cues
ity, past experience, and
fulfilled promises.

1209
Swift Trust in Web Vendors

Because individuals lack sufficient time to A Model of Swift Trust in Web


form perceptions of new trustees from scratch, Vendor
swift trust is usually “imported from other set-
tings and imposed quickly in categorical forms” Web design consists of two major dimensions:
(Meyerson et al., 1996, p. 174). To engender swift appearance and functionality (Zhang et al., 1999).
trust in online shopping, vendors embed cues in Appearance features refer to the look and feel of
the Web site design to facilitate the importing and a Web site. They give the first impression to the
categorization processes. Some cues provide links Web visitors. Functionality features are more
for consumers to import trust from other trusted task-related. They refer to a Web site’s ability to
sources (i.e., trust transference [Stewart, 2003]). supply information and conclude transactions.
For example, the presence of Visa and FedEx icons Here, we propose a model (see Figure 1) depict-
provides signal to consumers about safe payment ing how a set of appearance and functionality
and fast delivery. Other cues invoke categoriza- features induce swift trust and purchase intention.
tion processes to speed up the trust formation. In both appearance and functionality dimensions,
For instance, some vendors utilize page layout we examine features that have been previously
and menu sets similar with those known and studied and are often noted as salient to fostering
trusted Web sites to encourage new consumers to positive beliefs about a Web vendor and its Web
categorize them as familiar. Industry certificates site. While they are not exhaustive, the set of
or awards are usually placed prominently on the features will provide us some initial insight into
homepage for consumers to label the vendors as swift trust formation. In the following sections,
professional. More Web sites present BBB and we discuss these appearance and functionality
Trust-e seals, so consumers will classify them as features and their relationships with swift trust
reliable vendors. Thus, through embedding cues in and purchase intention.
Web design, new vendors can successfully foster
swift trust among first time visitors. Swift Trust and Purchase Intention

Purchase intention refers to online consumers’


willingness to purchase from a Web vendor

Figure 1. A model of swift trust in Web vendor

Appearance

Perceived H2
Normality Swift Purchase
H1
H3
Trust Intention
Perceived Social
Presence
H4

rd H5 H6 H7a H8a H7b H8b


Awareness of 3 -
Party Assurance

Perceived Perceived Effort Performance


Security Privacy Expectancy Expectancy

Functionality

1210
Swift Trust in Web Vendors

(Stewart, 2003). Prior research suggests that al., 2003a). Normality of a Web site signals to new
purchase intention is influenced by consumers’ consumers that the vendor will fulfill their expec-
trust in a Web vendor (Stewart, 2003; van der tations of a good shopping experience (i.e., product
Heijden, Verhagen, & Creemers, 2003; Gefen et information can be easily acquired, transactions
al., 2003a). Although formed quickly at the early are completed smoothly and safely, and expected
stage of a relationship, swift trust encourages fur- goods and services are received quickly) (Gefen et
ther interactions and facilitates the development al., 2003a). If a Web site is perceived as normal,
of purchase intention. If consumers grant swift consumers will categorize it as trustworthy and
trust towards a Web vendor, they tend to believe extend swift trust. Alternatively, an unusual Web
that this vendor will protect their privacy, sell site design that is not well-known and requires
high-quality products, and be capable of consum- learning new ways to access information may
mating online transactions. Thus, we hypothesize not be trusted.
that as swift trust in a vendor grows, consumers When visiting a new Web site, consumers
will report higher purchase intentions. assess normality through their quick interaction
with the Web site. Specifically, they evaluate
H1: Swift trust in a Web vendor positively affects their perceptions of: 1) the shopping process, 2)
purchase intention. the information required to complete a purchase,
and 3) the quality of their interaction with the
Appearance Features Web site (Gefen et al., 2003a). For example, most
consumers are familiar with one-click shopping.
A Web site is an electronic storefront of a Web If, on a new Web site, consumers find that they
vendor. For a pure e-commerce company, the Web must go through a complicated process and pro-
site is the only access point through which con- vide unusually detailed information to complete
sumers may collect information on, and interact a transaction, or they have a complicated or un-
with, the goods and services. Because of the Web pleasant communication with online representa-
site’s centrality in the consumer-Web vendor rela- tives, they will not trust the Web vendor and fail
tionship, a Web site’s appearance is an important to complete a transaction. Thus, we anticipate
source of signals about the vendor’s trustworthi- that consumers who perceive normality of a Web
ness. Prior trust literature has mentioned several site are more likely to form swift trust towards
appearance features that affect trust (McKnight a Web vendor.
et al., 2002; Gefen et al., 2003a; Gefen & Straub,
2004). Based on the literature, we believe online H2: Perceived normality positively affects swift
consumers will express swift trust when they trust in Web vendor.
perceive a Web site’s appearance as normal and
high on social presence, and they are aware of Perceived Social Presence
third-party assurances of trustworthiness. In the Perceived social presence refers to consumers’
following paragraphs, we explain the logic behind perception that there is personal, sociable, and
these relationships. sensitive human contact and/or peer community
on the Web site (Gefen & Straub, 2004). A Web
Perceived Normality site can project a social presence similar to what
Perceived normality refers to consumers’ per- consumers experience in off-line stores. Gefen
ception that a Web site performs in a normal or and Straub (2004) found that when a Web site
customary manner, which is consistent with their is perceived as having a higher level of social
experience with similar Web vendors (Gefen et presence, it enables more effective interaction

1211
Swift Trust in Web Vendors

and communication between a consumer and online transactions. Through prominently placing
Web vendor, and thus, renders consumers more seals and icons of these mechanisms, a Web site
inclined to trust the Web vendor. By projecting encourages consumers to extend trust based on
social presence similar to off-line stores, a Web their experiences with the established institutional
site may induce consumers to categorize the Web structures. Based on the prior studies (McKnight
site as part of a trustworthy group of known, off- et al., 2002; Stewart, 2003), consumers who are
line vendors. quickly aware of the third-party assurance on
In practice, although Web vendors are not an unfamiliar Web site are more likely to extend
able to use salespersons or store assistants to swift trust towards the Web vendor.
physically interact with customers, they send
cues or project social presence through images H4: Awareness of the third-party assurance posi-
of high-quality products, happy customers, and tively affects swift trust in Web vendor.
live-chat customer service functions. For example,
on Overstock.com, service representatives use Functionality Features
interactive, pop-up windows to ask online con-
sumers if they need more information, or offer Online customers’ impression of a Web site’s
to help complete a transaction. Many Web sites functionality, that is, its ability to successfully
foster virtual communities, where customers can complete a transaction, influences their trust in
participate in peer discussions, provide product/ the Web vendor. Although deep understanding
store reviews, and receive peer assistance (e.g., of functionality requires extensive interaction
newegg.com and target.com). If, during the first with a Web site, cues in the Web site design may
visit, consumers perceive a Web site as high on contribute to the formation of swift trust. Based
social presence, we anticipate they will extend on prior research, we direct our attention to four
swift trust to a Web vendor. perceptions of functionality that may influence
swift trust: security, privacy, effort expectancy
H3: Perceived social presence positively affects and performance expectancy (Gefen et al., 2003a;
swift trust in Web vendor. Thatcher & George, 2004).

Awareness of Third-Party Assurance Perceived Security


Awareness of third-party assurance refers to Perceived security refers to consumers’ perception
consumers’ knowledge and awareness of third- that a Web site can safely complete transactions
party institutional mechanisms that ensure a (Pavlou & Chellappa, 2001). When considering
trustworthy environment for Web site operations a transaction with an unknown party, consumers
and business transactions (McKnight et al., 2002). will evaluate the vendor’s ability to protect their
Several institutional mechanisms are widely used information, successfully complete a transaction,
in e-commerce, including seal programs (e.g., and offer help resolving a dispute. In the online
Trust-e reliability and privacy seal programs, environment, consumers usually perceive addi-
VeriSign security seal program), credit card pro- tional security threats than the traditional consum-
tections (e.g., Visa card security and protection ers, such as credit card theft, defective products/
programs, MasterCard security and credit basics), services, and delivery failure (Featherman &
and customer certified store ratings and feedback Pavlou, 2003). Consumers’ security concerns are
systems (e.g., BizRate.com) (Pavlou & Gefen, the major determinant of their online perceptions
2004). Prior research suggests that these mecha- and behaviors (Salisbury, Pearson, Pearson, &
nisms provide a safe and reliable environment for Miller, 2001).

1212
Swift Trust in Web Vendors

Consumers’ assessment of the Web site’s abil- data may be used by the firm (Gefen et al., 2003a;
ity to safely complete transactions contributes Gefen, Rao, & Tractinsky, 2003b). When consum-
to the formation of trust (Zhou, Dai, & Zhang, ers quickly recognize that a Web vendor complies
2007). For example, Amazon.com provides no- with the fair information practices and commits
tices when users sign in using the secure server. to protect its customers’ privacy, they are more
It gives customers easy access to the shipping likely to express swift trust. Hence:
and return policies on its homepage. By doing
so, Amazon.com sends signals about its ability to H6: Perceived privacy positively affects swift
facilitate safe, secure online shopping. Similarly, trust in Web vendor.
many Web vendors send signals about security
through providing order status, real-time ship- Effort Expectancy and Performance
ping tracking, and return assistance. When first Expectancy
time visitors feel fewer security concerns on a Within the domain of e-commerce, effort expec-
Web site, they will be more likely to report swift tancy and performance expectancy have been
trust in the vendor. tied to trust and purchase intentions (Pavlou,
2003; Gefen et al., 2003a). Effort expectancy (also
H5: Perceived security positively affects swift referred to as ease of use [Venkatesh, Morris,
trust in Web vendor. Davis, & Davis, 2003]) refers to the degree to
which the consumers believe that using a new Web
Perceived Privacy site would be free of effort (Gefen et al., 2003a).
Perceived privacy refers to the consumers’ belief Performance expectancy (also referred to as per-
that a Web vendor will protect consumers’ personal ceived usefulness [Venkatesh et al., 2003]) refers
and financial information (Pavlou & Chellappa, to the degree to which the consumers believe that
2001). An online transaction requires consumers using a new Web site would enhance their ability
to provide more personal and financial information to safely, effectively purchase goods (Gefen et al.,
than traditional shopping channels. For example, 2003a). Gefen et al. (2003a) found that customers’
online transactions require providing billing and effort expectancy in using a Web site influences
shipping addresses, something that is not typi- their trust. Their findings also indicate that trust,
cally requested by brick-and-mortar stores. Also, effort expectancy, and performance expectancy
e-commerce companies often use cookies to col- influence customers’ usage intentions towards
lect information and provide customized recom- the Web site. Consistent with prior research, we
mendations to consumers. Due to demands for hypothesize that when a new Web site is perceived
consumer information, as well as a Web vendor’s as low on effort expectancy, consumers will be
ability to collect information without consumers’ more likely to report swift trust as well as inten-
notice, privacy concerns have become a defining tions to use a Web site. Hence:
issue for many Web consumers (Malhotra, Kim,
& Agarwal, 2004). H7a: Effort expectancy negatively affects swift
When considering a transaction with an unfa- trust in Web vendor.
miliar store, privacy may be a salient issue for a
consumer (Liu, Marchewka, Lu, & Yu, 2005). A H7b: Effort expectancy negatively affects pur-
Web site may address privacy concerns through chase intentions.
prominently featuring links to the privacy notice
or policies, by briefly stating how they collect In a departure from Gefen et al. (2003a), we
data on consumers, and by illustrating how such argue that performance expectancy positively

1213
Swift Trust in Web Vendors

influences swift trust. Because first time visitors Study Procedure


lack extensive experience with the tools provided
by a Web vendor, expectations about performance The study consisted of a task and a survey. Partici-
reflect how they interpret cues embedded in a pants were assigned a shopping task on Overstock.
Web site. For example, a standard “useful” fea- com. Specifically, subjects were asked to shop for
ture of a shopping site is the ability to search for a birthday gift for their significant others. They
products. Similarly, a comparison tool can help were given 10 minutes to view the Web pages
consumers compare similar products in terms and research a purchase. No actual purchase was
of price, specifications and features in a more required in the shopping task. Following comple-
effective way. When consumers perceive that a tion of the task, subjects were asked to complete
Web site possesses tools that enable searching for a survey about their experiences on Overstock.
and acquiring products, they will be more likely com. The survey collected data on all constructs
to express positive expectations of performance. in the research model and demographic informa-
Given that swift trust reflects limited interaction tion. Respondents were required to complete the
with a Web site, we anticipate that the presence task and survey in 25 minutes, thus engendering
of these “useful” tools positively influences swift time pressure characteristic of situations involv-
trust and purchase intention. Hence: ing swift trust.

H8a: Performance expectancy positively affects Measurement Development


swift trust in Web vendor.
Where possible, we adapted measures from prior
H8b: Performance expectancy positively affects online trust research. Each item was measured us-
purchase intention. ing a seven-point Likert-type scale (anchored with
1=strongly disagree, 7=strongly agree). Purchase
intention was measured using four items adapted
METHOD from Pavlou’s work (Pavlou, 2003; Pavlou & Ge-
fen, 2004) (see Appendix I for the items).
A quasi-experimental survey design was used to Swift trust was operationalized as a second-
test the research model. We used Overstock.com order construct. It comprises three dimensions:
as our target Web site as it is a pure e-commerce competence, benevolence, and integrity. Each
company. Subjects would not confound trust dimension was measured using three to four items
in Overstock.com with their experiences at a adapted from McKnight et al. (2002).
real-world brick-and-mortar or click-and-mortar Measures of most appearance features were
business. adapted from prior studies. Perceived normality
We collected data at three public universities was measured by items adapted from Gefen et al.
in the United States. Junior- and senior-level col- (2003a). Perceived social presence was measured
lege students participated in the study. Although by items adapted from Gefen and Straub (2004).
student samples have been criticized in many IS To measure awareness of third-party assurance, we
research contexts, evidence suggests that college developed four items that directed attention to the
students are active online shoppers, who do not dif- presence of icons and links to trusted third parties
fer from their working peers (Sen, King, & Shaw, such as VeriSign security seal program or credit
2006), and thus, represent a useful sample for card partners that appear on Overstock.com.
testing our research model (Suh & Lee, 2005). Functionality features measures were adapted
from prior research. Effort expectancy and perfor-

1214
Swift Trust in Web Vendors

mance expectancy were adapted from Gefen et al. RESULTS


(2003a). Perceived security and perceived privacy
were adapted from Salisbury et al. (2001), Yang We used partial least square (PLS), a structural
and Jun (2002), and Smith et al. (1996). equation modeling technique, to evaluate the
hypotheses. Although a debate is emerging in
Pilot Study and Data Collection the MIS literature on the merits of PLS, in the
broader methods literature, PLS is considered
We conducted a pilot study to validate the mea- a useful tool for theory building, specifically
surement instruments and our quasi-experiment when one seeks to establish predictive validity
design. The instrument and task were well received in a structural model (Chwelos, Benbasat, &
by the pilot study participants. Minor changes were Dexter, 2001). As a result, we believe PLS is an
made in wording and the order of items. Also, we appropriate tool for estimating our measurement
received feedback that our task was meaningful and structural models. In the following sections,
and induced a feeling of time pressure among we present the results in two steps: measurement
our respondents. model and structural model.
Two-hundred eighty-one students participated
in this study. We dropped 14 responses due to Measurement Model
missing values. Also, we dropped 43 subjects
because they reported prior experience with Two measurement models were evaluated. Follow-
Overstock.com. This yielded a usable sample of ing Agarwal and Karahanna’s approach (2000),
224 respondents that included 118 females and we first evaluated the measurement model for
106 males, with substantial Internet and online swift trust—a second order construct. Swift trust
shopping experience. T-tests revealed that sample is formed by three reflective sub-dimensions:
characteristics and responses across research sites competence, benevolence and integrity. All items
were not significantly different. Table 3 presents yielded high loadings (ranging from 0.855 to 0.948,
sample characteristics. p < 0.000) on the appropriate sub-dimension. Also,
each sub-dimensions’ internal consistency score
(ranging from 0.908 to 0.941) and composite reli-

Table 3. Sample characteristics


Char. Number Percentage Char. Number Percentage
Gender Internet Experience – Years of Use
Female 118 52.7% 0-3 years 5 2.2%
Male 106 47.3% 4-7 years 103 46.0%
Age 8 years or above 116 51.8%
Under 21 85 37.9% Internet Experience - Often
21 - 45 133 59.4% Once a day or less 16 7.1%
Above 45 6 2.7% Many times a day 208 92.9%
Internet Level Online Shopping Experience – in past 2 years
Beginner 2 0.9% 0-2 times 31 13.8%
Intermediate 34 15.2% 3-10 times 67 29.9%
Proficient 188 83.9% 11 time or above 126 56.3%

1215
Swift Trust in Web Vendors

Table 4. Swift trust: Means, reliability and average variance extracted1


Dimension Means StDev CR Alpha 1 2 3
1. Competency 5.516 0.949 .952 .931 .832
2. Benevolence 5.022 1.095 .943 .908 .707 .848
3. Integrity 5.142 1.010 .958 .941 .698 .797 .852
Diagonal elements in the ‘correlation of constructs’ matrix are the square root of the average variance extracted. For adequate
1

discriminant validity, diagonal elements should be greater than corresponding off-diagonal elements.

Table 5. Overall measurement model: Means, reliability and average variance extracted1

.864
9

.926

.313
8

.921

.349

.087
7

.274

.145
.933

.583
6

.392

.545

.184
.375
.872
5

.918

.338

.361

.042
.360

.119
4

.501

.185
.942

.070

.717
.400

.316
3

.596
.530
.907

.538

.396

.520
.635

.211
2

.413

.064
.364

.566
.928

.533

.340

.341
.414
1
Alpha

.944

.932
.953

.940
.944

.892

.935

.949
.894

.960

.947
.957
.961

.933

.959

.964

.964
.927
CR

1.324

1.311
1.144
1.348

0.924

1.028

1.351

0.921
1.057
StDev

4.479

2.908
5.498
5.008

5.227

5.961

4.231

6.221
5.056
Means

5. Awareness of
Social Presence

7. Performance
1. Transaction

Online Store

8. Perceived

9. Perceived
3. Perceived

4. Perceived

Expectancy
Expectancy
Normality
2. Trust in
Intention

3rd-party
6. Effort

Security

Privacy

Diagonal elements in the ‘correlation of constructs’ matrix are the square root of the average variance extracted. For adequate
1

discriminant validity, diagonal elements should be greater than corresponding off-diagonal elements

1216
Swift Trust in Web Vendors

ability score (ranging from 0.943 to 0.958) were higher on the construct of interest than other
high. The detailed item loadings and reliability constructs. Every construct has square root of
scores are reported in Appendix II (a) and Table AVE (ranging from 0.864 – 0.942) greater than
4, respectively. a confirmatory factor analysis its correlations with any other constructs. Hence,
was performed to assess the convergent and dis- our analysis indicates adequate convergent and
criminant validity of the measures (see Appendix discriminant validity for the overall measure-
II (a) and Table 4). Briefly, all items load higher ment model.
on the corresponding sub-dimensions than on
other sub-dimensions, and each sub-dimension Structural Model
has a square root of the AVE (ranging from 0.832
to 0.852) greater than its correlations with other Figure 2 presents the structural model results.
sub-dimensions. Hence, our analysis indicates Appearance and functionality features explain
adequate reliability, convergent, and discriminant 61% of the variance in swift trust. Also, swift
validity of the swift trust measurement model. trust, effort expectancy, and performance expec-
Next, following Agarwal and Karahanna’s tancy explained 40% of the variance in purchase
approach (2000), we used the factor scores of the intention.
above three sub-dimensions as direct measures of
swift trust to assess the full measurement model. Swift Trust
Results are presented in Table 5 and Appendix II
(b). All indicators load on the construct of inter- In terms of appearance features, we found per-
est, with loadings ranging from 0.829 to 0.965 (p ceived normality (H2: .19, p<.01), perceived social
< 0.000). Internal consistency scores range from presence (H3: .13, p<.01), and awareness of 3rd
0.892 to 0.953. The composite reliability scores party assurance (H4: .28, p<.01) positively relating
range from 0.927 to 0.964. All indicators loaded to swift trust. In terms of functionality features,

Figure 2. Research model with PLS results

Appearance
Perceived
Normality
H2: .19**

Perceived Swift Purchase


Social Presence H3: .13** Trust H1: .35** Intention
R2 = .61 R2=. 40

H4: .28**
Awareness of
3rd-Party
Assurance H5: .26** ns H7a: -.13* ns H7b: -.12* H8b.44**

Perceived Perceived Effort Performance


Security Privacy Expectancy Expectancy

Functionality

1217
Swift Trust in Web Vendors

perceived security (H5: .26, p<.01) positively relat- uct lines/business models, our findings suggest
ed to swift trust while effort expectancy (H7a: -.13 patterning Web sites after successful e-commerce
p<.05) negatively related to swift trust. Perceived businesses in their market as a means to engender
privacy (H6: n.s.) and performance expectancy swift trust for first time visitors.
(H8a: n.s.) were not significantly related to swift The study also shows that social presence posi-
trust. Collectively, our results provide evidence tively related to swift trust in a Web vendor (H3:
that a Web site’s appearance and functionality Supported). An unfamiliar Web site that projects
attributes influence swift trust. social presence similar to real-world salespersons
and customer service representatives engenders
Purchase Intention feelings of competence, benevolence, and integ-
rity in online consumers (Gefen & Straub, 2004).
The overall model explains 40% of the variance Overstock.com uses images of satisfied customers
in purchase intention. Swift trust (H1: .36, p < and content employees to convey social presence.
0.01) and performance expectancy (H8b: .44, p In addition, when customers check a product, a
< 0.01) positively relates to purchase intention. live assistance chat window frequently pops up
Effort expectancy (H7b: -.12, p < 0.05) negatively and offers to provide more information or help
relates to purchase intention. completing a transaction. Further, below the
item description, customers can read the product
reviews posted by other customers or post their
DISCUSSION own reviews. All of these features give consumers
the feeling like they were shopping at a physical
This study examined sources and consequences store with helpful service representatives, and they
of swift trust in Web vendors. In the following were part of a community of shoppers (Thatcher
paragraphs, we discuss our findings and the & George, 2004). Thus, we suggest prominently
implications. projecting social presence as a useful means to
foster consumers’ swift trust.
Appearance Features and Swift Trust First-time visitors reported higher levels of
swift trust when they were aware of the third-
All of the three appearance features that we party assurance (H4: Supported). Overstock.com
examined positively relate to swift trust. The places icons of third-party assurance providers
study findings show that perceived normality on every Web page—including icons from credit
has a significant positive relationship with swift card companies, payment options, VeriSign BBB
trust (H2: Supported). Similar to other groups of Online, and BizRate.com. Our findings confirm
online consumers, our respondents had significant that third-party icons signal that Overstock.com
Internet and e-commerce experience. Among our is operated in a safe and reliable environment,
study participants, similarity with familiar Web and the transactions in the environment are well
sites engendered their swift trust in an unfamiliar protected (McKnight, Kacmar, & Choudhury,
Web site. This finding suggests that Web design- 2004). For practice, this finding directs Web
ers should carefully consider how they visually designers to carefully consider the placement of
and structurally differentiate their Web site from third-party assurance icons and seals. If properly
competitors. If a Web site employs an unusual or sized, placed at directly visible positions, repeated
novel design, first time visitors may not extend on different pages or different sections, the third-
trust to a Web vendor. For new Web vendors, in party assurance icons engender consumers’ swift
addition to relying on low price or novelty of prod- trust (McKnight et al., 2004).

1218
Swift Trust in Web Vendors

Overall, appearance features are salient to un- the presence of the security features to online
derstanding swift trust. This is probably because consumers.
the appearance features are mostly visible to the Contrary to our expectations, perceived pri-
first-time Web site visitors under time pressure. vacy was not significantly related to swift trust
Our findings suggest that consumer perceptions (H6: not supported). This finding may be a func-
of normality, social presence, and third-party tion of the task we used in our study. Participants
assurances lead to swift trust in a Web vendor. searched for information and assessed features
For Web designers, these findings suggest that of a Web site; but they did not provide personal
the look and feel of a Web site can be used to information required to complete a transaction.
foster swift trust and convert first-time visitors Perhaps due to the short and anonymous visit,
to paying customers. our participants were not concerned with privacy
issues associated with completing a purchase on
Functionality Features and Swift Overstock.com. Because swift trust develops prior
Trust to making a purchase, we are comfortable with the
strength of our research design. However, future
We found mixed results for functionality features’ research should examine ties between swift trust
relationship with swift trust. Perceived security and an antecedent to privacy concerns. Hence,
positively related to swift trust (H5: supported). even though we did not find a relationship between
Consumer perceptions of a Web site’s capability privacy and swift trust, there remains room for
to perform secure transactions influence their future research on privacy and swift trust.
perceptions of the Web vendors’ competence, Effort expectancy negatively related to swift
benevolence, and integrity. Fully understanding trust (H7a: Supported) and purchase intention
the security features on a Web site may require (H7b: Supported). When using a Web site was
time and extensive interaction. However, Web perceived as relatively free of effort, consumers
vendors can quickly signal their intention and were more likely to perceive the Web vendor as
ability to provide a safe shopping experience to competent, benevolent and high on integrity.
the first time visitors through security cues. For While consistent with prior online consumer
example, Overstock.com provides prominent links behavior research (Gefen et al., 2003a), our find-
to order tracking, shipping and return policies on ings confirm the nomological net of relationships
its homepage. The transparent transaction pro- between effort expectancy, trust, and purchase
cesses may create a sense of security for online intention. For practice, our findings underscore
consumers. In addition, the “golden lock” signs the importance of designing Web sites to be easy
on the login and check-out pages, as well as the to navigate, easy to find product information, and
timeouts and re-login requests, also provide quick easy to conclude transactions.
signals of a Web site’s security. If these security Performance expectancy was not related
cues are prominently placed, first-time visitors to swift trust (H8a: Not Supported) and it was
can quickly perceive good protection and make related to purchase intention (H8b: Supported).
quick inferences about the trustworthiness of a Performance expectancy’s influence on swift trust
Web vendor. While most Web vendors expend may have been washed out by perceived normal-
significant efforts in developing security features ity, perceived social presence, perceived security,
that consumers may not directly see or understand, and effort expectancy. While other perceptions
we suggest, to foster swift trust, Web vendors refer to specific aspects of a Web site, perfor-
should give special attentions to how to signal mance expectancy requires consumers to make
global assessments of a Web site. In light of this,

1219
Swift Trust in Web Vendors

it makes sense that perceptions of specific Web a Web vendor. For practice, although the casual
site attributes would exert greater influences on visitors led by search engines may have many
swift trust than a global assessment of a Web site. alternatives and move fast, Web vendors still can
An alternative explanation for this finding may be successfully foster swift trust and convert them
that it requires consumers’ substantial experience to paying customers through the proper use of the
to form perceptions of performance expectancy. appearance and functionality features.
Because swift trust reflects only limited interac-
tion with a Web site, online consumers may lack
the experience to form beliefs about performance CONCLUSION
expectancy. Thus, it is not surprising that perfor-
mance expectancy is not an antecedent to swift This study examined swift trust in Web vendors.
trust in a Web site. However, purchasing a product To induce conditions necessary for the formation
rests on general, well-developed perceptions of of swift trust, we constrained study participants’
a Web site’s functionality. Hence, performance time to visit a Web site. Also, we dropped responses
expectancy, as expected, directly influences from study participants with experience with our
purchase intention (Gefen et al., 2003a). target Web site—Overstock.com. Based on our
Overall, online consumers report higher levels study, we found substantial support for appearance
of swift trust when they report positive perceptions and functionality features positively relating to
of a Web site’s security and low levels of effort to swift trust. Also, we found supports for swift trust,
use. The other functionality features may require effort expectancy, and performance expectancy
more interactions to be realized and contribute influencing purchase intention.
to trust formation. Although perceptions of the For research, this study provides initial evi-
features may evolve over time, Web vendors who dence that swift trust has different antecedents
embed noticeable signals that they are capable to than other forms of experience-based trust iden-
complete secure, easy transactions are likely to tified in the e-commerce literature—it was not
engender consumers’ swift trust and purchasing affected by perceived privacy or performance
intention in their Web sites. expectancy. This suggests that future research
should examine additional factors that shape
Swift Trust and Purchase Intention the development of swift trust. For practice, our
research underscores the importance of fostering
Swift trust positively relates to purchase intention swift trust as a means to increase purchase inten-
(H1: Supported). When online consumers quickly tion. Our findings direct attention to approaches
form trust, they are more likely to participate in under managerial control such as normality (i.e.,
further interaction with the Web site and eventu- using Web site designs similar to known vendors)
ally make actual purchases from the vendor. This and social presence (i.e., engendering a sense of
finding is consistent with prior trust research (Ge- community or having interactive help features)
fen, 2002b; see Gefen et al., 2003a; 2003b; Pavlou that positively influence the formation of swift
& Gefen, 2004). Our finding also supports the idea trust.
that trust is relevant to purchasing intention even In light of findings, it is important to note sev-
when consumers have limited time to familiarize eral limitations and opportunities for extending
themselves with a Web vendor. Hence, this find- our research. An important limitation of this study
ing underscores the importance of the previous is the use of a single Web site, Overstock.com. We
findings on how appearance and functionality used this approach to control for variance in how
features influence the formation of swift trust in a Web site’s features and respondents’ experience

1220
Swift Trust in Web Vendors

influence swift trust. Overstock.com possesses Friedman, B., Kahn, P., & Howe, D. (2000). Trust
many favorable features such as recommendation online. Communications of the ACM, 43(12),
agents, icons, and product descriptions that may 34-40.
foster swift trust. In future research, it would be
Galleta, D., Henry, R., McCoy, S., & Polak, P.
useful for researchers to examine how manipu-
(2006). When the wait isn’t so bad: The interacting
lating these features would influence consumers’
effects of Web site speed, familiarity, and breadth.
swift trust in other unfamiliar Web vendors.
Information Systems Research, 17(1), 20-37.
Also, future research should examine how to
foster swift trust in diverse populations of online Ganesan, S. (1994). Determinants of long-term
consumers. In this research, we drew on students orientation in buyer-seller relationships. Journal
who are active online consumers to examine swift of Marketing, 58(2), 1-19.
trust. Although students are a valid sample for the
Gefen, D. (2002a). Nurturing clients’ trust to en-
current study, a pure student sample may include
courage engagement success during the customi-
less variance in age, occupation, and Internet
zation of ERP systems. Omega, 30, 287-299.
experience. For example, research suggests that
age influences perceptions of technology among Gefen, D. (2002b). Customer loyalty in e-com-
members of the workforce (Morris & Venkatesh, merce. Journal of Association for Information
2000). Hence, future research should examine Systems, 3, 27-51.
differences in demographic factors such as age,
Gefen, D., Karahanna, E., & Straub, D. (2003a).
income or occupation, which may relate to the
Trust and TAM in online shopping: An integrated
formation of swift trust in e-commerce.
model. MIS Quarterly, 27(1), 51-90.
Gefen, D., Rao, V., & Tractinsky, N. (2003b).
References The conceptualization of trust, risk and their re-
lationship in electronic commerce: The need for
Agarwal, R., & Karahanna, E. (2000). Time flies clarifications. Paper presented at the 36th Hawaii
when you’re having fun: Cognitive absorption and International Conference on System Sciences.
beliefs about information technology usage. MIS Big island, Hawaii.
Quarterly, 24(4), 665-694.
Gefen, D., & Straub, D. (2004). Consumer trust
Chwelos, P., Benbasat, I., & Dexter, A. (2001). in B2C e-commerce and the importance of so-
Research report: Empirical test of an EDI adop- cial presence: Experiments in e-products and
tion model. Information Systems Research, 12(3), e-services. Omega, 32, 407-424.
304-321.
Hosmer, L. (1995). Trust: The connecting link
Doney, P., & Cannon, J. (1997). An examination between organizational theory and philosophical
of the nature of trust in buyer-seller relationships. ethics. Academy of Management Review, 20(2),
Journal of Marketing, 61, 35-51. 379-403.
DoubleClick. (2004). DoubleClick Q3 2004 ad Jarvenpaa, S., & Leidner, D. (1999). Communica-
serving trend report. Doubleclick.com. tion and trust in global virtual teams. Organization
Science, 10(6), 791-815.
Featherman, M., & Pavlou, P. (2003). Predict-
ing e-services adoption: A perceived risk facets Kee, H., & Knox, R. (1970). Conceptual and
perspective. International Journal of Human- methodological considerations in the study of
Computer Studies, 59(4), 451-474. trust and suspicion. Journal of Conflict Resolu-
tion, 14(3), 357-366.

1221
Swift Trust in Web Vendors

Koufaris, M., & Hampton-Sosa, W. (2004). The Meyerson, D., Weick, K., & Kramer, R. (1996).
development of initial trust in an online company Swift trust and temporary groups. In: R. Kramer
by new customers. Information and Management, & T. Tyler (Eds.), Trust in organizations: Frontiers
41, 377-397. of theory and research, (pp. 166-195). Thousand
Oaks, CA: Sage Publications.
Kracher, B., Corritore, C., & Wiedenbeck, S.
(2005). A foundation for understanding online Mishra, A. (1996). Organizational responses to
trust in electronic commerce. Information, crisis: The centrality of trust. In: R. Kramer & T.
Communications and Ethics in Society, 3(3), Tyler (Eds.), Trust in organizations, (pp. 261-287).
131-141. Newbury Park, CA: Sage Publications.
Lee, M., & Turban, E. (2001). A trust model for Morris, M., & Venkatesh, V. (2000). Age dif-
consumer internet shopping. International Jour- ferences in technology adoption decisions: Im-
nal of Electronic Commerce, 6(1), 75-91. plications for a changing workforce. Personnel
Psychology, 53(2), 375-403.
Liu, C., Marchewka, J., Lu, J., & Yu, C. (2005).
Beyond concern—a privacy-trust-behavioral Pavlou, P. (2003). Consumer acceptance of elec-
intention model of electronic commerce. Informa- tronic commerce: Integrating trust and risk with
tion and Management, 42, 289-304. the technology acceptance model. International
Journal of Electronic commerce, 7(3), 101-134.
Malhotra, N., Kim, S., & Agarwal, J. (2004).
Internet users’ information privacy concerns Pavlou, P., & Chellappa, R. (2001). The role of
(IUIPC): The construct, the scale, and a causal perceived privacy and perceived security in the
model. Information Systems Research, 15(4), development of trust in electronic commerce
336-355. transaction. Unpublished work.
Mayer, R., Davis, J., & Schoorman, F. (1995). An Pavlou, P., & Gefen, D. (2004). Building effective
integrative model of organizational trust. Academy online marketplaces with institution-based trust.
of Management Review, 20(3), 709-734. Information Systems Research, 15(1), 37-59.
McKnight, D., Choudhury, V., & Kacmar, C. Ruyter, K., Moorman, L., & Lemmink, J.
(2002). Developing and validating trust measures (2001). Antecedents of commitment and trust in
for e-commerce: An integrative typology. Infor- customer-supplier relationships in high technol-
mation Systems Research, 13(3), 334-359. ogy markets. Industrial Marketing Management,
30, 271-286.
McKnight, D., Cummings, L., & Chervany, N.
(1998). Initial trust formation in new organiza- Salisbury, D., Pearson, R., Pearson, A., & Miller,
tional relationships. Academy of Management D. (2001). Perceived security and World Wide
Review, 23(3), 473-490. Web purchase. Industrial Management and Data
Systems, 101(3-4), 165-176.
McKnight, D., Kacmar, C., & Choudhury, V.
(2004). Shifting factors and the ineffectiveness of Sen, R., King, R., & Shaw, M. (2006). Buyers’
third party assurance seals: A two-stage model of choice of online search strategy and its managerial
initial trust in a Web business. Electronic Markets, implications. Journal of Management Information
14(3), 252-266. systems, 23(1), 211-238.

1222
Swift Trust in Web Vendors

Smith, H., Milberg, S., & Burke, S. (1996). Infor- Wang, T., & Emurian, H. (2005). An overview of
mation privacy: Measuring individuals’ concerns online trust: Concepts, elements, and implications.
about organizational practices. MIS Quarterly, Computers in Human Behavior, 21, 105-125.
20(2), 167-196.
Yang, Z., & Jun, M. (2002). Consumer perception
Stewart, K. (2003). Trust transfer on the World of e-service quality: From Internet purchaser and
Wide Web. Organizational Science, 14(1), 5-17. non-purchaser perspectives. Journal of Business
Strategies, 19(1), 19-41.
Suh, K., & Lee, Y. (2005). The effects of virtual
reality on consumer learning: An empirical in- Zhang, P., von Dran, G., Small, R., & Barcellos,
vestigation. MIS Quarterly, 29(4), 673-697. S. (1999). Web sites that satisfy users: A theoreti-
cal framework for Web user interface design and
Tan, T., & Thoen, W. (2002). Formal aspects of a
evaluation. Paper presented at the 32nd Hawaii
generic model of trust for electronic commerce.
International Conference on System Sciences.
Decision Support Systems, 33, 233-246.
Maui, HI.
Thatcher, J., & George, J. (2004). Commitment,
Zhou, L., Dai, L., & Zhang, D. (2007). Online
trust, and social involvement: An exploratory
shopping acceptance model—a critical survey of
study of antecedents to Web shopper loyalty.
consumer factors in online shopping. Journal of
Journal of Organizational Computing and Elec-
Electronic Commerce Research, 8(1), 41-62.
tronic Commerce, 14(4), 243-268.
van der Heijden, H., Verhagen, T., & Creemers,
M. (2003). Understanding online purchase inten-
Endnote
tions: Contributions from technology and trust
perspectives. European Journal of Information

1
We specially thank the special issue editor
Systems, 12, 41-48.
for granting extra space to include the ap-
Venkatesh, V., Morris, M., Davis, G., & Davis, pendices.
F. (2003). User acceptance of information tech-
nology: Toward a unified view. MIS Quarterly,
27(3), 425-478.

1223
Swift Trust in Web Vendors

Appendix I: Surveys and Measures

Construct Items
Appearance Features
The steps required at this Web site to search for products and make orders are typical of
other similar type Web sites.
Perceived Normality The information requested of me at this Web site is the type of information most similar
type Web sites request.
The nature of the interaction with this Web site is typical of other similar type Web sites.
There is a sense of human contact in this Web site.
There is a sense of personalness in this Web site.
Perceived Social
There is a sense of sociability in this Web site.
Presence
There is a sense of human warmth in this Web site.
There is a sense of human sensitivity in this Web site.
The display of the VeriSign icon on Overstock.com makes me feel that this Web site is
trustworthy.
The display of the BBBOnline icon on Overstock.com makes me feel that this Web site
Awareness of Third- is trustworthy.
Party Assurance The display of the Visa, MasterCard, American Express, and DiscoverCard icons on
Overstock.com makes me feel that this Web site is trustworthy.
The display of the BizRate.com icon on Overstock.com makes me feel that this Web site
is trustworthy.
Functionality Features
I find this Web site easy to use.

Effort Expectancy It would be easy for me to become skillful at using this Web site.
(Reverse Coded) Learning to use this Web site would be easy to me.
My interaction with this Web site is clear and understandable.
I find this Web site useful.

Performance Expec- Using this Web site can improve my shopping performance.
tancy Using this Web site can enhance my shopping effectiveness.
Using this Web site can increase my shopping productivity.
I feel secure in providing sensitive information (e.g., credit card number) when transact-
ing with Overstock.com.

Perceived Security I would feel totally safe providing sensitive information about myself to Overstock.com.
I would feel secure sending sensitive information to Overstock.com.
Overall, Overstock.com is a safe place to send sensitive information.

1224
Swift Trust in Web Vendors

Construct Items
I am concerned that Overstock.com is collecting too much information about me.
It bothers me when Overstock.com asks me for personal information.
I am concerned about my privacy when browsing Overstock.com.
Perceived Privacy
(Reverse Coded) I have doubts as to how well my privacy is protected on Overstock.com.
My personal information could be misused when transacting with Overstock.com.
My personal information could be accessed by unknown parties when transacting with
Overstock.com.
Swift Trust in Online Store
I believe this online store is effective in assisting and fulfilling my purchases.
This online store performs its role of e-vendor very well.
Competence
Overall, this online store is a capable and proficient e-vendor.
In general, this online store is very knowledgeable about the business it operates.
I believe that this online store would act in my best interest.
Benevolence If I required help, this online store would do its best to help me.
This online store is interested in my well-being, not just its own.
This online store is truthful in its dealings with me.
I would characterize this online store as honest.
Integrity
This online store would keep its commitments.
This online store is sincere and genuine.
Purchase Intention
Given the need, I intend to transact with Overstock.com.
Given the chance, I think that I would consider making purchases from Overstock.com.
I would probably purchase from Overstock.com when I have a need.
It is likely that I will actually buy products from Overstock.com in the near future.

Appendix II: Confirmatory Factor Analyses: Factor loadings &


Convergent Validity
(a) First-Order Constructs

Com Ben Int


Com1 0.855 0.568 0.581
Com2 0.948 0.664 0.650
Com3 0.929 0.679 0.678
Com4 0.915 0.666 0.636
Ben1 0.693 0.935 0.755
Ben2 0.664 0.914 0.706
Ben3 0.597 0.914 0.741
Int1 0.639 0.743 0.917
Int2 0.689 0.746 0.947
Int3 0.633 0.706 0.895
Int4 0.616 0.748 0.933

1225
Swift Trust in Web Vendors

(b) Second-Order Constructs

Social Aware- Effort Perf.


Intent Trust Normality Security Privacy
Presence ness Expect. Expect.
Intent1 0.936 0.489 0.336 0.309 0.370 0.298 0.475 0.382 0.080
Intent2 0.947 0.565 0.401 0.330 0.439 0.397 0.553 0.397 0.054
Intent3 0.958 0.498 0.344 0.353 0.395 0.350 0.556 0.395 0.084
Intent4 0.885 0.430 0.175 0.365 0.332 0.214 0.522 0.366 0.020
Com 0.479 0.886 0.561 0.267 0.567 0.532 0.480 0.494 0.206
Ben 0.499 0.925 0.449 0.405 0.562 0.445 0.505 0.566 0.214
Int 0.480 0.922 0.463 0.408 0.607 0.445 0.463 0.568 0.158
WN1 0.307 0.507 0.946 0.055 0.356 0.698 0.486 0.281 0.180
WN2 0.347 0.527 0.938 0.115 0.440 0.641 0.462 0.343 0.204
WN3 0.308 0.493 0.955 0.026 0.335 0.697 0.474 0.271 0.138
SP1 0.368 0.382 0.048 0.916 0.344 0.101 0.331 0.332 0.083
SP2 0.334 0.360 0.097 0.888 0.368 0.135 0.333 0.321 0.070
SP3 0.319 0.376 0.070 0.940 0.330 0.093 0.289 0.350 0.036
SP4 0.331 0.338 0.045 0.931 0.306 0.106 0.281 0.305 -0.044
SP5 0.326 0.368 0.064 0.937 0.311 0.115 0.324 0.354 0.041
AW1 0.348 0.612 0.410 0.289 0.893 0.397 0.354 0.500 0.214
AW2 0.307 0.544 0.289 0.274 0.886 0.317 0.364 0.444 0.164
AW3 0.374 0.537 0.393 0.322 0.848 0.328 0.311 0.445 0.119
AW4 0.426 0.524 0.301 0.384 0.875 0.261 0.342 0.519 0.138
PEOU1 0.334 0.520 0.688 0.128 0.368 0.950 0.583 0.273 0.186
PEOU2 0.302 0.474 0.706 0.080 0.361 0.921 0.531 0.233 0.123
PEOU3 0.306 0.444 0.619 0.103 0.280 0.923 0.512 0.217 0.136
PEOU4 0.333 0.505 0.672 0.133 0.391 0.953 0.554 0.300 0.098
PU1 0.482 0.480 0.478 0.271 0.340 0.589 0.836 0.330 0.159
PU2 0.548 0.496 0.463 0.299 0.386 0.512 0.965 0.315 0.041
PU3 0.538 0.500 0.450 0.337 0.373 0.541 0.958 0.317 0.026
PU4 0.526 0.485 0.466 0.343 0.350 0.519 0.936 0.332 0.102
Secu1 0.410 0.542 0.315 0.278 0.451 0.235 0.273 0.897 0.296
Secu2 0.364 0.511 0.251 0.378 0.493 0.217 0.338 0.939 0.300
Secu3 0.389 0.531 0.266 0.341 0.507 0.245 0.314 0.961 0.310
Secu4 0.375 0.620 0.335 0.347 0.567 0.312 0.367 0.924 0.263
Priv1 0.085 0.153 0.237 -0.057 0.174 0.230 0.098 0.218 0.829

1226
Swift Trust in Web Vendors

(b) continued
Priv2 0.054 0.178 0.190 -0.001 0.179 0.174 0.095 0.273 0.839
Priv3 0.051 0.174 0.160 0.041 0.170 0.132 0.058 0.265 0.878
Priv4 0.061 0.219 0.174 0.019 0.167 0.133 0.095 0.321 0.929
Priv5 0.025 0.187 0.112 0.096 0.154 0.051 0.040 0.280 0.877
Priv6 0.062 0.182 0.104 0.107 0.117 0.057 0.068 0.259 0.855

This work was previously published in Journal of Organizational and End User Computing, Vol. 21, Issue 1, edited by M.
Mahmood, pp. 88-108, copyright 2009 by IGI Publishing (an imprint of IGI Global).

1227
1228

Chapter 5.3
Understanding Brand
Website Positioning in the
New EU Member States:
The Case of the Czech Republic
Shintaro Okazaki
Universidad Autónoma de Madrid, Spain

Radoslav Škapa
Masaryk University Brno, Czech Republic

Abstract that ignoring the cultural, social, and economic


characteristics of particular markets would cause
This study examines Websites created by American psychological rejection by local consumers, thus
multinational corporations (MNCs) in the Czech decreasing profits in the long run. The debate has
Republic. Utilizing a content analysis technique, also produced a compromised or hybrid approach,
we scrutinized (1) the type of brand Website func- which suggests that whether to standardize or local-
tions, and (2) the similarity ratings between the ize advertising in a given market is a question of
home (US) sites and Czech sites. Implications are degree, and it is necessary to analyze many factors
discussed from the Website standardization versus on a case-by-case basis (Mueller, 1991).
localization perspective. This debate is not limited to traditional media.
As multinational corporations (MNCs) integrate
their marketing communication with an emergent
Introduction interactive medium, websites are becoming increas-
ingly important for brand marketing and customer
Both academics and practitioners have long de- relationship management in multiple markets. This
bated whether advertising messages should be is because the Internet is, by definition, a glocal
standardized. The proponents of standardization medium, which allows companies to create localized
argue that the use of uniform advertising would content with global access. In fact, many MNCs
provide significant cost benefits, thus improving have established so-called “global gateway” sites
company performance in the short run, while creat- with several language options. Consumers can first
ing a consistent brand image in multiple markets. choose the language, then seek the information
In contrast, the proponents of localization contend they desire. In this regard, the content of the local

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Understanding Brand Website Positioning in the New EU Member States

sites may need to be adapted to local consumers’ adding 10 countries: Cyprus, the Czech Republic,
tastes and preferences, in terms of design, lay- Estonia, Hungary, Latvia, Lithuania, Malta, Po-
out, copy, message, and so forth. (Okazaki and land, Slovakia, and Slovenia. In 2007, two more
Alonso, 2002). countries, Romania and Bulgaria became the mem-
Okazaki (2005) examined websites created by ber states, making the Union of 27 countries. This
American MNCs’ in four EU member states (i.e., drastic expansion changed the way multinational
the UK, France, Germany, and Spain), and found corporations (MNCs) operate their businesses in
a high level of localization in website communi- Europe. Because of these countries’ low labour
cation strategy. This research extends Okazaki’s costs and investment incentives (e.g., tax reduc-
exploration into the new EU member states, by tion, construction aid, etc.), many firms moved
conducting a content analysis of the MNCs’ web- their production facilities from other regions to
sites created in the Czech Republic. Specifically, these new member states. For example, Sheram
we address the following questions: (1) What types and Soubbotina (2000) report that “Countries seen
of brand website functions are used? (2) To what as more advanced in market reforms—the Czech
extent are the Czech sites standardized? and Slovak Republics, Hungary, and Poland—
attracted almost three-quarters of foreign invest-
ment” in transition economies. In fact, Poland
Significance of the Study received approximately $6.4 billion in foreign
direct investment in 2003, an increase of $360
This study will be an interesting addition to the million over the previous year (MacKay 2004).
literature of global information technology for As these new EU Member States experience
two reasons. First, prior research provides little rapid economic expansion, global marketing influ-
information on how the content created by the ences consumers in them more and more. Their
most globally diffused information technology, the product experiences increasingly resemble those
Internet, has been standardized in foreign markets. of their “Western” neighbours. In this light, it is
Information managers in global markets should reasonable to argue that the role of advertising
be aware of a question of transmitting culturally in everyday consumption has also undergone a
bound meanings into local sites. Secondly, this drastic transition, in both content and executions.
study addresses how design features and website For example, in the Czech Republic, advertising
functions can be used as a tool to create a uni- spending reached 563 million euros in 2004, while
versal imagery in global websites. Specifically, the average annual growth rate over the last 5 years
this study explores one of the most understudied has been 5%. Multinational corporations (MNCs)
countries in Europe: the Czech Republic. After are the largest advertisers in these countries.
joining the European Union, studies on informa-
tion technology in this new member state is almost
non-existent, thus, this research makes a unique Media Usage in Eastern
contribution to the literature. and Central Europe

The Czech Republic


Enlargement of the
European Union In the Czech Republic, television has tradition-
ally been the primary vehicle for advertising,
In 2004, the enlargement of the European Union accounting for 46% of the MNCs’ marketing
increased its member states from 15 to 25, by budgets. Print media is the second medium with

1229
Understanding Brand Website Positioning in the New EU Member States

34%, while outdoor advertising (i.e., billboard) most popular items in the Internet shopping are
is third, with 8% of total advertising expenditure electronics, books, journals, textbooks, tickets
(OMD Czech, 2005). However, the rapid growth and travel services, and accommodation. Online
of the Internet has significantly affected this media shopping is typically used more by men than
distribution. According to the Czech Publishers women, and by the younger generation groups,
Association, the share of Internet advertising between 25 and 45 years.
has been estimated at approximately 4%, or 25 Approximately 12,500 Czech enterprises pur-
million Euro (760 millions CZK), with a growth chased goods or services via the Internet in 2003,
rate of almost 80% in 2004 (Unie vydavatelů, almost 30% of the total number. The value of In-
2005). The telecommunications, financial, and ternet purchases reached 2.8% of total purchases,
automobile companies are the heaviest users and the value of Internet sales reached nearly 2.1%
of the Internet as an advertising medium (Unie of total sales in these enterprises.
vydavatelů, 2005).
In 2005, nearly 30% of Czech households had Poland and Hungary
a personal computer (Czech Statistical Office,
2005). Internet penetration is increasing steadily in Along with the Czech Republic, Poland and Hun-
the Czech Republic. Nowadays, 35% of the adult gary make up the fastest-growing economic region
population in the Czech Republic uses the Internet, within the new EU member states. For example,
almost twice the number of Internet users in 2000. the rapid transformation of the Polish economy is
The Czech Republic has thus clearly outmatched reflected in the accelerated growth of its advertising
other Central European countries: for example, market. Between 1996 and 1999, average annual
Bulgaria (16%), Hungary (22%), and Poland (31%). growth in advertising expenditure was more than
However, it has not yet achieved the levels of Internet 40%, which can be attributed to the drastic structural
penetration in Estonia (51%) or Slovenia (56%). changes, and the subsequent economic boom, in this
The most dynamic increase is found in older people period. With regard to media share, television was
(GfK, 2006), even though the Internet use remains the most popular (48%), with print media second
the domain of younger people. A quarter of Czech (35%). On-line advertising, including websites,
citizens have an Internet connection at home. remains far behind traditional media, representing
Searching for information is one of the most approximately 1% of total media spending (Zenith
frequent activities on the Internet, according to a Optimedia, 2004). The telecommunications, finan-
survey by the Czech Statistical Office (2005). In cial, and automobile industries are the heaviest
the most recent quarter to be surveyed, 62% of users of the Internet for advertising, promotion,
the Internet users used the Internet to find infor- and transactions (Agora, 2005).
mation about goods and services, 54% used it to In 2005, 30% of households in Poland had the
find and download professional texts, 38% looked technical possibility of the Internet access. In term
for services related to travel and accommodation, of quality of connections, the survey found that
36% to read and download on-line newspapers and only 16% of Polish households used a broadband
magazines, and 28% to play or download games or connection. The significant disparity in the Inter-
music. However, the number of individuals with net infrastructure is between urban areas and the
e-shopping experience increased rapidly between countryside, where the penetration of broadband
2005 and 2006 (the survey was carried out in the connections is four times lower than in urban
first quarters of 2005 and 2006): it amounted to areas (Eurostat, 2005).
14% in 2006, while a year before it had been The Internet usage by Polish enterprises is
only 6% (Czech Statistical Office, 2006). The below the EU average: in 2005, 87% of enterprises

1230
Understanding Brand Website Positioning in the New EU Member States

used the Internet connection. The share of broad- (Eurostat, 2006). The lowest rates of access were
band the Internet connections was 43%. More in Lithuania, for households (16%), and in Latvia,
than 67% of companies have their own website for enterprises (75%). The largest disparities in
homepage (Eurostat, 2006). Online purchases the Internet access between households and en-
have not yet become popular. Only 5% of Pol- terprises were recorded in the Czech Republic,
ish consumers ordered goods or services via the Lithuania, and Slovakia. The number of individu-
Internet in 2005 (Eurostat, 2006). According to als who have never used the Internet outweighs
the survey by GfK (2006), only 4% of the Internet the number of regular users in the new member
users make a purchase on the Internet once a month states. That differs from the situation in the old
or more. A further 6% buy online once every two member states.
to three months, while 18% go online sporadically There is also disparity in the presence of
with the intention of buying something. The most companies’ websites on the Internet. In January
frequently purchased items include books, CDs, 2005, 62% of enterprises in the EU were equipped
clothes, and shoes. Less frequently, people buy with a website, but only 49% in the new member
DVDs and air tickets (GfK, 2006). states. The lowest percentages of companies with
In B2B the situation is similar. In 2005, only websites were found in Latvia, Hungary, and
9% of the enterprises surveyed ordered products Lithuania. Most enterprises use the content of their
or services via the Internet. Sales via the Internet web presentations mainly to market their products.
were lower, with only 4% of enterprises selling via Less than half use it to display catalogues of their
the Internet. In 2005, turnover from e-commerce products, services, and prices. One quarter use
via the Internet amounted to 1.6% of total turnover websites to offer after-sales service to their custom-
(Eurostat, 2005). ers. Apart from the Czech Republic, enterprises
Similarly, the Hungarian market has shown in the new member states registered lower rates
a drastic growth in market size and advertising than the EU average for purchases, sales, and for
spending. According to the Budapest Business total sales on the Internet, as a percentage of their
Journal (BBJ, 2004), advertising expenditure overall turnover (Eurostat, 2006).
in television media reached 213 million euros The e-readiness rankings of the Economist In-
by 2003. The print media also showed a drastic telligence Unit can be seen as a complex indicator
growth, to spending of 212 million euros. In of the level of ICT of a country’s infrastructure.
2003, the online advertising market expanded by The index is a weighted collection of nearly 100
approximately 30%, achieving a 2% share of the quantitative and qualitative criteria, which assesses
total media market. The principal reasons for this the “state of play” of a country’s information and
growth were an increased number of the Internet communications technology (ICT) infrastructure,
users in younger generations, and the rapid pro- and the ability of its consumers, businesses, and
liferation of broadband high-speed connection. governments to use ICT to their benefit. In the
The largest online advertisers include car dealers, 2006 e-readiness rankings, Estonia achieved the
telecommunication companies, beer makers, and best position of all the new EU member states
cosmetics firms (BBJ, 2003). (27th), whereas Latvia (39th) was lowest. By
comparison, ten of the fifteen old EU members
Other EU Member States were in the top 20.
The Networked Readiness Index, published
In 2005, Slovenia had the highest rate of the annually in the Global Information Technology
Internet usage in the new member states, both Report, is a similar index. This index captures
for households (48%), and for enterprises (96%) such aspects as available ICT infrastructure, and

1231
Understanding Brand Website Positioning in the New EU Member States

Table 1. Descriptive statistics

Population1 GDP Advertising Advertising Internet Internet No of Online Internet


per spending3 spending penetration5 household local spending8 Advertising
capita2 as % of penetration6 domains7 spending9
GDP4
Czech
10,288.9 73.6 769,186 0.65 50 29 1,502,537 7 22,734
Rep.
Cyprus 776.0 88.9 89,073 0.54 33.6 37 75846 2 n.a.
Estonia 1,339.9 59.8 107,744 0.79 51.8 46 449,036 n.a. 3,607
Hungary 10,057.9 62.5 1,029,874 0.91 30.4 32 1,176,592 7 21,302
Latvia 3,385.7 48.6 129,961 0.81 45.2 42 132,204 1 7,277
Lithu-
2,280.5 52.1 150,07 0.50 35.9 35 240,592 2 3,086
ania
Malta 407.7 71.7 n.a. n.a. 33 53 20,673 n.a. n.a.
Poland 3,8101.8 49.7 1,862,672 0.55 29.9 36 5,001,786 6 32,885
Slovakia 5,391.6 57.1 n.a. n.a. 46.5 27 486,020 0 n.a.
Slovenia 2,010.3 81.9 242,656 0.64 55.5 54 64,284 9 5,484
EU 10 74,040.3 64.6 44 39 9,149,570
Note: 1. Data in thousands for the 1st of January 2007. Source: Eurostat (2007)
2. GDP (in PPS per capita) in 2005. EU25=100%. Source: Eurostat (2007).
3. Global advertising expenditure 2006. In $US Thousands. Initiative Innovation (2007).
4. Initiative Innovation (2007) and The World Factbook, Central Intelligence Agency (2007)
5. Internet Usage in the European Union. Penetration (% Population) in 2007. Source: Internet World Stats (2007).
6. Percentage of households who have Internet access at home in 2006. Source: Eurostat (2007).
7. Number of local domains based on number of top-level domain in January 2007. Source: ISC Internet Domain Survey (2007).
8. The Internet turnover as percentage of the total turnover of enterprises with 10 or more employees in 2006. Source: Eurostat (2007).
9. Global advertising expenditure 2006. In thousands of $. Initiative Innovation (2007).

actual levels of ICT usage, and its purpose is to illustrations, or copy, except for translation in
understand more thoroughly the impact of ICT on international markets (Onkvisit and Shaw, 1987).
the competitiveness of nations. In this index, Esto- The standardized school of thought argues that
nia again scored best amongst the new members. consumers anywhere in the world are likely to
Latvia and Poland had the lowest ratings. share the same wants and needs (Elinder, 1961;
Levitt, 1983). On the other hand, the localized
approach asserts that consumers differ across
Standardization vs. countries, and therefore advertising should be
Localization tailored according to culture, media availability,
product life cycle stages, and industry structure
The issue of standardization arises from the (Synodinos et al., 1989; Wind, 1986). Combining
desirability and feasibility of using a uniform these two extremes, the third school of thought
marketing mix (4Ps) across national markets states that the appropriateness of standardization
(Szymanski et al., 1993). Advertising has been depends on economic similarity, market position,
examined more often than the other elements of the nature of the product, the environment, and
this mix (Agrawal, 1995; Zandpour et al., 1994). organizational factors (Jain, 1989).
A standardized approach is the use of uniform In the 1970s, empirical evidence showed a
messages with no modification of headings, high degree of localization, due to both increasing

1232
Understanding Brand Website Positioning in the New EU Member States

nationalistic forces, and various well-publicized on-line marketing. This frequently entails creating
advertising blunders in the 1960s (Agrawal, 1995). a diverse range of websites in multiple markets
This trend reversed, to favour standardization, in (Donthu and Garcia, 1999). By 2001, more than
the 1980s, and went along with a drastic rise in 36 million domains for commercial websites had
the number of multinational advertising agen- already been established: these “dot coms” are
cies (Yin, 1999). During this period, a series of projected to attract an astonishing $6.8 trillion
content analysis studies attempted to identify in business by 2004 (Forrester, 2002; Internet
cross-cultural differences between Japanese and Software Consortium, 2001).
U.S. advertising (Hong et al., 1987; Madden et Such numbers incline observers to see the
al., 1986; Mueller, 1987). Internet as a door to the “global village won-
In the 1990s, localization seemed to remain derland”, as advocated by Levitt (1983): that is,
popular among MNCs operating in various regions an entity that creates an environment for more
of world markets. Harris (1994) found that 69% standardized marketing communication in world
of 38 MNCs (19 American and 19 European) markets. Product-based websites are replacing
standardized their advertising campaigns to such shopping venues as mail-order catalogues
some extent throughout the EC markets, whilst and television-based home shopping, and also
the rest of the sample localized. Interestingly, offer a new format for global advertising among
only 8% of the sample used totally standardized culturally and linguistically diverse groups (Pas-
advertising, providing “little evidence of any tor, 2001). An increase in the quantity and quality
widespread practice of standardized pan-European of product/brand information on the Internet is
advertising campaigns” (Harris, 1994). Kanso generating extraordinary consumer interest, which
and Nelson (2002) found that 62% of 193 firms extends beyond physical and political boundaries
(both American and non-American subsidiaries) (Donthu and Garcia, 1999). Accordingly, Roberts
in Finland and Sweden use localization, and place and Ko (2001) asserted that websites, with their
a strong emphasis on local cultures. Similarly, ability to uniformly blend textual and visual con-
Samiee et al. (2003) found that MNCs operating in tent, constitute the best communication medium
Southeast Asia tend to localize advertising. They in which to develop brand images.
examined 113 firms in Hong Kong, PRC, Taiwan, One roadblock that MNCs face involves local-
and Singapore, and found that both environmental ized websites: primarily, the need to satisfy the
and economic factors were the primary drivers of linguistic requirements of a diverse population
this tendency. of potential customers (Warden et al., 2002).
According to Quelch and Klein (1996), estab-
lishing localized relationships with international
Website Positioning as Global consumers is best achieved by creating regional
Information Management Web content. However, creating regional com-
mercial websites may not be cost-effective if, to
Although these issues have been debated for elicit return visits, a company is obliged to update
decades in traditional media, a new stream of re- information continuously. Such intense website
search has emerged recently, on the standardization maintenance on a regional level can jeopardize
versus localization of global websites in multiple consistent brand strategies, by eliminating the
markets. With the rapid expansion of the Internet, “advantage of centralized management of a firm’s
and the resulting connections between local, re- Websites” (Warden et al., 2002).
gional, and international markets, an increasing In a pioneering study, Okazaki and Alonso
number of MNCs are shifting from off-line to (2002) examined Japanese websites in Japan,

1233
Understanding Brand Website Positioning in the New EU Member States

Table 2. Network and e-readiness statistics

e-readiness Enterprises
Networked e-readiness e-readiness rankings, Enterprises availability
Networked Readiness rankings, rankings, Business selling via of the
Readiness Index e-readiness general Connectivity Environment Internet Internet
Index 1 (Rank) 1 rankings 2 index 2 index 2 index 2 2005 in % 3 2005 in % 3
Czech 0.36 32 32 6.14 4.90 7.39 13 92
Republic
Cyprus 0.36 33 n.a. n.a. n.a. n.a. 4 85
Estonia 0.96 23 27 6.71 6.60 7.81 8 90
Hungary 0.27 37 32 6.14 4.80 7.34 4 78
Latvia -0.03 52 39 5.30 3.95 7.21 1 75
Lithuania 0.08 44 38 5.45 4.65 7.28 6 86
Malta 0.51 30 n.a. n.a. n.a. n.a. 16 90
Poland -0.09 53 32 5.76 4.30 7.28 5 87
Slovakia 0.19 41 36 5.65 4.05 7.35 7 92
Slovenia 0.34 35 28 6.34 5.90 7.45 12 96
Note: n.a. = not available.
1. Global Information Technology Report 2005-2006
2. Economist Intelligence Unit (2006).
3. Eurostat (2005).

Spain, and the U.S.A., and found that cultural industrial goods.A summary of prior research on
dimensions (power distance, uncertainty avoid- website content analysis is shown in Table 3.
ance, masculinity – femininity, individualism-
collectivism, and long-term orientation) and
communication style (high context versus low Communication in the Global
context) were the primary drivers of cross-cultural Website Environment
differences in MNCs’ website communication
strategies. Focusing on more operational aspects, What is the primary factor influencing MNCs
Okazaki (2005) examined American brands’ web- that operate in European markets? They now
site standardization in France, Germany, Spain, face more and more pressure to generate more
and the UK. He argued that the progress of the comprehensive marketing strategies on the web.
EU enlargement and economic integration via Among the various forms of the online environ-
the euro provided firms with an incentive to use ment, websites have been one of the most popular
a uniform website communication across the EU platforms, allowing consumers to see, consult, and
member states. However, the findings were mixed, obtain product-related information at any time,
in that the level of standardization of American anywhere. Such websites can be seen as a new
brands’ websites in the European countries was form of global marketing communications, offer-
low, compared to their respective home-country ing opportunities to strengthen effective relational
(American) websites. On the other hand, differ- marketing in international markets (Robert and
ences within the EU were minimal: the websites Ko, 2002). The creation of a localized URL in
created within the European markets were some- Europe, therefore, may be a necessary strategic
what “regionalized”, especially for durable and move, because cultural and linguistic barriers are

1234
Understanding Brand Website Positioning in the New EU Member States

Table 3. Prior research on website content analysis

Year Authors Countries Unit of analysis Analyzed content Sample size Statistical design
examined
1999 Ju-Pak US, UK & Product-based web- Information content, 11 0 ( E E . U U . ) Chi-square, ANO-
S.Korea sites creative strategies, design 100 (UK) VA
100(S.Korea)
1999 Oh, Cho & U S & Target ads Information content, creative 50 for each coun- Chi-square
Leckenby S.Korea strategies try
1999 Yo o n & U S & Brand websites Information content, emo- 20 for each coun- Chi-square
Cropp S.Korea tional appeals, cultural as- try
pects
2000 Lee & Sego U S & Banners Information content, emo- 252 in total Chi-square
S.Korea tional appeals, colours, etc.
2000 Chung & Ahn U S & Banners Information content, 251 (EE.UU.) Chi-square
S.Korea “call-to-action” messages, 221 (S.Korea)
demographics, etc.
2000 Yoon US Product-based web- Information content, 200 in total Chi-square, ANO-
sites celebrity endorsement, etc. VA
2002 Okazaki & Japan, Spain & Product-based web- Information content, cultural 20 for each coun- Chi-square, ANO-
Alonso EE.UU. sites values, creative strategies try VA
2002 Dou, Nielsen Canada, Den- Commercial web- Communication systems, 150 for each coun- ANOVA
& Tan mark & Ma- sites transactional functions, etc. try
laysia
2002 Zahir, Dobing 26 countries National portals Linguistic aspects, design, 26 portals Descriptive stat
& Hunter colours, Hofstede’s cultural
dimensions, etc.
2003 Robbins & 16 countries Corporate websites Design, presentation, links, 90 in total ANOVA
Stylianou security, information con-
tent, financial content, cor-
porate information, etc.
2005 Okazaki US, UK, Brand websites Brand website functions, 244 ANOVA, discrimi-
France, Ger- similarity ratings nant analysis, mul-
many & tiple regression
Spain

perhaps the most difficult obstacles to overcome ers. Unfortunately, there seems to be a lack of
in marketing communications across European empirical research regarding the standardization
nations (Kahle, Beatty, and Mager, 1994). Such versus localization issue in the online environment,
localization, however, could cost a great deal. leaving important questions unanswered.
Hence, MNCs may intuitively favour standard- What are the determining factors in interna-
ization, given the benefits associated with offline tional marketing communications on the web? In
marketing standardization, such as consistent a recent criticism of the slow progress of interna-
brand image and corporate identity, cost savings, tional advertising research, Taylor and Johnson
and organizational control. Furthermore, websites (2001) argue that the standardization debate should
seem to be an effective medium for establishing a “focus on what executions can be standardized and
global brand image, by offering consistent textual when they can be standardized”. Following this
and visual information to international consum- suggestion, this study intends to fill this gap, by

1235
Understanding Brand Website Positioning in the New EU Member States

identifying to what extent MNCs have adopted a communication and transaction feature form two
standardized approach for their websites created ends of one axis, which should be balanced with
in European markets. In order to ensure cross- the other axis, consisting of fact and image. The
national data equivalency, we examined only resulting four quadrants need to be effectively
the websites created by America’s top brands combined to achieve the desired level of website
for the UK, France, Germany, and Spain. These standardization. The components in each quadrant
countries differ importantly in terms of cultural can be considered the most relevant programmes
and linguistic characteristics, but are relatively for website brand communications.
homogeneous in socioeconomic conditions and The extent of website standardization should
technological infrastructure, and have online be determined on the basis of the two major roles
markets of a reasonable size. of global online programmes: (1) to enhance
Furthermore, 3.3%, 6.5%, and 8.1% of the worldwide transactions by establishing a localized
world’s online population consist of French, relationship, and (2) to develop a standardized
German, and Spanish speaking consumers, brand image, using the appropriate combination
respectively, compared to 35.2% of English of content, graphics, backgrounds, and multimedia
speakers (Global Reach, 2003). Therefore, these effects in all the MNC’s websites in different lan-
four countries represent an important segment of guages (Roberts and Ko, 2003). In the following
world online consumers. On the other hand, the section, each principal feature of our proposed
languages spoken in the new EU member states, model is therefore analyzed in the light of these
such as Polish or Czech, account for a very small perspectives.
portion of the online population. In fact, the impact
of these countries, on both the world economy
and the world online population, is negligible Methodology
(Table 4). Thus, an important question arises: is
it worthwhile for MNCs to consider local adapta- This study adopts content analysis as a research
tion in such new markets? Or is it better to use a methodology. This method has been widely used in
standardization approach in these markets, to take cross-cultural research (Brislin, 1980), as well as
advantage of cost savings and efficient website in the Internet research (McMillan, 2000; Okazaki
maintenance? To address these questions, this and Alonso, 2002).
study will examine websites created by MNCs Company selection. A website content analysis
for the Czech Republic. was performed, to examine the degree of website
standardization of American brands’ websites
created for the Czech market. Methodological rec-
Conceptual Framework ommendations from prior research were adopted,
to establish a high reliability (Dou et al., 2002;
Figure 1 shows the conceptual framework for this Okazaki and Alonso, 2002; Philport and Arbittier,
study. These concepts are essentially based on the 1997). To create a dataset, a ranking of “The 100
matrix proposed by Quelch and Klein (1996), who Top Global Brands” from BusinessWeek (2002)
suggested two primary models of website: the was used. Only brands with America as country
communication model and the transaction model. of origin (by the classification of BusinessWeek)
Originally, their matrix was not intended to be a were chosen to match home versus host country
theoretical model for formal testing, but since then website pairs. In total, 66 brands were found,
it has been used as a conceptual base (e.g., see Dou, of which 34 brands had websites in the Czech
Nielsen, and Tan, 2002). In our modified matrix, Republic. Here, it is important to note that these

1236
Understanding Brand Website Positioning in the New EU Member States

Table 4. World online population and language use

Language type Internet % of Total GDP2 % of world GDP per


access 1 world pop.1 economy capita 3
online
pop.
English 238.5 35.2 508 n.a. n.a. n.a.
Non-English 439.8 64.8 5,822 n.a. n.a. n.a.
European Languages (non-English) 238.1 35.1 1,218 12,968 30.3 n.a.
Czech 4.0 n.a. 12 121 n.a. 10.0
Dutch 13.2 2.0 20 575 n.a. 28.5
Finnish 2.8 n.a. 6 142  n.a. 23.6
French 22.7 3.3 77 1517 4.2 19.7
German 44.4 6.5 100 2,679 5.8 26.8
Greek 2.0 n.a. 12 189 n.a. 15.8
Hungarian 1.6 n.a. 10 96 n.a. 9.6
Italian 24.1 3.6 62 1,251 3.6 20.1
Polish 6.9 n.a. 44 359 n.a. 8.1
Portuguese 19.3 2.8 176 1,487 3.6 8.4
Romanian 2.4 n.a. 26 108 n.a. 4.2
Russian 18.4 2.7 167 822 1.8 4.9
Scandinavian languages (total) 13.5 2.0 20 550 1.3 27.9
Serbo-Croatian 1.0 n.a. 20 n.a. n.a. n.a.
Slovak 1.0 n.a. 6 47 n.a. 8.7
Slovenian 0.7 n.a. 2 22.9 n.a. 10.9
Spanish 54.8 8.1 350 2500 8.9 7.1
Turkish 4.6 n.a.  67 431 n.a. 6.4
Ukrainian 0.9 n.a. 47 115 n.a. 2.3
TOTAL EUROPEAN LANGUAGES (non-English) 238.1 35.1 1,218 12,968 33.9 n.a.
TOTAL ASIAN LANGUAGES 201.7 29.7 n.a. n.a. n.a. n.a.
TOTAL WORLD 648.7 6,330 41,400 n.a. n.a.
Note: 1 US$ in million; 2 US$ in billion, 3 US$ in Thousand.
Source: Global Reach (2003)

firms are considered as representative of Ameri- who suggested 23 website communication func-
can firms doing business in the Czech Republic, tions. Similar categories have been suggested
because their Internet presence can be considered in the past (e.g., Ghose and Dou, 1998; Leogn,
as the initial step of market entry mode. Huang, and Stanners, 1998). Each variable was
Coding categories. Next, a detailed coding measured on a categorical dichotomy as to the
sheet was first developed, with detailed opera- existence of each function on a website. Values of
tional definitions. The variables include (1) brand “1” and “0” were assigned for answers of “Yes”
website functions, and (2) similarity ratings and “No”, respectively. For example, websites that
(Table 5). With regard to the former, the coding had appropriate information associated with “job/
categories were adopted from Okazaki (2005), career development” were assigned “1” for this

1237
Understanding Brand Website Positioning in the New EU Member States

Figure 1. Conceptualization of website program standardization

attribute. Those that did not were assigned “0”. age, which has been considered a central gate to
These values were considered dependent variables Web-based communication. This is appropriate,
in the analysis. The similarity rating refers to the given the primary objective of the study: to identify
degree of similarity between home-country and major differences in the main text, pictures, and
host-country websites. This “similarity rating” graphics. We examined the first and second levels
measure was also adopted from Okazaki (2005), of websites, because it is practically impossible
and was originally inspired by Mueller (1991). The to scrutinize every detail of an entire site. The
textual and visual components of websites created existence of online brand communications was
for local markets were assessed for the extent to primarily determined by the main menu or index
which they were similar or dissimilar to those provided on the homepage. For instance, if the
created at home. The similarity rating was coded menu included a link labelled “corporate informa-
for each pair of websites (i.e., U.S.A.-Czech sites) tion”, the site was coded as having this variable.
on a five-point semantic scale, ranging from “very The only exception occurred when analyzing
different” (coded as 1) to “very similar” (coded direct or indirect online transactions, because in
as 5), with an intermediate scale point “not deter- some cases these functions may not be listed on
minable” (coded as 3). The components included the main index. In this case, the coders were asked
copy, headlines, text, layout, colour, photographs to examine the submenu of the websites.
associated with the product, with human models, Coder training and reliability. Following the
and with background scenes, illustrations, charts, recommendations by Kolbe and Burnett (1991),
graphs, and interactive images. two bilingual Czech judges, both of whom were
Coding instrument. All coding instruments unaware of the purpose of the study, were hired
were originally prepared in English, and then and first trained to grasp the operational definitions
translated into Czech, using the “back translation” of all the variables. During the training period, the
technique. Each typology was supplemented with coders practised independently, by examining 20
additional examples, to give a better illustration. randomly chosen websites from non-American
The unit of analysis was determined as the homep- firms. Then, the coded results were compared,

1238
Understanding Brand Website Positioning in the New EU Member States

Table 5. Measurement schemes

Measure Coding categories Scale type


Brand website Global/Local site options, Corporate information, Corporate news release, Product/Brand Nominal scale
functions news release, General product information, Brand specific information, Investor relation- (Yes=1, No=0)
ships, Direct online transaction, Indirect online transaction, Office/Store locator, Country/
Language option, Search engine, Jobs/Career development, Promotion/Prizes/Sweepstakes,
Education/Training, Culture/Entertainment, Client registration/Log-in, Guest book/Customer
feedback, E-mail alert, FAQs, Free download, Sitemap, Links

Similarity ratings Company logo, Company logo placement, Major copy, Major copy placement, Major Interval scale
headline, Major headline placement, Major text, Layout in top half / right half, Layout (1=very different, 5=very
in bottom half / left half, Colour in top half / right half, Colour in bottom half / left half, similar)
Major photograph 1 (product), Major photograph 2 (human model), Major photograph 3
(background scene), Major illustrations, Major chart or graph, Interactive image 1 (flash as
opening), Interactive image 2 (pop-ups), Interactive image 3 (animated banners), Interactive
image 4 (layers, pop-unders, etc.)

and differences were reconciled through discus- (2005) previous exploration was used as a refer-
sion. An inter-judge reliability was calculated ence with regard to the U.S., UK, French, and
using the reliability index suggested by Perreault German markets. This comparison should help our
and Leigh’s reliability index (Ir) (1989). Various understanding of MNCs’ website standardization
researchers consider this estimation method to be in existing versus new EU member states.
the best among available alternatives (e.g., Kolbe The Chi-square analysis detected significant
and Burnett, 1991). differences in 8 categories: global/local site op-
As Table 6 shows, the majority of the resulting tions, general product information, investor rela-
indexes far exceeded a widely accepted minimum tions, online purchase, email contact, promotion/
.80???, and was thus deemed satisfactory. It was prizes/sweepstakes, culture/entertainment, and
recognized that there would be a potential loss guest book/customer feedback. It appears that
of information in similarity evaluation between American MNCs tend to apply a different web-
American and Czech sites, because non-native site communication strategy in the Czech market
English speakers had analyzed American web- because, in prior research, Okazaki (2005) found
sites. However, it was accepted that such poten- significant differences in only 3 of 23 variables,
tial bias was minimized by the coders’ extensive suggesting that the frequency of the usage of brand
preparation: the subjective interpretation of tex- website functions in the UK, France, Germany,
tual information was minimal, since the coders and Spain was relatively uniform. In observing
were responsible for examining only major copy, the frequencies of brand website functions in
headlines, and text on the websites. Otherwise, the Czech sites, email contact was used more
they were instructed to objectively measure the frequently, but the other tools were used much
similarity of non-textual information. less than in the other markets.
Next, in order to capture the relationships
between the brand website functions and country
Results domain, we performed a multiple correspondence
analysis via optimal scaling technique. This
Table 7 summarizes the frequency distribution method is appropriate for nominal variables, from
of brand website functions. For the sake of com- which a multidimensional map can be created. We
parison, the information provided in Okazaki’s used the existence of the brand website functions

1239
Understanding Brand Website Positioning in the New EU Member States

Table 6. Intercoder reliability

Measure Coding categories Perreault & Leigh’s Ir


Brand website functions Brand specific information 0.82
Client registration/Log-in 0.91
Corporate information 1.00
Corporate news release 0.97
Country/Language option 1.00
Culture/Entertainment 0.97
Direct online transaction 1.00
Education/Training 0.94
E-mail alert 1.00
FAQs 1.00
Free download 0.94
General product information 0.97
Global/Local site options 0.94
Guest book/Customer feedback 1.00
Indirect online transaction 1.00
Investor relations 0.97
Jobs/Career development 0.97
Links 0.91
Office/Store locator 1.00
Product/Brand news release 1.00
Promotion/Prizes/Sweepstakes 0.97
Search engine 1.00
Sitemap 1.00

Similarity ratings Text 0.91


Major photograph: product 0.75
Major photograph: model 0.56
Major photograph: background scene 0.92
Major illustrations 0.82
Major chart or graph
Layout in top half / right half 0.93
Layout in bottom half / left half 0.98
Headline placement 0.74
Headline 0.69
Copy placement 0.73
Copy 0.91
Company logo placement 0.95
Company logo 0.96
Colour in top half / right half 0.93
Colour in bottom half / left half 0.91

1240
Understanding Brand Website Positioning in the New EU Member States

Table 7. Results of brand website functions

1. Brand website features US UK France Germany Czech p


(n=66) (n=57) (n=49) (n=57) (n=34)
Global/Local site options 37.9 84.2 65.3 75.4 64.4 .000
Corporate information 89.4 86.0 87.8 84.2 86.1 .956
Corporate news release 53.0 54.4 55.1 59.6 55.6 .983
Product/Brand news release 51.5 49.1 53.1 52.6 55.6 .990
General product information 80.3 84.2 83.7 78.9 47.2 .000
Brand specific information 75.8 68.4 73.5 68.4 69.4 .858
Investor relationships 45.5 26.3 16.3 22.8 11.1 .001
Online purchase 71.2 42.1 42.6 43.9 25.0 .000
Email contact 22.7 31.6 28.6 24.6 77.8 .000
Office/Store locator 33.3 33.3 32.7 26.3 13.9 .346
Country/Language option 62.1 57.9 71.4 61.4 52.8 .536
Search engine 68.2 57.9 55.1 50.9 52.8 .442
Jobs/Career development 62.1 47.4 46.9 54.4 61.1 .374
Promotion/Prizes/Sweepstakes 56.1 63.2 44.9 47.4 11.1 .000
Education/Training 39.4 33.3 26.5 24.6 25.0 .331
Culture/Entertainment 47.0 57.9 53.1 42.1 13.9 .001
Client registration/Log-in 51.5 36.8 38.8 36.8 41.7 .398
Guest book/Customer feedback 78.8 82.5 75.5 77.2 22.2 .000
E-mail alert 25.8 15.8 20.4 19.3 25.0 .483
FAQs 18.2 22.8 16.3 19.3 8.3 .647
Free download 19.7 26.3 26.5 28.1 36.1 .643
Sitemap 45.5 42.1 44.9 36.8 38.9 .905
Links 4.5 12.3 8.2 3.5 19.4 .053
Note: The data of the US, UK, France and Germany are based on Okazaki (2005)

(yes or no) as descriptive variables, and the type upper right quadrant (where Czech sites appear
of country domain (U.S.A. or Czech Republic) as to dominate).
classification variables. Figure 2 shows the result- Lastly, Table 8 summarizes the results of
ing biplot component loadings. As clearly seen, the similarity ratings. A higher similarity rating
the majority of brand website functions are more indicates a higher degree of standardization. As
closely associated with U.S. sites (represented the results clearly show, the similarity between
by “1”), while only a limited number of brand the American and Czech sites was notably higher,
website functions are associated with Czech sites especially in logo, copy, and colour. In comparison
(represented by “2”). Specifically, global/local site with the other sites, headlines and major pho-
options, links, and indirect online transactions are tographs also exhibit higher similarity. On this
concentrated in the lower left quadrant (which basis, it appears clear that the American MNCs
U.S. sites appear to dominate), but the rest of the tend to create highly standardized websites in the
brand website functions are concentrated in the Czech Republic.

1241
Understanding Brand Website Positioning in the New EU Member States

Figure 2. Multiple correspondence analysis

Discussion growing market that attracts more and more foreign


direct investment, this case could be considered
This study attempts to examine the website com- indicative of the general tendency in the other
munication strategy used by American MNCs in new EU member states.
a new EU member state, the Czech Republic. First, American MNCs tend to use general
We performed a content analysis of 34 Czech product information less frequently in the Czech
sites created by America’s top brands in terms of market than in the other EU markets, which sug-
brand website functions, and similarity between gests two possible scenarios. First, they have not
the home and host sites. The findings indicate yet commercialized their products in this market,
that American MNCs appear to standardize their and therefore dispose of little information. In this
Czech sites. Given that the Czech Republic is a case, the primary objective of their websites would

1242
Understanding Brand Website Positioning in the New EU Member States

Table 8. Results of similarity ratings

Components UK France Germany Czech


(n=57) (n=49) (n=57) (n=34)
Company logo 4.51 4.61 4.39 4.58
Major copy 1.43 1.36 1.14 4.00
Major headline 1.23 1.17 1.02 3.47
Major text 2.84 2.53 2.68 2.32
Layout 3.35 3.32 3.29 3.88

be to provide a preliminary information platform customer feedback indicates that American brands
in a new market. Second, they might have needed are less willing to offer personalized contact to the
to localize product information to a great extent, local Czech consumers, probably because of the
especially in the local language, with more adapted unavailability of local outlets, representatives, or
product usage. However, it appears that the most staff. In contrast, in the Czech Republic, they offer
logical conclusion would be that the lack of online general email addresses more frequently than in
product information is because American firms the other countries, as an alternative contact mode
are still in the very early mode of market entry in for general inquiries.
the Czech Republic. Third, by the same token, culture/entertain-
The second scenario seems very unlikely ment and promotion/prices/sweepstakes are used
when we address the following question: Why much less in the Czech market than in the others,
would large multinational firms devote resources because these elements need to be matched to local
to extensive adaptations of Czech websites? The consumer tastes, and require more personalized
country’s population is only 10 million and, ac- content. It would make little sense to offer presents
cording to the most recent Eurostat (2006), only or incentives when the companies actually have
19% of Czech homes had the Internet access in no local sales activities.
2005, including 5% with broadband connections. Finally, a surprisingly high level of similarity
As much as 63% of the population has never used ratings for both textual and visual components in-
the Internet. Only one in six people who used the dicates a lack of any cultural adaptation of websites
Internet (5.5%) bought anything online in 2005, to the Czech market. This contrasts with Okazaki’s
and these purchases were limited largely to elec- (2005) findings regarding American MNCs’ web-
tronic goods (2.1%), books (1.6%), and clothing site strategy in the UK, France, Germany, and
(1.1%). Therefore, if the total market in a given Spain, where clearly localized websites have been
product category is currently only 100,000 or so, created in the existing EU member states. This
and if adapting the website is only going to improve finding is consistent with the frequency of brand
website effectiveness by 5% to 10%, is there any website functions: the Czech sites use far fewer
incentive to adapt and then to continue managing brand website functions with highly standardized
that adaptation? Consistent with this argument, our textual and visual components.
findings indicate that online purchase functions are It is clear that American MNCs consider the
rarely used in the Czech sites. This suggests that EU a single market, and one that is strategically
American MNCs may have neither distribution dissimilar to their home market. If we observe
channels nor local investors in the Czech market. only the websites created in the “older” member
Similarly, the much less frequent use of guest book/ states, there seems to exist a “regionalization”

1243
Understanding Brand Website Positioning in the New EU Member States

strategy across Europe, in that the level of simi- was limited to the menu and submenu of the
larity ratings among the European samples was homepages. However, it is possible that a more
relatively uniform. This may be due to the close localized strategy might have been observed in
geographical proximity of the three countries, further links. Finally, we examined only one of the
which would, logically, provide more opportuni- new EU member states, thus, any generalization
ties for personal interaction and the accumulation of our findings should be treated with caution.
of greater knowledge. However, in the case of the
Czech Republic, website adaptation has not yet
advanced, probably due to many unknown factors: Future Research Directions
in particular, specific information regarding local
consumers’ tastes and preferences. First, future extensions should examine websites
To make our findings more objective, we created for the other new EU member states,
should recognize a few limitations. First, the such as Poland, Hungary, Latvia, Lithuania, and
current study examined only one country that Malta. Information technology management for
has recently joined the European Union. Future website positioning in these countries is virtually
research should expand this study into other unknown, and analyzing brands’ websites posi-
new EU member states, especially Poland and tioning in these countries should help us to draw
Hungary, because these two countries, along more generalisable implications. In particular,
with the Czech Republic, are the most economi- researchers are planning to examine Polish and
cally developed regions. Because of the extreme Hungarian websites in the next stage because, in
scarcity of research related to these countries, these countries, the total online as well as offline
any such extension will contribute significantly advertising spending is substantial, in comparison
to the literature. Second, while content analysis with the other new EU member sates.
could provide useful information regarding the Second, in furthering our explorations, con-
manifest content, the findings should be treated tent analysis methodology should be improved.
with caution. The findings by no means explain Specifically, we need to examine the level of
practitioners’ “true” intentions in website com- standardization or localization at deeper levels
munication strategy. In this regard, it will be of websites. While this study scrutinized the
interesting to extend this study in the future, by first level of websites or homepages, some may
conducting a questionnaire survey of advertisers claim that much information was lost by ignor-
and marketers who are actually responsible for ing the second and third levels of websites. For
the new EU member states. example, the lack of direct online transactions
need not necessarily mean that the website does
not have a link to the online shopping sites of dif-
Limitations ferent companies. This was the case for consumer
electronics, in that computer or office machine
While this study makes significant contributions products are sold on “general” e-commerce (or
to the global information management literature, even auction) sites. More specific coding instruc-
some important limitations must also be recog- tions should be established, to enable the coders
nized. First, content analysis is, by definition, an to improve their analysis with a higher level of
observational method that examines only mani- inter-coder reliability.
fest content. Our findings have little or nothing Third, we also should conduct a survey that
to do with marketers’ “true” intentions on global targets foreign subsidiaries’ managers. It will be
website positioning. Second, our unit of analysis interesting to compare the findings of this paper

1244
Understanding Brand Website Positioning in the New EU Member States

with the managers’ perceptions. In particular, Czech, O. M. D. (2005). Odhady reklamních


several questions appear of special interest. For výdajů.
example, are their websites created or controlled
Czech Publishers Association. (2006). Internetová
locally or globally? What level of control do
reklama v roce 2005 přesáhla 1 miliardu Kč.
senior executives of foreign subsidiaries actually
Press release,
have of their electronic commerce planning and
executions? Czech Statistical Office (2006), The survey on
Finally, in an attempt to capture a clearer ICT usage in households and by individuals in
picture of global website positioning in multiple the Czech Republic in 2006 (in Czech).
markets, more collaboration will be needed by
Donthu, N., & Garcia, A. (1999). The Internet
researchers in information systems management
Shopper. Journal of Advertising Research, 39(5),
and other disciplines: in particular, marketing
52–58.
management. Needless to say, a higher level of
international cooperation is necessary to conduct Dou, W., Nielsen, U. O., & Tan, C. M. (2002).
more objective and reliable data collection in Using corporate websites for exporting market-
multiple markets. ing. Journal of Advertising Research, 42(5),
105–115.
Economist Intelligence Unit (2006), The 2006
References
e-readiness rankings.
Agora (2005). Advertising market. Elinder, E. (1961). How International Can Ad-
Agrawal, M. (1995). Review of a 40-year vertising Be?” in S. W. Dunn (ed.), International
debate in international advertising. Inter- Handbook of Advertising, NY: McGraw-Hill,
national Marketing Review, 12(1), 26–48. 59-71.
doi:10.1108/02651339510080089 Eurostat (2005), Europe in figures: Eurostat
ARBOmedia. (2005a). Standardní reklamní rok yearbook 2005.
přinese médiím 17,8 miliard. Eurostat (2006), Community survey on ICT usage
ARBOmedia. (2005b). Standardní reklamní rok in households and by individuals.
přinese médiím 17,8 miliard“. Factum, T. N. S. (2004). Firemní weby a market-
Brislin, R. W. (1980). Translation and Content ing.
Analysis of Oral and Written Material. In H. C. Factum, T. N. S. (2005). Češi a reklama v roce
Triandis and Berry, J. W. (eds.), Handbook of 2004.
Cross-cultural Psychology (Vol. 2, pp. 389-444),
Boston: Allyn and Bacon. Forrester (2002), Forrester’s Global eCommerce
Predictions For 2004.
Budapest business journal (2003). Web-based ad-
vertising making ground. September 16, 2003, GfK. (2006). Increasingly more kids use the In-
ternet. One in three Polish Internet users willing
Budapest business journal (2004). Online adver- to buy music on the Internet. Press release,
tising boom. August 18, 2004,
BusinessWeek (2002). The 100 Top Brands. Au-
gust 5, 95-99.

1245
Understanding Brand Website Positioning in the New EU Member States

Ghose, S., & Dou, W. (1998). Interactive Func- Leong, E. K. F., Huang, X., & Stanners, P. J.
tions and Their Impacts on the Appeal of Internet (1998). Comparing the Effectiveness of the Web
Presence Sites. Journal of Advertising Research, site with Traditional Media. Journal of Advertis-
38(2), 29–43. ing Research, 38(5), 44–51.
Harris, G. (1994). International Advertising Stan- Levitt, T. (1983). The Globalization of Mar-
dardization: What Do the Multinationals Actually ket. Harvard Business Review, 61(May/June),
Standardize? Journal of International Marketing, 92–102.
2(4), 13–30.
Mackay, S. (2004). Parliament Approves New
Hong, J. W., Muderrisoglu, A., & Zinkhan, G. Government. Executive Perspectives: Poland,
M. (1987). Cultural differences and advertising Price Waterhouse Coopers.
expression: a comparative content analysis of
Madden, C. S., Caballero, M. J., & Matsukubo,
Japanese and U.S. magazine advertising. Journal
S. (1986). Analysis of information content in U.S.
of Advertising, 16(1), 55–62, 68.
and Japanese magazine advertising. Journal of
Internet Software Consortium. (2001), Distribu- Advertising, 15(3), 38–45.
tion of Top-Level Domain Names by Host Count,
McMillan, S. J. (2000). The Microscope and
Jul 2002.
The Moving Target: The Challenge of Apply-
Jain, S. C. (1989). Standardization of International ing Content Analysis to The World Wide Web.
Marketing Strategy: Some Research Hypoth- Journalism & Mass Communication Quarterly,
eses. Journal of Marketing, 53(January), 70–79. 77(1), 80–98.
doi:10.2307/1251525
Mueller, B. (1991). An analysis of information
Ju-Pak, K. H. (1999). Content dimensions of Web content in standardized vs. specialized multina-
advertising: a cross-national comparison. Interna- tional advertisements. Journal of International
tional Journal of Advertising, 18(2), 207–231. Business Studies, First Quarter, 23-39.
Kanso, A., & Nelson, R. A. (2002). Advertising OBP. (2001). Advertising in Poland.
Localization Overshadows Standardization. Jour-
Oh, K. W., Cho, C. H., & Leckenby, J. D. (1999).
nal of Advertising Research, 42(1), 79–89.
A Comparative Analysis of Korean and U.S. Web
Kolbe, R. H., & Burnett, M. S. (1991). Content- Advertising. Proceedings of the 1999 Conference
Analysis Research: An Examination of Applica- of the American Academy of Advertising, Gaines-
tions with Directives for Improving Research ville: University of Florida, 73-77.
Reliability and Objectivity . The Journal of
Okazaki, S. (2005). Searching the Web for
Consumer Research, 18(September), 243–250.
global brands: How American brands stan-
doi:10.1086/209256
dardise their websites in Europe. Europe-
Koudelova, R., & Whitelock, J. (2001). A an Journal of Marketing, 39(1/2), 87–109.
cross-cultural analysis of television advertis- doi:10.1108/03090560510572034
ing in the UK and the Czech Republic. Inter-
Okazaki, S., & Alonso, J. (2002). A content analysis
national Marketing Review, 18(3), 286–300.
of Web communication strategies: Cross-cultural
doi:10.1108/02651330110695611
research framework and pre-testing. Internet
Research: Electronic Networking . Applications
and Policy, 12(5), 380–390.

1246
Understanding Brand Website Positioning in the New EU Member States

Onkvisit, S., & Shaw, J. J. (1987). Standardized Szymanski, D. M., Bharadwaj, S. G., & Vara-
International Advertising: A Review and Critical darajan, P. R. (1993). Standardization versus
Evaluation of the Theoretical and Empirical Evi- Adaptation of International Marketing Strategy:
dence. The Columbia Journal of World Business, An Empirical Investigation. Journal of Marketing,
22(Fall), 43–55. 57(October), 1–17.
Optimedia, Z. (2004). Poland’s advertising market Taylor, C. R., & Johnson, C. M. (2002). Stan-
in 2004. dardized vs. specialized international advertising
campaigns: What we have learned from aca-
Perreault, W. D., & Leigh, L. E. (1989). Reliability
demic research in the 1990s. New Directions in
of nominal data based on qualitative judgments.
International Advertising Research, 12, 45–66.
JMR, Journal of Marketing Research, 26(May),
doi:10.1016/S1474-7979(02)12019-9
135–148. doi:10.2307/3172601
Unie vydavatelů (2005). Internetová reklama v
Philport, J. C., & Arbittier, J. (1997). Advertis-
roce 2004 dosáhla 760 milionů Kč.
ing: Brand communications styles in established
media and the Internet. Journal of Advertising van REPEN, Erica, Rik Pieters, Jana Fidrmucova
Research, 37(2), 68–76. and Peter Roosenboom. (2000). The Information
Content of Magazine Advertising in Market and
Quelch, J. A., & Klein, L. R. (1996). The Internet
Transition Economies. Journal of Consumer Policy,
and International Marketing. Sloan Management
23, 257–283. doi:10.1023/A:1007138219753
Review, 38(Spring), 60–75.
West, D. C., & Paliwoda, S. J. (1996). Advertising
Roberts, M. S., & Ko, H. (2001). Global Interactive
adoption in a developing market economy: The case
Advertising: Defining What We Mean and Using
of Poland. International Marketing Review, 13(4),
What We Have Learned. Journal of Interactive
82–101. doi:10.1108/02651339610127266
Advertising, 1 (2),
World Advertising Resource Center (WARC)
Root, F. R. (1994), Entry Strategies for In-
(2004). European Marketing Pocket Book.
ternational Markets. Heath, Washington DC:
Lexington. Zandpour, F., Campos, V., Catalano, J., Chang, C.,
Cho, Y. D., & Hoobyar, R. (1994). Global Reach
Sheram, K., & Soubbotina, T. P. (2000), Beyond
and Local Touch: Achieving Cultural Fitness in
Economic Growth: Meeting the Challenges of
TV Advertising. Journal of Advertising Research,
Global Development, Washington D.C.: World
34(5), 35–63.
Bank. Taylor, Charles R., P. Greg Bonner and
Michael Dolezal (2002). Advertising in the Czech
Republic: Czech perceptions of effective advertis-
ing and advertising clutter. In Charles R. Taylor Additional Reading
(ed.), New direction in International Advertising
Research (Vol 12, pp. 137-149), San Diego, CA: Batra, R., Myers, J. G., & Aaker, D. A. (1996).
Elsevier. Advertising Management, Prentice Hall, Engle-
wood Cliffs, NJ.
Synodinos, N., Keown, C., & Jacobs, L. (1989).
Transitional Advertising Practices: A Survey of De Mooij, M. (1998). Global marketing and ad-
Leading Brand Advertisers. Journal of Advertising vertising: Understanding cultural paradox. Sage
Research, 29(2), 43–50. Publications, Thousand Oaks, CA.

1247
Understanding Brand Website Positioning in the New EU Member States

Duncan, T., & Ramaprasad, J. (1995). Standard- Lynch, P. D., Kent, R. J., & Srinivasan, S. S.
ized multinational advertising: The influencing (2001). The global Internet shopper: Evidence
factors. Journal of Advertising, 24(3), 55–67. from shopping tasks in twelve countries. Journal
of Advertising Research, 41(3), 15–22.
Ghose, S., & Dou, W. (1998). Interactive func-
tions and their impacts on the appeal of Internet Okazaki, S. (2004). Do multinationals standardise
presence sites. Journal of Advertising Research, or localise? The cross-cultural dimensionality
38(3), 29–43. of product-based Web sites. Internet Research:
Electronic Networking . Applications and Policy,
Ha, L., & James, E. L. (1998). Interactivity re-
14(1), 81–94.
examined: A baseline analysis of early business
Web sites. Journal of Broadcasting & Electronic Okazaki, S., & Alonso, J. (2003). Right mes-
Media, 42(4), 457–469. sages at the right site: on-line creative strategies
by Japanese multinational corporations. Journal
Hwang, J. S., McMillan, S. J., & Lee, G. (2003).
of Marketing Communications, 9(4), 221–239.
Corporate Web sites as advertising: An analysis of
doi:10.1080/1352726032000129908
function, audience, and message strategy. Journal
of Interactive Advertising, 3(2). Okazaki, S., & Alonso, J. (2003). Beyond the Net:
Cultural Values Reflected in Japanese Multination-
Jain, S. (1989). Standardization of interna-
als’ Web Communication Strategies. Journal of
tional marketing strategy: Some research hy-
International Consumer Marketing, 16(1), 47–70.
potheses. Journal of Marketing, 53(1), 70–79.
doi:10.1300/J046v16n01_04
doi:10.2307/1251525
Tharp, M., & Jeong, J. (2001). Executive insights:
Krippendorff, K. (1980). Content analysis: An
The global network communications agency.
introduction to its methodology. Sage Publica-
Journal of International Marketing, 9(4), 111–131.
tions, Newbury Park, CA.
doi:10.1509/jimk.9.4.111.19939
Laroche, M., Kirpalani, V. H., Pens, F., & Zhou,
Warden, C. A., Lai, M., & Wu, W. Y. (2003).
L. (2001). A model of advertising standardiza-
How world-wide is marketing communication
tion in multinational corporations. Journal of
on the World Wide Web? Journal of Advertising
International Business Studies, 32(2), 250–265.
Research, 43(5), 72–84.
doi:10.1057/palgrave.jibs.8490951

This work was previously published in Encyclopedia of Data Warehousing and Mining, edited by J. Wang, pp. 885-890, copy-
right 2005 by Information Science Reference (an imprint of IGI Global).

1248
1249

Chapter 5.4
WEB 2.0, Social Marketing
Strategies and Distribution
Channels for City Destinations:
Enhancing the Participatory
Role of Travelers and Exploiting
their Collective Intelligence
Marianna Sigala
University of the Aegean, Greece

ABSTRACT responsible for the development of city portals about


(a) the use of the major Web 2.0 tools in tourism
During the last decades, the use of Web 2.0 ap- and their impact on the tourism demand and sup-
plications for the generation, dissemination, and ply; and (b) the ways and practices for integrating
sharing of user-generated content (UGC) and the the use of Web 2.0 into their e-business model and
creation of new value added services are enormous. e-marketing practices.
Web 2.0 tools have tremendously changed the way
people search, find, read, gather, share, develop, and
consume information, as well as on the way people INTRODUCTION
communicate with each other and collaboratively
create new knowledge. UGC and Web 2.0 are During the last years, the number and use of numer-
also having a tremendous impact not only on the ous Web 2.0 tools, whereby Internet users produce,
behaviour and decision-making of Internet users, read and share multimedia content (User Gener-
but also on the e-business model that organizations ated Content, UGC), is mushrooming (eMarketer,
need to develop and/or adapt in order to conduct 2007a). It is estimated (eMarketer, 2007b) that 75.2
business on the Internet. Organizations responsible million USA Internet users currently use UGC, and
to market and promote cities on the Internet are this is expected to increase to 101 million by 2011.
not an exception from these developments. This eMarketer (2007c) also found that over 25 million
chapter aims to inform city tourism organizations USA adults regularly share advice on products or
services online.
DOI: 10.4018/978-1-60566-134-6.ch011

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

The Web 2.0 technologies and applications impact on tourism demand and supply; and b) the
(e.g. tags, RSS, blogs, wikis, podcasts, etc.) are ways and practices for integrating the use of Web
considered as the tools of mass collaboration, since 2.0 into their e-business model and e-marketing
they empower Internet users to collaboratively practices.
produce, consume and distribute information
and knowledge. In other words, Web 2.0 tools do
nothing more than realizing and exploiting the full Web 2.0 TOOLS IN TOURISM:
potential of the genuine concept and role of the USE, IMPACT AND APPLICATIONS
Internet (i.e. the network of the networks that is IN CITY MARKETING
created and exists for its users). This has tremen-
dously changed the way people search, find, read, RSS (Really Simple Syndication)
gather, share, develop and consume information, as
well as on the way people communicate with each Definition, Features and Use
other and collaboratively create new knowledge
(Sigala, 2008). UGC and Web 2.0 technologies are RSS allow users to subscribe to a webpage for
also having a tremendous impact not only on the receiving new content, e.g. subscribe to receive
behavior and decision-making of Internet users, online distributions of news, blogs, podcasts etc,
but also on the e-business model that organizations and so, RSS allows the creation of links and in-
need to develop and/or adapt in order to conduct teractive communication amongst other Web 2.0
business on the Internet (Bughin, 2007). applications and users. This is done either through
The tourism industry is not an exception from a news aggregator (similar to an email inbox) or a
such developments. On the contrary, as informa- news reader (a web-based environment) (Winer,
tion is the lifeblood of tourism, the use and dif- 2005). By doing so, one does not have to visit
fusion of Web 2.0 technologies have a substantial each individual website that he/she is interesting
impact of both tourism demand and supply. Indeed, to read any new information, but rather the RSS
more than ¼ of Internet users have used a weblog feeds all new updated information to the users’
to review information about a destination or travel RSS reader. RSS readers enable Internet users to
supplier in the last 12 months (Harteveldt, Johnson, gather and read all new information that is custom-
Epps & Tesch, 2006), many new Web 2.0 enabled ized to the user’s profile within one consolidated
tourism cyber-intermediaries have risen challeng- message. Many free RSS exist on the Internet, e.g.
ing the e-business model of existing online tourism FeedDemon, NewsGator, Rojo, software on the
suppliers and intermediaries who in turn need to website of Google™, MyYahoo®, etc.
transform their e-business model and e-marketing RSS allows new communication and interac-
practices in order to survive (Adam, Cobos & Liu, tion modes with information (Figure 1) (Farmer,
2007). As the Internet plays an important role 2004). In e-mail, the control of the communication
for the e-marketing of city destinations (Sigala, channel is held entirely by the instigator of the
2003; Yuan, Gretzel, & Fesenmaier, 2006), Web communication. Consequently, e-mail commu-
2.0 tools and applications also create both threats nication is characterized at times by flame wars,
and opportunities for organizations developing antisocial behavior and feelings of intrusion by
and maintaining destination management systems the participants, while the information artifact is
and portals. In this vein, this chapter aims to in- transitory, unfixed and not archived except in indi-
form city tourism organizations responsible for vidual instances. In discussion boards, information
the development of city portals about: a) the use artifacts are fixed, frequently archived and can be
of the major Web 2.0 tools in tourism and their interacted with through threading and comments,

1250
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

Figure 1. Communication types

but for accessing information, the user must delib- Business Applications
erately visit a dedicated online area. In contrast, for City Marketing
in RSS, both the communicator and the reader of
information have control of the communication As RSS is an information distribution technology
process, i.e. the former sends information only to that is characterized as a demand pull rather than
those that users have selected to aggregate the RSS a supply push model, many tourism destination
feed, while the later select from where and how organizations have adopted and incorporated RSS
(e.g. summaries, titles, or full entries) to receive feeds in their websites in order to communicate
communication. Further, as most RSS aggregators with their potential and current travelers in a less
are either integrated with or stand-alone desktop/ disrupted and personal way. Some examples of
web applications, readers need only to check the RSS include:
aggregator for new items.
• Keep a communication with their travel-
Impact on Tourism Demand ers such as sending them Newsletters and/
or updates of the programme of cultural
RSS feeders have a tremendous impact on the way events organized in the city.
consumers search and read information nowadays. • RSS helps organizations to enhance their
RSS has the following benefits for users: Search Engine Optimization by creating
inbound links to a company’s website and
• saves a lot of time spent on information by informing search engines whenever
searching; new content is uploaded on a website, so
• provides users with consolidated personal- that they can index it.
ized information; • RSS is used for syndicating content to other
• is less obstructive and more personalized Websites expanding the original website’s
to users’ interests than other Internet based readership and reach.
communication, and so, RSS entice sub-
scribers to visit the related websites, thus For example, the destination marketing or-
helping in building website traffic and ganization of Las Vegas has included an RSS on
visitation; its website (visitlasvegas.com), whereby users
• RSS boosts viral marketing and online can subscribe to feeds that automatically notify
word-of-mouth as users tend to forward them of current travel specials. Almost all of the
items in RSS feeds to their friends, fam- information (e.g. events’ news, weather updates,
ily and co-workers, much like the ‘forward special offers, etc.) on the official city website of
this message’ feature in eNewsletters. Dublin created by the city tourism board (visitdub-

1251
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

lin.com) is available to any traveler and/or other blog’’ collaborative feature enabling also
website through RSS. RSS are offered for free for multi-person weblog.
anyone for reading and/or enriching his/her own • Web-based: blogs are easy to access and
web site, provided that the latter follows the proper frequently maintain by simply using a web
format, terms and conditions and attribution, e.g. browser.
attribution such as “Content provided by Dublin • Community-supported: Weblogs can link
Tourism”. In such a way, visitdublin.com aims to other weblogs and websites (e.g. photos,
to enhance readership of its content, continuous videos, web-texts), enabling the synthesis
personal communication with its customers, viral and linkage of ideas from different users,
marketing, and search engine optimization through and so, stimulating meta-knowledge, i.e.
content syndication and incoming links. the generation of new knowledge through
sharing.
Blogs (or Weblogs) • Automated: Blogging tools are easy to
create and maintain without the need of
Definition, Features and Use HTML programming skills and knowl-
edge; so, bloggers can solely concentrate
Weblogs began as personal writing spaces that on the content.
store and update regularly multimedia content (in
reverse chronological order) and links of interest Du & Wagner (2006) also identified 3 types
to the author. Thus, blogs are used for recording of blogging tools and features: type I features of
its author’s journey and sharing it with others by blogging (such as text, diary, hyperlinks, user
using links, RSS, trackbacks, comments, taglines, friendly editor) provide easy-to-use and learn tools
archives, permanent links, blogrolls, etc. (Blood, for editing, presentation, publishing and interlink-
2000). Weblogs are defined as a ‘‘… site consists ing of content. Such blogs are heavily used by
of dated entries’’ (Blood, 2000), whereby entries those that solely seek a channel of self-expression.
are episodic or conversational in a diary or ‘‘story Type II blogging tools (such as Indexed archive,
telling’’ format. Motivated by different reasons search, “permalink”: a permanent URL for each
(Forrester Research, 2006), such as documenting weblog entry enabling referencing of specific
one’s life, providing commentary and opinions, past entries like other web-source. ‘‘Trackback’’:
expressing deeply felt emotions, articulating ideas a reverse hyperlink tracking the referrer weblogs
through writing, and forming and maintaining ‘‘making these formally invisible connections
community forum, weblogs (or blogs) are a ‘‘new visible’’, categorisation & syndication) are used
form of mainstream personal communication’’ by bloggers who wish to easily share rich media
(Rosenbloom, 2004, p. 31) for millions of people (e.g. videos, pictures etc), to have a sophisti-
to publish and exchange knowledge/information, cated content management system and to enable
and to establish networks or build relationships between-blog commenting or hyper linking, e.g.
in the world of all blogs. Indeed, blogging tools through ‘‘permalink’’ or trackback (Blood, 2000).
enabling between-blog interactivity are building Nowadays, the emerging Type III of blogging
up the ‘‘blogosphere’’ whereby social networks tools provide improved content distribution and
among bloggers are created. Du and Wagner (2006) between-blog connectivity (e.g. ‘‘pingback’’, alert
identified the following characteristics of blogs: of other bloggers’ comments or new posts), as well
as integrated applications for further enhancing
• Personalized: blogs are designed for in- social networking and community building such
dividual use and their style is personal as the following examples:
and informal. Blogger.com offers a ‘‘team

1252
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

• Workflow or project management (e.g. Internet users and increase traffic on a website, 2)
Lycos Circles workflow for a party, from create Interest to users who can now seek more
invitation to management of responses and and additional information, 3) develop someone’s
to travel directions), Desire to also visit a destination and/or buy the
• Polls, Intrasite messaging (e.g. ModBlog product and 4) foster an Action (e.g. book a hotel
allows users to track friends’ newest en- or organize a trip to a destination). Of course, it
tries, or to know who are the most ‘‘recent should be noted that the power of blogs can also be
visitors’’), negative, i.e. spread a bad experience of a tourist
• Web invitation, picture/music sharing (e.g. to million of online Internet users. Therefore, it is
MSN Space picture/music sharing, and re- very important that tourism companies authorize
mote posting of updates via email or mo- a public relations staff as the responsible repre-
bile devices). sentative of the company to first scan and read
blogs’ content and then respond to formally any
Impact on Tourism Demand positive and negative users’ comments. Guidelines
and corporate policies for responding to UGC
Numerous examples of general and-or (content or should also be established. Nowadays, many
user) specific blogs have been created in the tour- blogs take the form and are presented in a video
ism industry, such as tripadvisor.com, hotelchatter. format (Vlog, video blog). The first travel website
com, bugbitten.com, placeblogger.com, realtravel. to implement vlogs exclusively is endlesseurope.
com, travelpod.com, igougo.com, gazetters.com com. Due to the multimedia features of video
(a B2B blog for travel agents). Many travelers content and the intangible nature of the tourism
and tourists also develop and maintain their own product, it is argued that vlogs are going to have
blogs for sharing their travel experiences with a much greater impact and influence on travelers’
others and distributing their feedback (reviews) decision making and evaluation of alternative
of travel suppliers for achieving fun, social rec- tourism products and suppliers.
ognition, prestige and-or self-expression. Due to
the unbiased information shared in blogs based Business Applications
on first-hand authentic travel experiences, many for City Marketing
travelers tend to use and trust blogs’ information
for searching for travel information, tips and Blogs generate and distribute a plethora of UGC
selecting travel suppliers and destinations. Blogs related to travelers’ experiences, suppliers’ and
have the power of the impartial information and destination reviews, travel tips and advices. City
the electronic word-of-mouth that is diffusing destination management organizations can exploit
online like a virus. Hence, blogs are becoming and use such content for:
a very important information source for interna-
tional travelers for getting travel advice and sug- • monitoring and influencing electronic
gestions. Moreover, when reading others’ travel word-of-mouth;
experiences through weblogs, this also creates to • conducting an easy, free, timely and reli-
the reader the willingness to travel and visit the able market research about travelers’ pref-
same destination or suppliers. Indeed, research erences, feedback and profile;
has shown that UGC at blogs has a similar AIDA • communicating with current and prospec-
effect to users as paid advertisements have. The tive travelers in a very personal and infor-
latter is because blog’s content can (Lin & Huang, mal way;
2006): 1) Attract the attention, eyeballs of other

1253
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

• gathering travelers’ feedback and respond- the triplog) enabling Dutch travelers and locals
ing to customers’ complaints; to share and post the experiences they lived in
• enhancing search engine optimization: Holland (http://us.holland.com/blog.php?sf=e.
blogs are becoming very important tools pagerank&so=DESC&sel=popular). In this vein,
affecting information search since their blogs are becoming a useful tool for enabling local
links, content and popularity can dictate communities to get more involved in destination
the position of a company on a search en- marketing, communicate and blur with tourists.
gine search. As a result, blogs and web 2.0 tools can support
and foster community participation in city tourism
There are many search engines that one can use development and marketing practices. Community
for identifying and locating weblogs. The most participation can ensure a better blend between
popular one is technorati.com, which also provides locals and tourists reducing any inter-cultural
statistics about the online activity of weblogs, e.g. conflicts, creating social relations and respect
about the popularity of a blog and its potential and understanding amongst different cultures as
influence on search engines results. well as enabling multi-stakeholder understanding
Apart from exploiting others’ blogs, many and communication in tourism decision making
tourism suppliers and destination management and activities.
organizations have also adopted a pro-active
strategy by creating and incorporating blogs on Social Networking–
their own websites. For example, Marriott has Collaborative Networking
created its own blog on its website (www.blogs.
marriott.com), while Starwood has created a blog Definition, Features and Use
to communicate with its Preferred Guests and
enhance their loyalty through the website www. Social networking websites enable users to create
thelobby.com. Company initiated and moderated their profile and invite others with similar profile
blogs can offer the following benefits: solicit and to take part in their online community. The most
gather feedback from customers; conduct free popular websites such as myspace.com and bebo.
online market research; become recognized as an com reflect the willingness of Internet users to
expert on a specific topic; communicate and update transform websites as a gathering place of people
your customers with latest news; and use others’ with similar profiles.
customers’ suggestions for helping customers
select and evaluate products such as what amazon. Impact on Tourism Demand
com is doing by allowing users to upload books’
reviews on its website. For example, Eurostar has Many social networking websites have been cre-
initiated a blog (www.voiceofacity.com) whereby ated in the tourism industry allowing travelers
it has commissioned local Parisians to post blogs and prospective travelers to network with one
for creating a travelers’ guidebook with a truly another based on shared interests or attributes,
ground-roots feel. The destination organization such as tripmates.com, gusto.com, triporama.com,
of the city of Los Angeles has created a blog triphub.com, traveltogether.com and wayn.com.
supporting the sharing of bloggers’ experiences Travelers log into websites and create a personal
and insights on their adventures of the diversity profile detailing their travel experience and inter-
of Los Angeles’ arts and culture (http://blog.expe- ests, then network with others o share travel advice
riencela.com/). The fully Web 2.0 enabled official and stories, and even plan trips together. Hence,
portal of Holland features a blog capability (i.e. social networking websites have a tremendous

1254
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

impact on how tourists nowadays create, organize the website, upload their experiences, stories,
and consume tourism experiences. Many tourists pictures, videos and comments for sharing them
nowadays prefer to have the reassurance of other with other website visitors and users. The social
users sharing similar profile with them that the networking of Sheraton’s website enables potential
trip, the travel company, destination and/or the travelers to organize and book their holiday and
itinerary that they have scheduled is a good one hotel experience at any Sheraton property that
and it matches their preferences and tastes. Many matches their profile and preferences by reading
tourists also wish to use the Internet for collabora- and reflecting on the comments and first-hand
tively organizing a group trip with their friends. experiences written by previous Sheraton custom-
Websites such as tripup.com, traveltogether.com ers. The impact of social networking features for
and travelpost.com enable tourists to create an persuading potential travelers to select a particular
itinerary, e-mail and share it online with others, hotel and/or destination is very powerful, because
who in turn can edit, modify and enhance it, post it through social networking websites, travelers can
back to others for further comments and / or invite search website content based on keywords and
and read other travelers’ comments and advices on stories contributed by other travelers that may
the trip they organized in order to finally achieve be more relevant and make more sense to them
a consensus and proceed to a group booking. than keywords and experiences being created and
pushed by the website developers themselves. For
Business Applications example, on Sheraton’s website one can search an
for City Marketing hotel experience based on the comments and tags
contributed by a previous guests referring to “nice
Since sharing travel experiences in a social website walking in beaches nearby Sheraton hotels” or
can significantly inspire travel and boosts one’s “relaxing family holidays in Sheraton properties”
willingness to visit a destination or supplier, rather than using the Sheraton’s search engine to
several tourism websites are incorporating social find hotel based on its location, facilities etc.
networking tools in their e-business models. For Nowadays, many city destination organiza-
example, existing cyberintermediaries, such as tions have also incorporated social networking
Yahoo!® Trip Planner has adopted a collabora- features into their e-business model and strategy
tive trip organizing and booking tools. The of- in order to further enhance their communication
ficial website of the city of Philadelphia has also with customers and take benefit of the electronic
features a collaborative trip planning tool (www. word-of-mouth that they can create. For example,
planit.pcvb.org), that potential travelers can use the official website of Los Angeles invites any
for organizing their itinerary in Philadelphia with cultural and art organization-institution to become
friends as well as soliciting feedback and com- a partner with experiencela.com in order to share
ments from other travelers and locals. and distribute related content on their website.
Lufthansa created and operates its own social Experiencela.com has also created a cooperation
networking website, named as Jetfriends, for en- e-business strategy with the social networking
abling its young flyers to share flight experiences websites clickr.com and myspace.com. Specifi-
and indoctrinate them into the Lufthansa brand cally, the destination management organization of
and frequent flyers’ club (http://www.jetfriends. LA created a special webpage for LA on flickr.com
com/jetfriends/kids/). Sheraton also re-organized (http://www.flickr.com/groups/21164279@N00/
and re-designed its website (which is nowadays pool/) and myspace.com (http://www.myspace.
titled as the Sheraton Belong Neighborhood), com/experiencela) in order to enable the users
whereby Sheraton’s guests can subscribe to of the formers that are also funs and travelers

1255
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

of LA to share personal photographs, comments the search behavior of users with similar profiles
and stories about LA. The city of LA has realized and mental maps with them (personalized social
that such UGC can crucially drive traffic to their searching). In this vein, tagging has a great effect
website, boost their search engine optimization on how search engines identify and present infor-
strategy, instil travelers’ desire to visit LA and mation results in keyword searches to users.
use customer intelligence for providing reliable Flickr.com represents the first wide-spread
and timely advice and suggestions for trips to of tags, whereby users can add their own tags to
potential travelers to LA. any photo they wish to share, aggregate pictures
into photosets, create public or private groups,
Tagging (Social Search search photos by tags and easily add flickr-stored
and Tag clouds) photos to a blog. Nowadays, there are numerous
websites enabling tagging and searching based
Definition, Features and Use on tags, e.g. del.icio.us, a bookmarking service,
Technorati, a blog cataloging site, and digg, a
Tagging represents a new way for categorizing gathering place for tech fans. These sites create
information. Users tag a piece of content (e.g. an clickable “tag clouds” for resources, groupings of
audio, a picture, a word) with a meaning (a word tags arranged alphabetically, with the most used
or phrase) and then this information is categorized or popular keywords shown in a larger font. In
in categories based on this meaning. Community this way, these websites present other websites
tagging is a bottom-up, grass-roots phenomenon, that users think are important or relevant to them.
in which users classify resources with searchable Many such sites make use of RSS to notify in-
keywords. The tags are free-form labels chosen by terested users of changes and new developments,
the user, not selected from a controlled vocabulary. e.g. in flickr.com, RSS feeds can be attached to
Tagging is also known as consumer-generated tax- individual tags, or to photos and discussions.
onomy. Forrester (2006) defines tagging as ‘the act In addition to RSS, flickr.com and other social
of categorizing and retrieving Web content using networking sites typically offer functions such
open-ended labels called tags’. Tagging provides as search (for users and tags), comments (and
customer value, because it allows them to assign comment trails), and APIs (application program
their own word or words to mark products and interfaces) for posting to or from the tools, that
content online in order to categorize content that can be used in combination with blogs. An inter-
they find relevant, i.e. such as what bookmarking esting use of RSS that is combined with tagging
allows users to do. Words that users choose for is the Flashcard exchange, where, one can view
categorizing website and content then become a or subscribe to all flashcards posted for learning
navigation shortcut that a person can use to browse Spanish (or other languages).
and search content throughout that site. Although the tagging process is by no means
Tagging is used not only for saving and sort- simply technical—a way of categorizing resourc-
ing a user’s content but also for sharing content es—it has a strong social dimension as users of the
with others. Websites with tagging capabilities website find common interests and create on-line
can also allow users to share their personal tags communities. It represents another example of
and navigation ways with other users. Moreover, the fuzziness separating consumers and creators
some tag enabled websites enable users not only to on the Web. A contribution to a tagging site,
share their tag navigations, but also their profile. In seen by other users, may cause additional tags
this way, users can see who has tagged something, or comments to be added, automatically build-
and try to search and find information based on ing and updating and thus ultimately defining a

1256
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

resource. Instead of one person making a judg- Business Applications


ment about a blog entry, photo, or other resource, for City Marketing
a consensual classification is created. In effect,
a text or object identifies itself over time. This Because of the power of the folsconomy to provide
creation of “folksonomies” (as the user defines enhanced user-value and influence the results/page
how to sort information which in turn defines how ranking of search engines’ search, many tourism
others search and find information) can be seen firms nowadays include and consider tagging when
as a democratic implementation of the Semantic designing their websites and e-business strategy.
Web. Thus, for some, tagging helps and boosts the For example, Thomson’s website provides an
creation of the semantic web (Web 3.0), whereby affiliate link to deli.ci.ous.com (http://www.thom-
web content and search is directly related to its son.co.uk/), so that its users can tag and sort its
meaning for the users. website content through this technology.
However, tourism firms may adopt different
Impact on Tourism Demand strategies regarding the way they use and incor-
porate tagging into their websites (Epps et al.,
Several websites offer the capability for users to 2007). For example, Triporama.com has launched
sort, share and categories travel related content its tagging system, titled “Triporama Bookmarks”,
based on tagging, e.g. flickr.com (for pictures which allows its website users to download a free
sorting and sharing), travbuddy.com (for travel software in their PC for tagging content in their
experiences sharing), travelistic.com (for tag- own words from anywhere on the Web to their
ging video content). However, although more Triporama group trip plan, which they can then
and more users are using collaborative tools for share with their travel companions. Such a solu-
identifying and sorting travel content, tagging is tion provides differentiation customer value for
still an emerging technology: only 5% of USA Triporama, because as travelers visit many differ-
online leisure travelers—slightly more than 5 mil- ent Websites while planning a trip, Triporama’s
lion of the 114 million USA adults who travel for bookmarking tool lets travelers collect, label, and
leisure and go online regularly—tag Web pages or share the content they have found on the Web with
other content on sites like del.icio.us or Flickr™ other members of their group. Users also have the
(Epps, 2007). Moreover, the social capabilities of option to make their tags publicly available, while
tagging for community building and social col- Triporama also aims to edit and curate these public
laboratively construction of concepts’ meaning tags into features like “top 10” lists to give other
and of travel experiences could have numerous users ideas for planning their own group trips.
innovative applications in tourism as well. For When redesigning its website, Sheraton introduced
examples, travelers may be enabled and offered its “Vacation Ideas” feature whereby guests are
the opportunity to build structural tags in a text invited to write stories about their hotel stay, users
using XML for creating word groups or simply give Sheraton their consent for publishing their
finding appropriate keywords to describe a travel stories online and the entire story becomes a tag.
experience. This would offer additional options A tag cloud is created, titled the “Buzz Barom-
to Internet users to collaboratively develop travel eter”, whereby word occurring most frequently
itineraries and search of travel information with in stories appear in bigger fonts, while based on
others sharing a similar profile with them. the number of stories shared containing different
words a “Vacation Ideas page” is created (http://
www.starwoodhotels.com/sheraton/index.html).
For example, by clicking on the “Beach” guide

1257
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

brings up the five hotels whose stories mention users to tag their own or others’ Trip Plans with
“beach” most frequently (weighted by the number suggested or custom tags, which are later analyzed
of stories relative to the size of the property). By and used by Yahoo! for identifying and feeding
making storytelling the method by which the tags recommendations on the Yahoo! Travel home.
are created, Sheraton has made tagging so friendly Users can set their preferred level of privacy at
and easy for its guests that they do not even know the level of the Trip Plan (private, shared with
that they are tagging content. Sheraton benefits invited friends, or visible to any user). Users are
from this tagging strategy because: provided around 30 tags (e.g. (“budget,” “luxury”,
“weekend”, “honeymoon”) to choose from for la-
• It helps first time website users: Vacation beling trip plans from the style of the trip, however,
Ideas gives travelers, not knowing where Yahoo! monitors the tags used most frequently by
they want to go and/or not familiar with its users for augmenting its list of proposed tags.
the Sheraton brand, a more creative, user By using taxonomy-directed tagging, Yahoo!®
friendly and understandable way to search eliminates many of the inherent problems of folk-
and book hotels than the customary desti- sonomies created when users label similar things
nation-based and company pushed search. differently using synonyms or different forms of
• It helps Sheraton to build long lasting rela- the same word (e.g. lodging, accommodation,
tions with its guests by maintaining a close hotel etc.). Tagging has helped Yahoo!® to: a)
relationship with the travelers before, dur- make its content (750,000 Trips Maps, photos,
ing and after the trip. This is because the users’ comments etc.) more useful, accessible,
website provides guests having stayed at a searchable and understandable to its users; and
Sheraton’s hotel to return to the Sheraton’s b) to gather, analyze and use customers’ intelli-
website in order to contribute, solicit or gence (where they live, where they have traveled,
read other travelers’ stories. and what content they have viewed) and further
• It improves organizational learning, since refines that knowledge through the lens of the
Sheraton gets insights into Sheraton’s ho- tags they use to search in order to create targeted,
tel properties and customers’ experiences. personalized recommendations for destinations
Instead of conducting expensive and time and deals sold through its vertical search website
consuming research, Sheraton uses tagging Yahoo!® FareChase. In other words, tagging helps
as a simple and reliable way for gathering to further refine the collaborative filtering process
customer feedback and intelligence about that Yahoo! uses in order to provide personalized
its products and services. Based on the cus- recommendations and suggestions to its users.
tomer knowledge that is gained, operation Personalized suggestions for cross and up sales
managers can improve organizational pro- can significantly drive and enhance booking and
cesses, while marketers get to know what sales levels as well as provide additional functional
and how customers think and talk about and emotional value to website users that in turn
the brand, in order to better position the enhances customer loyalty.
Sheraton brand in the market and enhance In the same vein, the official destination portal
the guest experience at the hotels in ways of Holland uses tagging technology in order to
that reinforce guests’ perspective of the provide travelers an easy way to search the website
Sheraton brand. content and its multimedia information (video,
photos etc.) (http://us.holland.com/). Actually, tags
Yahoo!® Travel introduced new tagging fea- are used as an user constructed and defined search
tures into its Yahoo!® TripPlanner that enable engine rather than providing a search engine de-

1258
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

signed by the website developer that reflects a top videos, and podcasts), they later struggle to
down business defined search process. Tag clouds make this UGC relevant and accessible to
appear on the left with different font sizes to reflect their users and tags can help in addressing
words used more or less frequently, while “Top the latter.
10” suggestions for each tag (e.g. restaurant) are • City tourism organizations can gather reli-
also constructed and updated continuously when able and timely customer intelligence and
new content and UGC is shared on the website. feedback regarding the image of their des-
The portal also provides users the possibility to tination, the mental maps of their travel-
comment each others’contributions and comments ers and how they view and perceive their
as well as to tag the Holland’s webpage content destination etc. Such customer knowledge
by using different social booking technology such can be later used for marketing campaigns
as Digg and Furl. as well as for improving the products and
Overall, it becomes evident that city tourism services of the city as a destination.
organizations should consider including tagging • Customer information gathered through
into their websites, as tagging can help them social tagging can also be used for improv-
overcome the following issues (Epps, Harteverldt ing search engine optimization campaigns.
& McGowan, 2007): For example, words used frequently as
tags by travelers can be used as metatags-
• Very frequently websites do not speak the metadata for building the portal’s website
same language and they do not use the as well as for spending money on keyword
same terminology as their users. City des- sensitive search engines such as Google™
tination organizations should consider us- AdWords.
ing tagging in order to make their website
content more accessible, understandable However, when deciding whether and how to
and appealing to its users. For example, use tagging, city tourism organizations responsible
the marketer of a city might promote as the for the development of the city portal should also
major value of the destination its easy ac- decide the process, the policy and the way for
cessibility by air transportation, however, developing their tagging system regarding the
travelers may perceive as the most valu- following issues: a) does the company edit the tags
able feature of the destination the fact that incorporated by users? This is important specifi-
it is “a safe city to walk around”. cally if tags are uploaded with spelling mistakes or
• Tagging can help and further enhance they include anti-social and embarrassing words.
keyword search by supporting nuanced, In other words, editing and a clear policy may be
adjective-based searches. Tagging also en- required in order to protect the consistency, the
ables social search whereby users can see ethos and the good image of the website; b) are
who has tagged something, how credible the tags and taggers’ profiles made publicly avail-
or relevant its suggestions are based on his/ able for everyone? What consent and agreement
her profile and his/her evaluation by other are required to take from the users and how the
users. privacy policy of the website should be amended
• Tagging helps organize and display user- to incorporate the former?; c) is a software going
generated content uploaded on websites. to be purchased to manage the tagging process or
As more and more city tourism organiza- is this going to be done manually? Are the required
tions invite their users to upload and share and skilled labor sources available? and d) how tags
their UGC (e.g. reviews, itineraries, photos, are going to be created? Are the tags going to be

1259
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

provided by the users or are the tags restricted by words, ‘a wiki can be a blog, but a blog does not
the website owner? Forrester (2006) recommends have to be a wiki’.
that companies use taxonomy-directed tagging, as
it makes the tagging process more efficient and Impact on Tourism Demand
easier to use and it promotes consistency among
tags. This is because when users create a tag, they The most popular wiki is the famous online en-
can choose from existing suggested tags, or they cyclopedia, titled wikipedia.com, that is created
can add their own. and continually updated by its users. In tourism,
wikitravel.org represents a wiki based effort of
Wikis Internet users to collaboratively create and con-
tinuously update an online global travel guide
Definition, Features and Use including world-wide destinations. The number
of readers, creators and content at wikitravel.org
Wiki is a piece of server software permitting users are continuously mushrooming. At wikitravel.
to freely create and edit (hyperlinked) content via org, one can find guides for any destination ir-
any browser and without the need to have access respective of its size and/or geographic location,
to and know to use any programming language. as well as create a guide for any destination that
A wiki is a collaborative website whose content he/she wishes. Wikitravel.com is further enriched
can be edited by anyone who has access to it. Wiki with other web 2.0 tools and technologies such
features include easy editing, versioning capabili- as maps, tags, podcasts etc.
ties, and article discussions. So, wiki technologies
enable users to add, delete, and in general edit the Business Applications
content of a website. Wiki users-creators are noti- for City Marketing
fied about new content, and they review only new
content. As a result, such websites are developed Many tourism organizations take the opportunity
collaboratively through their users, and a wiki to promote and create links to their websites
becomes a collaboratively expandable collection through wikitravel.com in order to create and
of interlinked webpages, a hypertext system for drive traffic to their own websites (http://wiki-
storing and modifying information—a database, travel.org/en/London). Many other destination
where each page is easily edited by any user with management organizations exploit and incorpo-
a forms-capable Web browser client. Neus and rate the wiki technology in their website portals
Scherf (2005) defined wiki as web content man- in order to enable its users (travelers and locals)
agement systems allowing collaborative creation, to collaboratively create and share their percep-
connection and edition of contents, while Pereira tions and mental images and opinions about their
and Soares (2007) defined wiki as a shared infor- destination. For example, the National Library of
mation work space that facilitates access to infor- Australia has included a wiki on its portal (http://
mation content, organizational communications, wiki.nla.gov.au) inviting users to share their un-
and group collaboration. In other words, wikis derstanding and knowledge of local Australian
represent another way of content publishing and dances as well as negotiate their meaning and
communication as well as for group collaboration. create metaknowledge by synthesizing different
In this vein, wikis and blogs have some similarities views and perspectives. The National Library of
but they differ regarding the notification of new Australia has also developed a wiki and social net-
content, editing format, and structure. In other work website (www.pictureaustralia.org) whereby

1260
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

users can share their pictures about Australia and creating content (audio or video) for an audience
tell their story. In this way, the Library aims to that wants to listen when they want, where they
help democratize history and establish a collective want, and how they want. Podcasting differs
memory of places and events around the country. from webcasting. A podcast has a persistent site,
Ancient Times website (http://ancient.arts.ubc.ca/ capable of synchronizing with a portable multime-
community.html) includes several collaborative dia device, e.g. an MP3 player or iPod, whereas
tools, such as a wiki, blog and an arts metaverse webcasting is streamed from the internet and
enabling any user and history student to collab- requires the user to be connected to the internet
oratively develop and negotiate the meaning and while playing or viewing the webcast files. Web-
construction of old cities and destinations, such casting is closely related to real-time downloading
as Giza and Athens. These cultural guides can and synchronous broadcasting. Podcasting adds
significantly enhance the appeal and the interpre- spatial flexibility to the temporal flexibility that
tation of the cultural artifacts of historical cities webcasting offers and affords itself for creating
and destination by providing several edutain- personally-customizable media environments.
ment services and benefits to their users/visitors Podcasting offers customer value in terms of the
(Sigala, 2005a). Other wiki applications can also flexibility possibilities to hear personalized con-
be provided on the city portals in order to boost tent whenever and whatever device one wishes,
website loyalty, repeat traffic, and travelers’ desire e.g. one can download the “Economist”’s or
to visit the destination. For example, a destination “CNN”’s personalized news’ and press releases to
organization can design and incorporate a wiki his/her iPod and listen to its favorite news while
on its portal for enabling potential travelers and he/she driving at work. As podcasting does not
locals to exchange and collaboratively develop rely on the visual senses, it allows users to carry
recipes of local dishes and food. out other tasks while listening.

Podcasting and Online Video Impact on Tourism Demand

Definition, Features and Use Tourism experiences are intangible. One can-
not experience, feel and try a travel experience
Podcasting refers to the uploading of audio and before he/she buys and before he/she travels to
video files by users on websites. The most popular a destination. As a result, the purchase risk of a
website for sharing such content with others is travel—tourism experience is high and it is diffi-
youtobe.com. Podcasting represents repositories cult to persuade a user for the qualities of a tourism
of audio (podcasts) and video (vodcasts) or “video service. Due to its multimedia features, podcasting
podcasting” materials that can be “pushed” to sub- helps users to better and easier evaluate travel
scribers, even without user intervention, through alternatives by experiencing in someway a travel
RSS aggregate feeds of audio and video content experience before they decide to buy and consume
facilitating users to search the latest services. it and/or travel to a destination. This is because
Podcasting-capable aggregators or “podcatchers” audio and video files of hotels, destinations, and
are used to download podcasts. These files can other travel products created and uploaded for
also be downloaded to portable media players that sharing by other users are considered as more
can be taken anywhere, providing the potential unbiased information and not staged experiences
for “anytime, anywhere” learning experiences produced by the supplier aiming to promote his/
(mobile learning). Podcasting’s essence is about her own product as the best one. Podcasting has

1261
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

also been used as mobile guides for travelers, e.g. Metaverses: Massively
Virgin Atlantic provides through its website free Multiplayer Online Role
podcasts-guides of cities whereby they fly to. Playing Game (MMORPG)

Impact on Tourism Supply Definition, Features and Use

Many tourism suppliers are using Podcasting Metaverses are three dimensional virtual worlds
as a marketing, information and customer com- whereby Internet users collaboratively play “on-
munication tool. For example, Jumeirah hotel line MMORPG games” with others. However,
uploads podcasts on its website for delivering and these platforms are wrongly perceived as “simple
updating its potential guests about what is hap- games” and “virtual” worlds, since they frequently
pening in its properties at every single day, and-or represent an extension to our physical day-to-day
delivering to website users and potential buyers world to which users add new socio-economic and
the experiences of VIPs that have stayed at their political situations. MMORPG are games that are
property. Tate Gallery enables their visitors that played by numerous players (e.g. millions of users)
have experienced their paintings and exhibition to and they could be considered as an intermediate
record themselves, upload their audio-video on the step from ‘computer’ to ‘ambient’ era. Some of
Tate Gallery website, and which others can later these games (e.g. World of Warcraft) develop
download and use them as a mobile interpreta- around a theme defining the goals of the game,
tion guide while visiting the gallery. Orbitz.com while other games, such as SecondLife.com, there.
provides podcasting of many destinations that com, cokemusic.com, habbohotel.com and http://
travelers can download to their MP3 players and play.toontown.com/about.php, encourage a free-
use them as guides while visiting the destination. style of playing, allowing the users to make what
In a similar way, MGM Grand Hotel Las Vegas they want out of it. Metaverse environments are
has lauded online video on its website under the internet-based 3-D virtual world whereby their
title “Maximum Vegas” in order to better illustrate users, called residents, can interact with each other
to its potential guests the experience and services through motional avatars (an internet user’s rep-
of its hotel and gaming resort. Similarly, city des- resentation of him/herself) providing an advanced
tination organizations should consider enhancing level of a social network service. Although it is
the content and marketing appeal of their website difficult to measure the size and growth of such
portals by enabling podcasting opportunities, i.e. games, it is estimated that the market for massively
either allow users to share content or push their multiplayer online games is now worth more than
own created podcasting content (e.g. http://www. $1bn in the West world (Book, 2003). For example,
visitlondon.com/maps/podcasts/, Podcasts at the one can simply consider the size of and growth
official portal of London). For developing pod- of Second Life®® itself. Second Life® has more
casts, city tourism organizations can outsource than 5 million users, while about half a billion
this function to companies such as soundwalk. US$ are being transacted every year on Second
com, podtrip.com and heartbeatguides.com that Life’s® website (as reported on SecondLife.com
specialize in the development and dissemination on April 2007).
of destination podcasts.
Impact on Tourism Demand

Tourists and travelers participate in such games


either for fun and-or for ways of expression of

1262
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

oneself and for achieving satisfaction through of its new brand Scion on SecondLife® and then,
task—accomplishment, self-actualization and widely broadcasted the new campaign in real life.
creation—design of something new. For example, Other companies, use SecondLife® for recruiting
many people dream and try to become and excel and identifying new talents e.g. CNN does head
on a profession that they could not achieve in hunting of new journalists online.
their real life, others try to design a new product Many destinations are also moving into the
and service hoping that other players will adapt futuristic world of virtual reality and metaverse,
and pay for it and so they can gain money and/or as many city and country destination organiza-
head hunters would spot their talents and recruit tions create their virtual destinations. Netherlands
them in their real or virtual companies. Tourism Board recently opened a national tourism
board in SecondLife® (http://us.holland.com/
Business Applications secondlife.php), the city of Galveston launched
for City Marketing a virtual replica of itself in SecondLife® (http://
www.galveston.com/secondlife/), providing their
Many tourism and travel related companies have visitors with the chance to become part of an in-
already created their representative offices and teractive community (Figure 10 and 11). The aim
headquarters in metaverse environments such as is to provide digital travelers the chance to take
SecondLife.com. Embassies (e.g. that of Sweden), guided virtual tours, learn about the history, culture
Tourism Authorities (e.g. that of Maldives) of and daily life of the destination, and interact with
many countries and many tourism companies (e.g. new virtual friends from around the world. Tourism
TUI, Burj Al Arab Hotel, Marriott, Costa Cruises) Ireland has also launched the world’s first tourism
have created their offices and companies on islands marketing campaign in SecondLife® (http://dub-
of SecondLife.com for boosting their marketing linsl.com/index.php) including the sponsorship of
practices such as enhancing customer communica- a range of events and activities, including concerts,
tion and education about their products/services, fashion shows, and photographic exhibitions, in
building brand reputation and user communities, Second Life’s® replica city of Dublin. Dublin’s
and achieving word of mouth (WOM) and adver- representation in SecondLife.com is the first place-
tising. Hyatt used residents of Second Life® and location in Ireland that the Tourism Board created
exploited their intelligence and knowledge for its representation in Second Life®. Similiar to
designing a new hotel concept, named as Aloft; the Dublin creation, Amsterdam in Second Life®
architects and guests were involved in designing comes complete with Dutch signs, canals, trams
the hotel providing their feedback, preferences and a lot of attention to detail. Overall, when
and specialist knowledge (read the related blog at investigating the impact of SecondLife.com on
http://www.virtualaloft.com/). As a result of the its residents’ behavior, it becomes evident that
popularity of the new hotel, the first Aloft hotel historical landmarks and buildings such as Tour
will open and operate in real life in New York Eiffel, London Bridge, Ajax Football stadium etc.,
in January 2008. Apart from collaborative new have a great effect in building virtual communities
product development, a firm can further exploit of people spending a lot of time on dwelling them.
the social intelligence gathered and generated at Since it is apparent that real world modern-day
SecondLife.com and other metaverse environ- cities and their landmark attraction are probably
ments in order to conduct market research and the most effective at driving and retaining visitor
to test new product ideas and new advertising traffic, city tourism organizations should exploit
campaigns, e.g. Toyota first tested the campaign this inherent advantage and exploit their cultural

1263
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

and heritage assets in metaverse environments triptie.com and the “Trip Planner” function on
for boosting their city brand name, recognition Yahoo!® Travel, which allow users to integrate
and promotion. content from other websites into the user’s own
itinerary planning toolkit on the host website.
Mash-Ups
Impact on Tourism Demand
Definition, Features and Use
Travel decisions are complex and involve the
Mash-ups describe the seamlessly combination of searching, comparison and combination of several
two or more different sources of content and-or information located in many different websites.
software for creating a new value added service For example, a decision to travel to a destination
to users. Many mash-ups enrich their services requires various and a plethora of information
with some geographical content such as Google™ about weather conditions, exchange rates, travel
Maps; e.g. The New York Times Travel Section’s and accommodation alternatives and prices, at-
“36 Hours In…” mash-up, which allows users to tractions etc. As a result, tourists increasingly
search the “36 Hours in …” story archive from a demand and expect to combine and cross-check
Google™ Map. For example, when visiting the information from different sources, so that they
website traintimes.org.uk, one can see on real can better and easier make a holistic decision.
time where trains are located and when they will For example, tourists may not be able to clearly
arrive at destinations, since the website combines understand where a hotel may be located when
information from Google™ maps, and informa- the description of the hotel websites states that
tion from the British rail website about train time the hotel is located on the beach, near the beach
tables, delays etc. etc. Tourists easily get confused from different
There are several mash-up applications in tour- descriptions found in different websites. On the
ism such as new cyberintermediaries including contrary, mash-up websites empowered with
mapping services (e.g. earthbooker.com, tripmojo. maps (e.g. earthbooker.com) enable users to see
com, reservemy.com) and meta-search engines where exactly a hotel or other attraction is located
such as farecompare.com. Other examples include: (sometimes even locate the exact orientation and
www.43places.com that combines Flickr photos, view of a hotel room and then decide whether to
RSS feeds and Google™ Maps with tagging and book this room at this hotel).
user-generated content, allowing users to share
their favorite destinations; www.randomdayout. Business Implications for City Marketing
co.uk combines a number of data sources to create
a mapped itinerary, using Virtual Earth (Micro- Mash-up applications have empowered the rise
soft’s equivalent of Google™ Maps). An innovate of new cyber- and info- intermediaries offering
example related to destination marketing manage- new sophisticated information services (e.g.
ment is the case of the city of Pennsylvania (http:// flightcompare.com search and compare all flight
www.visitpa.com): based on a project amongst information from different websites in order to
Google™ Earth, Carnegie Mellon University, provide comparable flight information within
NASA, the Pennsylvania Tourism Office and the one webpage to its users). Moreover, many tour-
National Civil War Museum, virtual tourists would ism suppliers and organizations also enrich their
have the chance to view Pennsylvania’s Civil website content with maps in order to make it
War trails online. More sophisticated examples more user friendly and useful to their visitors,
of mash-ups are the “Marco Polo” function on e.g. the official website of London and Dublin use

1264
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

Google™ Maps with geotags for enabling tourists trip, an adventure travel etc. In this vein, tourism
to identify points of interest, hotels etc. Moreover, decisions are very complex and risky. Indeed, the
many companies leave their software as an Open literature about information search in the tourism
Application Programme Interface (API), so that field has recognized the important role of WOM
users can create limitless combinations of their in travel planning and decision making (Hwang,
services. For example, backstage.bbc.co.uk rep- Gretzell, Xiang & Fesenmaier, 2006; Murphy,
resents BCC’s services and opportunities offered Moscardo & Benckendorff, 2007). WOM has been
to its users, who are enabled to take content from found to be one of the most influential information
the BBC, re-structure it and present it the way they sources for travel (Morrison, 2002). Research has
prefer. Enabling user innovation is another way also shown that those with past experience with a
that companies aim to exploit on users’ creativity specific travel destination and that engage in digital
and intelligence instead of investigating solely on word-of-mouth communication are most likely
company’s R&D efforts. to be the most preferred and the most influential
source of information in the pre-trip stage of travel
decision making (Crotts, 1999).
FUTURE TRENDS To make travel decisions easier, travelers
need to reduce the inherent information com-
It has become evident from the above mentioned plexity of travel decisions as well as the risk
analysis that the two major impacts of web 2.0 related to the service firm (i.e. is that a good and
and its UGC on consumer behavior and marketing reliable company), the service risk (i.e. is that
practices are: 1) the electronic word-of-mouth that a service that fits my preferences and needs?)
is created; and b) the opportunities to build and and the purchasing risk (i.e. is that a trustworthy
maintain customer communities for enhancing the booking and buying channel to use for buying
practices of Customer relationship Management a travel service?). To achieve that, consumers
and social marketing. Exploiting web 2.0 for city use recommendation-based heuristics and other
marketing can have a tremendous effect on the users’ feedback to reduce uncertainty, eliminate
marketing effectiveness, since, as the following the related risks as well as filter and process the
analysis and discussion illustrates, both previous plethora of information that must be processed
issues significantly affect consumer loyalty and when making decisions (Olshavsky & Granbois,
purchasing behavior. 1979). WOM-based information is heavily used
and trusted by consumers for taking travel deci-
Web 2.0 and Electronic sions, because it is seen as more vivid, easier to
Word-of-Mouth (WOM) use, and more trustworthy as it is based on actual
experience and typically provided without direct
Word-of-mouth (WOM) is very important in benefits (Smith, Menon & Sivakumar, 2005).
tourism and in services in general, since objective As demonstrated in the above mentioned
information about a service experience cannot be analysis of UGC in Web 2.0 websites, electronic
easily provided before one buys and consumes the WOM can take different names and forms such
services themselves. Services are intangible and so as virtual opinion platforms, consumer portals,
they are difficult to be tested, tried and evaluated social networking, blogs’ comments, tag words,
before buying them. Consumers also tend to rely podcasting, virtual communities and online feed-
more on consumer reviews when purchasing high back mechanisms (Armstrong & Hagel, 1997;
involvement products (Park, Lee & Han, 2007), Bellman, 2006; Sigala, 2008). Users of Web 2.0
such as several travel products e.g. a honeymoon websites and tools may post their own experi-

1265
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

ences, videos, share their opinion, give advice, or eMarketer (2007d) reports that nearly 6/10 con-
look for answers to their questions. Consumers sumers prefer websites with peer-written reviews,
also perceive electronic WOM to be a reliable and that websites with reviews experience greater
source of information (Gruen, Osmonbekov & conversion rates.
Czaplewski, 2006). Dellarocas (2003) identified Overall, Dellarocas (2003) summarized or-
three different characteristics of online WOM ganizations’ benefits of electronic WOM in the
relative to traditional WOM: 1) electronic WOM following: brand building; customer relationship
is larger in scale (both in terms of quantity and management; customer acquisition; addressing
people impact) due to the Internet’s low-cost and customer complaints; market research; product
networking features; 2) electronic WOM is a development; quality control and supply chain
powerful and reliable market research tool giving quality assurance activities.
organizations the ability to monitor on real time
their operations; and 3) it is difficult to convey Web 2.0, Customer Relationship
contextual cues (e.g. facial expression) through Management (CRM) and
the Internet and peer review websites for example, Social Marketing
and so not knowing or seeing who the informa-
tion provider is makes, it is harder to interpret The major aim of CRM is to personalize business
the subjective information in online interaction. services and products as well as develop a 1:1
To address this problem, websites often display communications and long lasting relation with
demographic or other data about reviewers (for profitable customers (Sigala, 2005b). eCRM also
example, the length of membership, their location, requires the development of customers’ communi-
etc.) in order to help build credibility and trust. ties for providing loyal customers with functional,
Websites may even provide the possibility to users emotional and social benefits and value (Sigala,
to upload and share their own feedback and evalu- 2006). The previous section provided practical
ation (by incorporating each review into a rating examples illustrating the way in which web 2.0
of the reviewer) about the quality of the reviews applications and tools enable the formation and
written by other members. Moreover, because Web development of customer virtual communities.
2.0 enables users to identify and use personalized By identifying and reviewing the limited related
and contextual information (e.g. look at what oth- studies that have been conducted so far, the fol-
ers’ with similar profiles are saying), electronic lowing analysis further supports the capability
WOM is considered as both more relevant and of web 2.0 to build virtual communities of users
unbiased than traditional WOM, whereby one and enhance the community benefits (functional
cannot easily track and relate the content with and emotional/social) to its users.
the profile of its original messenger. Ying and Davis (2007) and, Lento, Welser,
Smith, Menon and Sivakumar (2005) claim Gu and Smith (2006) illustrated how blogs create
that consumers prefer such peer recommenda- and maintain strong online communities through
tions over other forms of input, while Amis their social ties tools such as blogrolls, perma-
(2007) advocated that social network sites have links, comments and trackbacks. Indeed, many
as much influence on consumers as television and authors (e.g. Lin, Su & Chien, 2006; Ying &
more than newspapers. Statistics actually provide Davis, 2007) have started to apply social network
evidence of consumers’ reliance on electronic analysis for measuring and illustrating the social
word-of-mouth. More than 80% of web shoppers bonds, networking and communication structures
said they use other consumers’ reviews when created within the blogsphere. Li and Stronberg
making purchasing decisions (Forrester, 2006). (2007) summarized blogs’ benefits for firms as

1266
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

follows: search engine optimization; e-word-of- Fesenmaier (2004) illustrated that virtual tourist
mouth (eWOW); improved brand perception and communities are useful for managing customer
visibility; instantaneous client feedback; market relations by: attracting customers through in-
research and insight; increased sales efficiency; depth, focused and member-generated content;
and reduced impact from negative user-generated engaging customers through social interactions;
content. Damianos et al. (2007) advocated that and retaining customers through relation building
social bookmarking generates social influence with other members. Online communities also
and bonds as well as creates value by: enabling build customer value (Wang & Fesenmaier, 2004)
resource management, information sharing and by generating users with all types of relational
discovery, expert finding, and social networking; benefits namely functional, social, hedonic and
providing teams with a place to share resources; psychological (Gwinner, Gremmler & Bitner,
forming and supporting social networks around 1998). Kim, Lee and Hiemstra (2004) provided
interest areas; and feeding expertise finding and evidence of the impact of virtual communities on
user profiling. Awad and Zhang (2007) discussed travelers’ loyalty and product purchase decision
the marketing benefits of eWOW generated in making. Andersen (2005) explored the use of on-
online review communities and debated firms’ line brand communities for developing interactive
efforts and strategies addressing it. By examin- communication channels and establishing social
ing the communication tools and social cues of and structural bonds with devoted users. Jang,
myspace.com, Dwyer (2007) demonstrated the Ko and Koh (2007) showed that online brand
impact of social networking sites on developing communities posses and develop features - such
customer interrelations and communities. In their as, quality and credibility of information, service
study of videos’ tags on Del.ici.ous, Paolillo and quality, member interaction and leadership, brand
Penumarthy (2007) found that social tagging reputation and (intrinsic and extrinsic) rewards
can generate community benefits such as: easy for members’ activities- that in turn, contribute
retrieval (as users use words they can remember to increased users’ brand loyalty, commitment
and have useful meaning to them); contribution and sales. Erat, Desouza, Schafer-Juger and
and sharing; attract attention; opinion expression; Kurzawa (2006) discussed how different types
play; and self-presentation. Thus, since tagging can of communities of practice (e.g. B2C, C2C)
be used for providing functional services, creating can be used for acquiring and sharing customer
social ties, market research on users’ opinions and knowledge in order to improve business processes
interests, and WOW, social tagging’s ability in and performance. Beyond collecting customer
creating user communities is evident. Forrester knowledge, online communities can also be used
(2006) demonstrated web 2.0’s ability to generate for co-operating with customers for New Product
customer and business value in different processes: Development (NPD) and innovation (Rowley,
customer service (e.g. community self-service Teahan & Leeming, 2007). A plethora of cases
savings); sales (e.g. community loyalty and sales and research studies (e.g. in Lagrosen, 2005; Pitta
reduces commissions and price competitions); & Fowler, 2006) reflects the possibility to use
marketing (e.g. credibility of eWOW); production virtual communities for NPD as well.
(e.g. co-design reduces waste); and R&D (e.g.
community input raises success rate).
A significant amount of literature also high-
lights the business benefits from developing virtual
communities particularly in the area of CRM
and social marketing. Analytically, Wang and

1267
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

Table 1. Web 2.0 extended CRM implementa-


Proposed Models for Exploiting
tion
Web 2.0 in Enhancing Marketing
Low market integration High market integration Communication and CRM
Many-to-one Many-to-many
High customer integration

Target: clients’ networks Co-exploitation of The previous analysis and industry examples il-
Active customers’ customers’ profiles with
involvement other network partners
lustrate that web 2.0 tools and applications have a
e.g. Lonelyplanet.com, e.g. mash-ups, earthbook- twofold impact on the way CRM is implemented:
Sheraton.com er.com, flightcompare.com 1) web 2.0’s networking and connectivity capabili-
ties provide enormous opportunities to communi-
cate and co-operate with customers and industry
One-to-one One-to-many partners in many different directions (e.g. many-
Low customer integration

Target: individual custom- Ecosystems of partners to-one, many-to-many) (Table 1) the social intel-
ers offering a seamless
experience to individual ligence and knowledge created collaboratively in
clients (cross-selling, web 2.0 platforms (i.e. the user-generated content)
products’ bundling)
e.g. travelocity.com
can be exploited in different ways for identifying,
developing, enhancing and maintaining relations
with profitable customers (Table 2).

Table 2. Exploiting social intelligence for managing and enhancing customer relationships through
their lifecycle

Phase Type of customer information/ CRM implementation activities


intelligence
Acquisition Of the customer information: Create brand awareness and recognition amongst customers and virtual communities
transaction and personal data by building and supporting electronic word-of-mouth
Develop brand reinforcement and trust by educating and informing customers about
the brand, its services, functionalities etc
Use customer intelligence in order to identify and target new customers, e.g. clone
the profile of existing product-service users, use the connections and recommenda-
tions of existing customers etc.
Use customer intelligence to understand how customers use the service, what func-
tionalities they prefer or not
Use customer intelligence for profiling customers
Retention For the customer information: Use customer intelligence for enhancing customer service and transactions
relationship and product data
Use customer intelligence for personalizing services and products
Build and develop community of customers-users
Use customer intelligence for innovation & NPD
Expansion For the customer information: Use customer intelligence for cross selling, e.g. suggest compatible products based
relationship and product data on other users’ purchases
Use customer intelligence for up-selling
Use customer intelligence for developing affiliation and loyalty programmes
Win back By the customer information: Use customer intelligence (feedback, reviews etc) for identifying pitfalls and faults
feedback and monitoring data
Use customer intelligence and communities for handling customers’ complaints
Use customer intelligence and communities for monitoring and managing the firm’s
reputation, status and prestige

1268
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

In other words, CRM cannot anymore be con- lect and use three forms of customer information
sidered as synonymous to one-to-one communica- / intelligence in order to manage each relational
tion and personalized service at an individual basis. phase. “Of-the-customer” information includes
Web 2.0 augment CRM practices and implementa- customers’ personal and transaction data for un-
tion to include various forms of communications derstanding and measuring their profile, e.g. sales,
with clients and business partners. Following Gib- profitability, purchasing patterns, preferences.
bert, Leibold and Probst (2002), Table 3 reflects “For-the-customer” information refers to product,
a two dimensional matrix, whereby the vertical service and firm information perceived as useful
axis represents how firms integrate customers by clients for making more informed decisions.
into their value chains and the horizontal axis “By-the-customer” information reflects customer
represents the integration of business partners feedback (e.g. customer complaints, suggestions,
into the firm’s value chain. Companies can use reviews) used for new product development or
web 2.0 technologies to communicate and enable business improvement. As illustrated previously,
dialogues and interactions not only between them Web 2.0’s user-generated content mushrooms
and their customers, but also between customers these three types of customer information and
themselves (C2C), between business partners, provides firms with several opportunities not only
among all of them etc. When engaged in two to collect, but also to get access to such types of
directional communication both customers and customer intelligence. In other words, Web 2.0
partners, firms can involve the former in their platforms can be exploited as a free and real time
value chain in order to create customer value and market research and intelligence tool. Table 2
benefits. For example, as explained earlier, when summarizes how firms can exploit web 2.0 tools
customers communicate with other customers in and platforms for collecting and analyzing this
virtual communities, customers provide social and customer intelligence for augmenting and sup-
emotional support to others as well as functional porting their CRM practices.
benefits (e.g. free consultancy in trip planning). Overall, it becomes evident that web 2.0 en-
Also, when co-operating and sharing content abled CRM reflects a cultural shift from product
and applications with other businesses (e.g. in ‘designing for customers’ to ‘designing with’ and
mash-up websites), firms can collaborate with ‘design by’ customers. For firms to achieve such
and integrate other partners in their value chain a cultural shift, crucial organizational changes
in order to provide additional services to their should also take place. Importantly, the role of
clients, e.g. a holistic tourism product-services marketers should be changed from being sales
such as a dynamic packaging. people to becoming community builders and
Moreover, in developing successful rela- perceiving customers not as targets to identify
tionships with profitable clients, firms need and sell, but as partners to collaborate with. Firms
to understand and manage all phases through should also realize that they should use customer
which relations are evolved, as each phase is intelligence not only for learning about their
characterized by differences in behaviors and customers and identifying new target markets
orientations and so, it requires different CRM (opportunistic behavior), that they should also use
approaches. Theory and practical evidence has customer intelligence for learning and improving
shown that customer relations evolve over three processes and products with their customers as
major distinct phases related to the customer life- well as with different business partners (partner-
cycle (see Sigala, 2008): initiation, maintenance ship relation). In other words, firms derive and
and retention or termination. Hence, all CRM realize maximum benefits when they exploit web
implementation models reflect practices that col- 2.0 tools for establishing and maintaining co-

1269
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

creation and co-learning adaptable and flexible REFERENCES


ecosystems with their customers and business
partners (Table 3). Adam, J., Cobos, X., & Liu, S. (2007). Travel 2.0:
Trends in Industry Awareness and Adoption. New
York University and PhoCusWright Inc.
Conclusion Amis, R. (2007, May). You can’t ignore social
media: How to measure Internet efforts to your
Internet users and travelers are nowadays empow- organization’s best advantage. Tactics, 10.
ered to create and synthesize in their own way
the travel content that they also wish to distribute Andersen, P. H. (2005). RM and brand involvement
and share it with others through users’ controlled of professionals through Web-enhanced brand
distribution channels. In this vein, Web 2.0 tech- communities: Coloplast case. Industrial Market-
nologies enable Internet users to become the ing Management, 34(3), 285–297. doi:10.1016/j.
co-producers, the co-designers, the co-marketers indmarman.2004.07.007
and the co-distributors of tourism experiences and
Armstrong, A., & Hagel, J. (1997). Net gain:
services as well as the co-entrepreneurs of new
Expanding markets through virtual communities.
tourism products and new e-business models. As
Boston, MA: Harvard Business School Press.
the diffusion of Web 2.0 applications becomes
wide and consumers incorporate them within Awad, N. F., & Zhang, S. (2007, January). Stay
their daily and professional life, travelers expect out of my forum! Evaluating firm involvement in
tourism firms and organizations to provide similar online ratings communities. Paper presented at the
Web 2.0 enabled services. The previous analysis 40th HICSS, Waikoloa, Big Island, Hawaii.
aimed at identifying and illustrating the business
Bellman, S., Johnson, E., Lohse, G., & Mandel,
implications created for tourism and hospitality
N. (2006). Designing marketplaces of the artificial
enterprises as well as strategies and tactics that
with consumers in mind. Journal of Interactive
they can adopt for eliminating threats while ex-
Marketing, 20(1), 21–33. doi:10.1002/dir.20053
ploiting the arising opportunities. Therefore, as
Web 2.0 is here to stay, it is evident that unless Bickart, B., & Schindler, R. M. (2001). Internet
a city tourism organization adopts and incor- forums as influential sources of consumer infor-
porates Web 2.0 tools into its e-business model mation. Journal of Interactive Marketing, 15(3),
and strategies for marketing and managing its 31–40. doi:10.1002/dir.1014
destination, the competitiveness of the latter is
Blood, R. (2000). Weblogs: A history and perspec-
threaten. Nevertheless, in order to be successful,
tive. Retrieved October 15, 2007, from http://www.
the adoption and use of any web 2.0 tools should
rebeccablood.net/essays/weblog_history.html
be accompanied with appropriate organizational
and cultural changes within the firm regarding Book, B. (2003, July). Traveling through cyber-
the roles, job descriptions and tasks of its staff, space: Tourism and photography in virtual worlds.
users and business partners. Further research is Paper presented at the conference Tourism &
required in order to understand and examine how Photography: Still Visions - Changing Lives in
firms are achieving and trying to implement such Sheffield, UK.
organizational changes when incorporating web
2.0 into their e-business model.

1270
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

Bughin, J. R. (2007, August). How companies Epps, S. R., Harteveldt, H. H., & McGowan,
can make the most of user-generated content. B. (2007). Executive Q&A: Social tagging for
The McKinsey Quarterly, Research in Brief, Web ebusiness. Answers to E-Business professionals’
Exclusive. common questions about social tagging. Forrester
Research.
Crotts, J. (1999). Consumer decision making and
prepurchase information search. In Y. Mansfield Erat, P., Desouza, K., Schafer-Jugel, A., & Kurza-
& A. Pizam (Eds.), Consumer behavior in travel wa, M. (2006). Business customer communities
and tourism (pp. 149-168). Binghamton, NY: and knowledge sharing: studying the critical is-
Haworth Press. sues. European Journal of IS, 15, 511–524.
Damianos, L., Cuomo, D., & Griffith, J. Hirst, Farmer, J. (2004). Communication dynamics:
D., & Smallwood, J. (2007, January). Adoption, Discussion boards, Weblogs and the development
utility, and social influences of social bookmark- of communities of inquiry in online learning
ing. Paper presented at the 40th HICSS, Waikoloa, environments. In R. Atkinson, C. McBeath, D.
Big Island, Hawaii. Jonas-Dwyer, & R. Phillips (Eds.), Beyond the
comfort zone (pp. 274-283). Proceedings of the
Dellarocas, C. (2003). The digitization of word-of-
21st ASCILITE Conference.
mouth: Promise and challenges of online feedback
mechanisms. Management Science, 49(10), 1407– Forrester Research. (2006). Social computing.
1424. doi:10.1287/mnsc.49.10.1407.17308 Retrieved September 14, 2007, from http://www.
forrester.com.
Du, H. S., & Wagner, C. (2006). Weblog success:
Exploring the role of technology. International Gibbert, M., Leibold, M., & Probst, G. (2002).
Journal of Human-Computer Studies, 64, 789– Five styles of CKM, and how smart companies
798. doi:10.1016/j.ijhcs.2006.04.002 use them to create value. European Management
Journal, 20(5), 459–469. doi:10.1016/S0263-
Dwyer, C. (2007, January). Digital relationships
2373(02)00101-9
in the ‘MySpace’generation: Results from a quali-
tative study. Paper presented at the 40th HICSS, Gruen, T. W., Osmonbekov, T., & Czaplewski,
Waikoloa, Big Island, Hawaii. eMarketer (2007a). A. J. (2006). eWOM: The impact of customer-to-
Web 2.0 sites draw more visitors. Retrieved May customer online know-how exchange on customer
2, 2007, from http://www.eMarketer.com. value and loyalty. Journal of Business Research,
59, 449–456. doi:10.1016/j.jbusres.2005.10.004
eMarketer (2007b). UGC users outnumber cre-
ators. Retrieved July 2, 2007, from http://www. Gwinner, K. P., Gremmler, D. D., & Bitner, M.
eMarketer.com. J. (1998). Relational benefits in services: The
customer’s perspective. Journal of the Acad-
eMarketer (2007c). The rising roar of word-of-
emy of Marketing Science, 26(2), 101–114.
mouth. Retrieved June 29, 2007, from http://www.
doi:10.1177/0092070398262002
eMarketer.com.
Harteveldt, H., Johnson, C. A., Epps, S. R.,
eMarketer (2007d). Reviews boost e-commerce
& Tesch, B. (2006).Travelers embrace social
conversions. Retrieved May 25, 2007, from http://
computing technologies. Guidelines for travel
www.eMarketer.com.
e-commerce and marketing executives and manag-
Epps, S. R. (2007). Demystifying tagging for travel ers. Cambridge, MA, USA: Forrester Research.
sellers. Forrester Research Report.

1271
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

Haven, B. (2007). Making podcasts work for Morrison, A. (2002). Hospitality and tourism
your brand. Cambridge, MA, USA: Forrester marketing (3rd ed.). Albany, NY: Delmar.
Research.
Murphy, L., Moscardo, G., & Benckendorff, P.
Hwang, Y., Gretzel, U., Xiang, Z., & Fesenmaier, (2007). Exploring word-of-mouth influences on
D. (2006). Information search for travel decisions. travel decisions: friends and relatives vs. other
In D. Fesenmaier, H. Werthner & K. Wöber (Eds.), travelers. International Journal of Consumer
Destination recommendation systems: Behavioral Studies, 31(5), 517–527. doi:10.1111/j.1470-
foundations and applications (pp. 3-16). Cam- 6431.2007.00608.x
bridge, MA: CAB International.
Neus, A., & Scherf, P. (2005). Opening minds:
Jang, H. Y., Ko, I. S., & Koh, J. (2007, Janu- Cultural change with the introduction of open-
ary). The influence of online brand community source collaboration methods. IBM Systems
characteristics on community commitment and Journal, 44(2), 215–225.
brand loyalty. Paper presented at the 40th HICSS,
Olshavsky, R. W., & Granbois, D. H. (1979).
Waikoloa, Big Island, Hawaii.
Consumer decision making: Fact or Fiction?
Kim, W. G., Lee, C., & Hiemstra, S. J. (2004). The Journal of Consumer Research, 6, 93–100.
Effects of an online virtual community on cus- doi:10.1086/208753
tomer loyalty and travel product purchases. Tour-
Paolillo, J. C., & Penumarthy, S. (2007, January).
ism Management, 25(3), 343–355. doi:10.1016/
The social structure of tagging internet video on
S0261-5177(03)00142-0
del.icio.us. Paper presented at the 40th HICSS,
Lagrosen, S. (2005). Customer involvement Waikoloa, Big Island, Hawaii.
in NPD: A relationship marketing perspective.
Park, D. H., Lee, J., & Han, J. (2007). The ef-
European Journal of Innovation, 8(4), 424–436.
fect of online consumer reviews on consumer
doi:10.1108/14601060510627803
purchasing intention: The moderating role of
Lento, T., Welser, H. T., Gu, L., & Smith, M. involvement. International Journal of Elec-
(2006). The ties that blog: Relationship between tronic Commerce, 11(4), 125–148. doi:10.2753/
social ties and continued participation in blogs. JEC1086-4415110405
Workshop on Weblogging Edinburgh.
Pereira, C. S., & Soares, A. L. (2007). Improving
Li, C., & Stromberg, C. (2007). The ROI of blog- the quality of collaboration requirements for IM
ging. Cambridge, MA: Forrester Research. through social networks analysis. International
Journal of Information Management, 27, 86–103.
Lin, Y., & Huang, J. (2006). Internet blogs as a tour-
doi:10.1016/j.ijinfomgt.2006.10.003
ism marketing medium: A case study. Journal of
Business Research, 59, 1201–1205. doi:10.1016/j. Pitta, D., & Fowler, D. (2005). Online con-
jbusres.2005.11.005 sumer communities and their value to new
product developers. Journal of Product
Lin, Y., Su, H. Y., & Chien, S. (2006).
and Brand Management, 14(5), 283–291.
Knowledge-enabled procedure for customer
doi:10.1108/10610420510616313
relationship management. Industrial Marketing
Management, 35, 446–456. doi:10.1016/j.indmar- Rosenbloom, A. (2004). The blogosphere.
man.2005.04.002 Communications of the ACM, 47(12), 31–33.
doi:10.1145/1035134.1035161

1272
WEB 2.0, Social Marketing Strategies and Distribution Channels for City Destinations

Rowley, J., Teahan, B., & Leeming, E. (2007). Sigala, M. (2008, January). Developing and
Customer community and co-creation: A case implementing an eCRM 2.0 strategy: Usage and
study. Marketing Intelligence & Planning, 25(2), readiness of Greek tourism firms. ENTER 2008
136–146. doi:10.1108/02634500710737924 conference, Innsbruck, Austria.
Sigala, M. (2003). Developing and benchmarking Smith, D., Menon, S., & Sivakumar, K. (2005).
internet marketing strategies in the hotel sector Online peer and editorial recommendations,
in Greece. Journal of Hospitality & Tourism Re- trust, and choice in virtual markets. Journal of
search (Washington, D.C. Print), 27(4), 375–401. Interactive Marketing, 19(3), 15–37. doi:10.1002/
doi:10.1177/10963480030274001 dir.20041
Sigala, M. (2005a). In search of online post- Wang, Y., & Fesenmaier, D. R. (2004). Model-
modern authenticity: Assessing the quality of ing participation in an online travel community.
learning experiences at eternalegypt.org. In M. Journal of Travel Research, 42(3), 261–270.
SIGALA & D. Leslie (Eds.), International Cul- doi:10.1177/0047287503258824
tural Tourism: Management, implications and
Winer, D. (2005). What is a. River of News. style
cases (pp. 123–136). Oxford, UK: Butterworth
aggregator? Really Simple Syndication. Retrieved
Heinemann, Elsevier.
June 22, 2005, from http://www.reallysimplesyn-
Sigala, M. (2005b). Integrating customer relation- dication.com/riverOfNews
ship management in hotel operations: Managerial
Ying, Z., & Davis, J. (2007, January). Web com-
and operational implications. International Jour-
munities in blogspace. Paper presented at the 40th
nal of Hospitality Management, 24(3), 391–413.
HICSS, Waikoloa, Big Island, Hawaii.
doi:10.1016/j.ijhm.2004.08.008
Yuan, Y., Gretzel, U., & Fesenmaier, D. R. (2006).
Sigala, M. (2006). e-Customer relationship man-
The role of information technology use in Ameri-
agement in the hotel sector: Guests’ perceptions
can convention and visitors bureaus. Tourism
of perceived e-service quality levels. Tourism:
Management, 27(2), 326–341. doi:10.1016/j.
An International Interdisciplinary Journal, 54(4),
tourman.2004.12.001
333–344.

This work was previously published in Information Communication Technologies and City Marketing: Digital Opportunities
for Cities Around the World, edited by M. Gasco-Hernandez; T. Torres-Coronas, pp. 221-245, copyright 2009 by Information
Science Reference (an imprint of IGI Global).

1273
1274

Chapter 5.5
City Brands and their
Communication
through Web Sites:
Identification of Problems and
Proposals for Improvement
José Fernández-Cavia
Universitat Pompeu Fabra, Spain

Assumpció Huertas-Roig
Universitat Rovira i Virgili, Spain

ABSTRACT INTRODUCTION

City marketing tries to position cities in the mind In the current world of cities, competition has
of the public, although the process of creating and increased and the centre of interest has moved to
communicating city brands is still at an early stage include much broader spheres. Already, cities do
of its development. One of the main tools for the not try only to be just significant tourist nuclei, but
communication of these brands is now the World they also compete in aspects such as quality of life,
Wide Web. This chapter describes the results of two economic development and sustainability. Aside
combined studies (qualitative and quantitative) that from tourist interest, cities try to position themselves
analyzes a sample of official city Web sites. The as comfortable areas to live and important centers
results show that official Web sites of cities give of economic development that attract all types of
much attention to ease of navigation, but interactiv- investment.
ity is much less implemented, especially between To achieve this, it is necessary to know the
users. Furthermore, some lack of attention to the opinions and evaluations of the publics (Prebensen,
communication aspects of city brands can also be 2007)1, to find out what image they have of the
found. Finally, the chapter submits a number of city and determine the positioning that it would be
improvement proposals. desirable to achieve. Therefore, a vision of the city
must be formulated and, consequently, a program of
identity must be created that is transmitted through
a brand and a visual logo and, later, an adequate
DOI: 10.4018/978-1-60566-134-6.ch002 and effective communication program must be

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
City Brands and their Communication through Web Sites

run. With this objective, citymarketing tries to with the brand itself and the branding done by
position cities in the minds of the public. In spite the destination.
of this, the process of creating and disseminat- The city brand is a new concept and is not
ing city brands is still in a very early stage of its very well defined. It is currently very much in
development. fashion, and many people have theories, but few
This chapter tries to show, on one hand, that have dared to define it. It is a construct composed
the concept and application of city brands are of a name, a logo, some symbols and some values
still very incipient and, on the other hand, that the that we try to associate with a city, representing its
official websites, in part due to city brands being identity, with the objective of creating a position
underdeveloped, do not pay enough attention to and a vision of the city in the minds of the public.
the dissemination of the graphic, functional, and Each city must have its own brand, and each city
emotional aspects of the brand. With this, there brand must be the result of a citymarketing plan
is still a long way to go in the dissemination of and a competitive city strategy.
cities through their brands on the Internet. A very complete definition of destination
brand, fully applicable to the city brand, which is
based on the previous definitions of Aaker (1991)
THE CONCEPT OF CITY BRAND and Ritchie and Ritchie (1998) is that of Blain,
Levy and Brent Ritchie (2005), which implies:
The concept of brand applied to destinations,
places or cities is relatively new. It started to The creation of a name, symbol, logo, word mark
spread with the Travel and Tourism Research or other graphic that both identify and differenti-
Association’s Annual Conference in 1998 (Blain, ate a destination; that convey the promise of a
Levy & Brent Ritchie, 2005). From that moment, memorable travel experience that is uniquely
the concept has been developed widely and has associated with the destination; and that serve to
been studied from diverse perspectives, especially consolidate and reinforce the emotional connec-
from the point of view of tourism. All in all, tion between the visitor and the destination; that
studies on city brands and destinations are still reduce consumer search costs and perceived risk;
under developed and knowledge on the subject all with the intent purpose of creating a destina-
is limited. Some authors consider that it’s not tion image that positively influences consumer
correct to talk about branding or place branding destination choice. (p. 337)
in relation to territories, cities or countries. They
believe that it is incorrect to associate commu- However, the brand image is the result of the
nicational and marketing terms to realities with branding process, which is the perception created
their own identity like cities. Nevertheless, the in the minds of individuals. It is “networks of
majority or authors appreciate that the territories knowledge elements stored in long-term memory,
and the cities do not have the same characteristics and the core of such a network is the brand name
as commercial products, but agree that they can which is linked to a number of other knowledge ele-
apply the same marketing strategies to the ter- ments and/or associations” (Riezebos, 2003).
ritories (Olins, 2002). As confirmed by Bill Baker:2 “A destination
The first difficulty we come up against in the without a clear and attractive brand image is like
study of city brands is the confusion of concepts. a person without a personality. They blend into
Therefore, it is fundamental to distinguish between the crowd, are seen as uninteresting, and don’t
city brand and brand image (Cai, 2002). Many get the attention they deserve”.
studies confuse the analysis of the brand image As consequence of this dichotomy between

1275
City Brands and their Communication through Web Sites

city brand and brand image, there are two types seller’s good or service as distinct from those of
of studies on the topic. On one hand, those that other sellers”.
analyze the brand names themselves, their names, Blain, Levy and Brent Ritchie (2005) add three
symbols, logos, their identification, purposes and concrete purposes of city brands to the two clas-
meanings. Our research is included within this sics stated previously by Aaker (1991). On one
first type. And on the other, those that are based hand, to give the visitors the security of a quality
on the associations and relationships that the brand experience at the destination. On the other, to
names create with the public, that is, the brand reduce the search costs on the part of the visitors
image that is created among consumers. and, finally, to offer a single purchase proposal.
We also consider it is important to distinguish But we must be conscious of the difficulties of
another conceptual aspect. Place branding is not creating city brands, and the limitations that still
the same as destination branding. Place branding is exist with respect to their functions.
based on the construction of a global image of the Moreover, Hankinson (2004) uses the concept
territory that promotes the place in its globality: of brand networks, in which the destination and
economical, touristic and as a place of residence. city brands have four functions: brands as com-
Even in some occasions, tourist branding can be municators that represent a differentiation between
contradictory with place branding. This is why we cities, brands as perceptual entities that appeal to
consider it necessary to give a definition of place the senses and emotions, brands as values, and
branding. It is the sum of beliefs and impressions brands as relationships.
people hold about places. Images represent a Having reviewed the theoretical framework
simplification of a large number of associations of destination and city brands, and defined the
and pieces of information connected with a place. concept of city brand, from which the study starts,
They are a product of the mind trying to process we will now make a simple classification. As we
and pick out essential information from huge understand, city brands can be classified accord-
amounts of data about a place (Kotler, Haider & ing to their degree of evolution and development
Rein, 1993). in the following categories:
Having clarified these concepts, we will con-
centrate on the city brand, which is the purpose 1. The graphic brand, which only implies the
of our analysis. This, like all brands, has its raison creation of a symbol and a logo. Logos are
d’être, which is based on two basic functions the basic element for the creation of a brand
(Aaker, 1991). One of these is the identification and the main vehicle for communicating an
of the brand with the town and the attribution of image.3
a symbology and some values to the destination. 2. The functional conceptual brand. This type
The cities must have new signs of identity, an im- of brand adds the symbolization of some
age and a position. Therefore, the first function of of the territory’s characteristics to the logo,
the brand is to attribute functional and emotional which are real and tangible, and which are
values to a city that identify the different cities to be promoted, being adopted as strong
globally and by consensus. points of the city. These attributes may be:
The second function of the brand is based on good climate, beaches, nightlife, quality of
differentiating the cities from each other. This life and, level of innovation.
has always been the principal mission of all 3. The emotional conceptual brand, created by a
brands. According to the American Marketing body, entity or public institution, also trying
Association: “the brand is a name, term, design, to transmit abstract, symbolic and personifi-
symbol, or any other feature that identifies one able values to the city, such as innovation,

1276
City Brands and their Communication through Web Sites

multiculturality, modernization, passion, etc. adopted by all the public, starting with the city’s
With this combination of values, the aim is own residents, companies and institutions; and
to position and distinguish the image of the for this purpose it is fundamental that these are
city from the competition. Previous studies involved in its creation, that they adopt it as their
(Ekinci & Hosany, 2006; Hosany, Ekinci & own and thus help with its dissemination. A study
Uysal, 2006)4 have shown that the emotional by Blain, Levy and Brent Ritchie (2005), based
and personifiable values of the destination on interviews with the heads of marketing of
brands have positive influences in the prior the destinations, showed that the opinion of the
choices of purchase and recommendation residents and the visitors must be fundamental in
of these destinations. the process of creating city brands.
So, in light of the classification above, an exist-
Various authors agree that the brand image of ing city brand may be understood as more or less
a destination has two basic dimensions (Lawson developed according to its degree of preparation.
& Band-Bovy, 1977): cognitive and affective, Some city brands are simply logos that do not
which would correspond with the emotional represent any specific aspect of the destination,
and functional conceptual brand. The cognitive while others have elaborate brands agreed by
component would be made up of the beliefs and consensus, which come from a prior marketing
knowledge of the physical attributes of a city, plan and represent functional and emotional values
the functional conceptual brand; meanwhile, the that can be identified with the city.
affective component would refer to the feelings
about these attributes, the emotional conceptual
brand (Baloglu & McCleary, 1999). From the QUALITATIVE ANALYSIS
perspective of understanding the brand image OF CITY BRANDS
as a “cluster of attributes and associations that
the consumers connect to a brand”, Biel (1997) Having stated the definition, functions and clas-
understands the existence of “hard” associa- sification of city brands, we performed an initial
tions, which refer to the tangible and functional qualitative study on eight brands of important
attributes, and “soft” associations, emotional at- world cities on their official websites. The study
tributes. Biel acknowledges that the personality of showed the limitations of city brands in their
the brand belongs to the emotional aspect of the current process of creation, implementation and
brand image. Along the same lines, authors such dissemination.
as Etchner and Brent Ritchie (1991), Kapferer The objective of the investigation centered on
(1997) or De Chernatory and Dall’ Olmo Riley finding out the degree of evolution and develop-
(1997) confirm that the brand and its image are ment of a sample of city brands, and their dis-
composed of two attributes: the functional or semination through the websites of their official
tangible, and the symbolic or intangible5. institutions. The analysis consisted of two stages.
However, the emotional conceptual brand must The first centered on the prior examination of
be agreed upon by consensus, created jointly by the city brand itself and its degree of develop-
public and private institutions and citizens of the ment. The method used in this part of the study
region, which involves both the internal and ex- was based on the classification of the degree
ternal public, and is not identified with or property of advance preparation and evolution. In other
of a single institution, but of the whole region. words, in the analysis of the graphics and logo,
The valid significance of a brand is that reg- the functional and emotional values assigned
istered by its public. Certainly, a brand must be and the knowledge of the moment of creation,

1277
City Brands and their Communication through Web Sites

motivation and agents involved in the process of burgh, Amsterdam, Cincinnati, Toronto, Dubai
creating the city brand. and Hong Kong.
The second phase of the investigation centered
on the study of how the city brands were treated
on the official websites of the destinations. The RESULTS OF THE QUALITATIVE
analytical method used was the BIWAM (Brand STUDY: EVOLUTION AND
Identity Web Analysis Method). This is a technique LIMITATIONS OF CITY BRANDS
for qualitative analysis of the establishment of
brands on the web, created by Martín Barbero and The results of the study showed that city brands
Sandulli (2005), which includes eight dimensions are a concept of recent creation and still incipient
of analysis, of which we applied six to our study development. Actually, the majority of brands
of city brands: analyzed in the study were created in 2005. Many
large world cities still have not created their city
1. Analysis of the Appearance, which corre- brands. And among those that have, the degree
sponds to the strong, real and objective points of evolution of their brands is still mostly in the
of the destination (the functional element of initial stages. The vast majority are stuck in the
the brand) and how it is communicated on stage of creating logos based on strong points or
the web. characteristics of the city that they wish to boost.
2. Analysis of the Personality. This implies the However, very few cities try to identify themselves
assessment of the symbolic and emotional with an emotional conceptual brand, based on
elements that are attributed to city brands the appropriation of personifiable values and the
and their treatment on the websites. creation of a city marketing strategy.
3. Analysis of Humanity. This refers to the Our results coincide with those of a study by
interactivity of the page. ESADE (2004) on the evolution of the positioning
4. Analysis of the Style. This analyses the of Spanish tourist destinations and their tourism
graphic part of the brand, specifically, the brands. This stated that the brands, as strategic
relationship of the logo with the colors and realities of tourist destinations, are usually fairly
the typography of the website. general, based on functional values without dealing
5. Analysis of the Medium, and the commu- with emotional aspects. They do not segment their
nicative functionality of the websites. range much and are only transmitted externally,
6. Analysis of the Credibility. This refers to forgetting the internal public, and do not evolve
errors, slow loading speed, internal coher- with the passage of time.
ence, etc. Other studies in tourist marketing have shown
that, in general, the application of marketing tech-
We only applied six dimensions of analysis niques in destinations is still scarcely developed
created by Martín Barbero and Sandulli (2005) (Gnoth, 1998; Pritchard & Morgan, 2002).
because the other two could be applied to product The causes of the lack of evolution of city
brands, but no to city brands. brands are related to a series of limitations the
The sample consisted of eight city brands brands experience in their implementation. The
that correspond to international tourism capitals, first consists of the complexity of combining
which have created their city brands, but show a segmentation strategy with the creation of a
different degrees of evolution according to the single brand image. The cities are directed at
typology stated above. The city brands selected diverse sectors of the public (citizens, investors,
for the sample were: Barcelona, Madrid, Edin- businesspersons, tourists) with whom they wish

1278
City Brands and their Communication through Web Sites

to communicate. As each sector of the public or characteristics at the same time, perhaps to be
has certain interests and certain needs, the cities attractive to more sectors of the public, maybe
generally create different marketing strategies for because they are values that are currently very
each of them. For this reason, it may be compli- attractive in society and which they do not want
cated, or even contradictory, to create different to renounce although they have been adopted by
strategies and integrate them into one single brand other destinations. This is shown in the eight city
positioning. brands analyzed, Amsterdam and Toronto identify
Another difficulty is found in the existence of themselves as creative; Barcelona and Dubai,
more than one brand per city. It often occurs that adventurous; Barcelona and Edinburgh, friendly;
different institutions create city brands for their Edinburgh and Toronto, imaginative; Barcelona,
websites or for independent use. The result is Amsterdam and Edinburgh, diverse; Hong Kong,
dispersion, incoherence and the impossibility of Barcelona and Amsterdam, cosmopolitan; and
creating a single image that is recognizable and finally, Barcelona, Amsterdam, Edinburgh and
adopted throughout the community. For example, Toronto, innovative. In this sense, the city brand
the City Council of Madrid has a municipal web- completely loses its distinctive or differentiating
site (munimadrid) with a brand and a logo that function and this limits the creation of a single
has nothing to do with the Madrid brand on the image for every destination. Similarly, Morgan,
municipal tourism portal (esmadrid), or with the Pritchard and Piggott (2002) also showed that the
website of the region (turismomadrid). It is easy to images created of destinations and cities are not
understand that it is absolutely essential to make the different and do not usually contain a single idea
effort to coordinate and negotiate when creating a or single purchase proposition.
single brand that is not property of one institution Our research also revealed that the majority
in particular, but of all the citizens, and applicable of city brands have been created according to a
to all the websites related to a destination. specific event. For example, Madrid’s brand was
With respect to the second function of the created to promote the destination internationally
brand, that of differentiating cities from each other, when the city was selected as a possible candi-
it is possible that different cities try to identify date for the Olympic games of 2012. Barcelona,
themselves with the same values. The research however, created its brand to promote the Forum
that we have performed shows that the emotional 2004, and Cincinnati, after the results of an eco-
conceptual brand, that which attributes some nomic study, took notice of the need to connect
personifiable values to a city with the purpose the three States of Ohio, Kentucky and Indiana,
of differentiating it from other cities, is usually and as a result, created its brand with this unifying
created in a very broad and ambiguous way, and objective. This fact may also be negative for city
this does not fulfill its differentiating function. brands, which should never be associated with a
The majority of cities do not identify with a single specific political event, as they must represent a
value, but with many, some of which are shared city and not a municipal government, a social or
by different brands, which, in fact, encourages sporting event or any private interest.
confusion. The motive that generally moves public bodies
Kotler (1993) already stated that all images of to create city brands is mainly based on tourist
a destination must be simple and distinctive. The or economic interests, and thus the brand created
main function of a brand must be to differentiate usually is identified with these interests. This is
it with respect to the competition. However, in an error, a reductionist conception that limits the
the study it was demonstrated that the majority of potential of the city brand. In addition, on occa-
cities prefer to be identified with diverse values sions, after an election or change of party in the

1279
City Brands and their Communication through Web Sites

town halls, the city brands and their representation only as tourist destinations, but also as business
and meanings are changed, precisely with the in- centers and residential areas, portals or broader
tention of breaking away from the previous image websites are starting to be created, which offer,
and promoting a new improved image related to in addition to tourist information, business and
the political party that has formed the municipal leisure information for the citizens.
government. These changes create dysfunction, In this sense, and from the field of communi-
as the brands need time to be implemented and cation, the brand websites have been marked as
require their evolution to be homogenous and the future of marketing communication on the
coherent. The persistence and durability of a brand Internet, as they have the potential to provide
is key for its implementation and acceptance by high levels of information and, in addition, create
all sectors of the public. Therefore, brand changes virtual product experiences (Klein, 2003). Brand
only create more confusion in the identities and websites are capable of combining both of the basic
images of the cities. objectives of commercial communication in this
Finally, but importantly, difficulty in the cre- channel: to create a brand image and achieve a
ation of city brands lies in coordination, taking into direct response (Hollis, 2005). As Cho and Cheon
account the umbrella brands of destinations greater (2005) describe, the websites may serve for diverse
than the cities. In tourism, the broader destinations communication purposes: public relations, sales
(regions, nations, states or countries) include those promotion, advertising or direct marketing.
within them in their brand for the promotion of In the second part of the study, the results of
tourism. So this presents us with a number of our analysis of city brands on the web showed
questions: Should the umbrella brands take into that the aspects that make up the corporate image,
account and be coherent with the city brands that that is, the colors, the lines and the logo of the
they include? Should they all be related? Should city brand, in general, are used very little to create
they be coherent with each other? Should they graphic coherence and brand image throughout the
have common features? Should the attributes of website. There are some exceptions, such as the
the umbrella brands be shared by the brands of sites of Amsterdam or Madrid, that show effective
the respective cities that are represented? graphic coherence for transmitting the brand, but
habitually the typography and the colors of the
logo are only used in auxiliary hyperlinked pages,
TREATMENT OF CITY but not the whole official site.
BRANDS ON THE WEB The study showed that what is best transmitted
through the web is the functional conceptual brand
The Internet and new information technologies of the cities, the strong points or the potential that
play a key role in communicating the cities and they wish to promote; but in no way the emotional
their brands. They are an important source of conceptual brand, which ascribes personifiable
information. Destination Management Systems values to the destination. The exception is a hy-
are more than simple websites. In addition to the perlink that some official websites have, such as
information, they offer advertising, marketing and Edinburgh or Amsterdam, which links to a page
sales applications, and have interactive resources exclusively dedicated to explaining the emotional
that, in an entertaining way, provide services and brand.
attract the attention of the users. Starting with the results of the studies of
Currently, through a city’s tourism website, Hosany, Ekinci and Uysal (2006), where it is
you can get information, make reservations, demonstrated that the emotional and personifiable
etc. However, in the promotion of the cities, not values have positive influences on the intention

1280
City Brands and their Communication through Web Sites

to visit, purchase and recommend destinations basic iconography and ease of use, such as that
on the part of consumers, it is surprising that it is of Barcelona, would transmit a good city brand
actually the emotional aspect and the personality image. However, extensive and complicated sites
of the brand that are the least developed aspects that are slow to open and that have a confused
on the official websites of the cities analyzed. The internal structure, such as that of Toronto, cause
heads of marketing of the cities should develop less positive or even unfavorable attitudes to be
strategies and campaigns that promote the dis- created.
tinctive personality of the destinations, based on Along the same lines, Jared M. Spool (1996)6
the emotional components of these cities, which did a study comparing websites, which demon-
create better positioning and a more favorable strated that the usability of a website considerably
image among users. and positively affects the brand and the branding
Thus, in conclusion, the under use of websites process. His results showed that the users that
to promote city brands has been verified. The navigate more easily through a website and find
websites centre on the functional conceptual the information that they want quickly end up with
aspects of the brand, that is, all the strong points a better impression of the brand, as it has satisfied
that are notable in the city, but lack, in general, their expectations to a greater degree. Contrarily,
coherent graphic treatment and the expression of the obstacles that the users find when navigating
the emotional conceptual brand. The design of the negatively and directly affect their perception
websites tries to be useful and functional to provide of the brand. Therefore, usability is essential for
the users with the information and services they effective branding.
wish to obtain, but they are not at all creative to Regarding interactivity, Liu (2003) defends the
disseminate the emotional values attributable to the idea that the concept of interactivity unites three
city through its brand. McMillan (2004) coincides correlated but different factors: the active control
with our statements, arguing that advertising on of information, bidirectional communication and
the Internet and websites must be more creative, the synchronicity or simultaneity of communica-
that is, better designed, with greater impact, more tion. In a previous article (Liu & Shrum, 2002), the
varied and more entertaining. same author classified the brand websites in the
Once at this point, and in light of other studies maximum range of the three factors stated. This
and a bibliography centered on more technical study showed, by the way of bidirectional commu-
questions of the websites, we decided to analyze nication, that the Internet is the only medium that
other characteristic aspects of the websites that can be used for commercial transactions without
also influence the dissemination and perception the help of other tools, since necessary activities
of city brands. We considered that dealing with such as showing the product, placing orders,
city brands should not be limited to an analysis making payments or even, in categories such as
of the websites’ content only, but should take music, software or transport titles, distributing the
into account aspects such as the interactivity or product can happen through the web.
usability of their pages. Other authors (Cho & Cheon, 2005) prefer
Different research on websites shows that us- to divide the concept of interactivity into three
ability is a key aspect in the creation of a good fields of action or types: consumer-message
brand image. The sites that seem to be or are interactivity, consumer-consumer interactivity
easier to open, navigate or use, create a more and consumer-marketer interactivity. Consumer-
favorable attitude and image among users (Chen message interactivity refers to the capability of
& Wells, 1999; Chen, Gillenson & Sherrell, 2002; the user to personalize his or her relationship
Heijden, 2003). Thus, small websites, with very with the contents of the page according to his or

1281
City Brands and their Communication through Web Sites

her interests and motives. Consumer-marketer For this empirical study, we used a quanta-
interactivity centers on the communication be- tive method based on a WTO (World Tourism
tween the user of a website and the organizers or Organization, 1999) analysis model. This was
those responsible for the content; this relationship extended with the contribution of recent studies
may be bidirectional, from user to administrator (McMillan, 2003; Liu, 2003; Cho & Cheon, 2005)
(questions, suggestions, complaints) or from and adding newly created interactive resources,
administrator to user (obtaining personal data, which appeared as new features on destination
answering questions, etc.). Consumer-consumer websites at the time the study was carried out.
interactivity is the relationship that may be created At the same time, the model was also extended
between the people that access a website (virtual with aspects to analyze about how city brands
communities, chats, forums, etc.). are dealt with on the web, the databases that the
More recent studies (Sicilia, Ruiz & Munuera, websites may obtain about their users, and other
2005; Ko, Cho & Roberts, 2005) show that inter- aspects of website information not considered in
activity enables the information to be processed the initial model.
better and generates more favorable attitudes The analysis was applied to 40 official websites
towards the website and towards the product and of important tourist cities of the five continents
the brand, and greater intention to purchase. during 2006. This sample was selected by means
Based on these previous studies on the usability of a ranking evaluating the main tourist cities of
and interactivity of websites we decided to make the world, taking into account the World’s Top
a broader quantitative analysis about the treatment Tourism Destinations of the WTO (World Tour-
of city brands on the Internet. In addition to the ism Organization), the number of visitors and the
items related to graphic, functional and emotional importance of the city as a tourist destination.
aspects of the brand, in the trial we analyzed We observed 135 indicators, of which 87 are
usability and interactivity characteristics of the representative of the three variables analyzed:
websites as elements that also influence the dis- usability, interactivity and the brand. Interactivity
semination of a good brand image. is analyzed in the three factors mentioned previ-
ously: consumer-message interactivity, consumer-
marketer interactivity and consumer-consumer
PURPOSE AND METHOD OF interactivity. Its analysis, through the SPSS
QUANTITATIVE STUDY. USABILITY, program, was centered on descriptive statistics
INTERACTIVITY AND THE and the combination of variables, using Gamma
CITY BRAND ON THE WEB as a correlation index.

The first qualitative study was wider. It analyzed RESULTS OF THE QUANTATIVE
the city brands and their treatment in websites in STUDY. USABILITY
depth. But later, we decided to carry out a quan-
titative study analyzing more webs and variables. The concept of usability is defined as “the extent
We had to develop a different questionnaire of to which a product can be used by specified us-
analysis that could be measured by quantitative ers to achieve specified goals with effectiveness,
methods. efficiency and satisfaction in a specified context
The main objective of this quantitative study of use”7. This ease of use, nevertheless, is related
consisted in analyzing the degree of usability, to very diverse aspects, that go from the page
interactivity and treatment of city brands on the design, to the content quality, the ease of locating
official websites of tourist cities. information and the simplicity of navigation, all

1282
City Brands and their Communication through Web Sites

Table 1. Items of usability (Source: Authors’ Table 2. Ranking of websites with the most and
compilation) least points with respect to usability (Source:
Authors’ compilation)
Indexes of usability Frequency Percentage
List of contents on all pages 36 90.0% Web sites with high scores Web sites with low scores
in structure, design and in structure, design and
Link the home page 37 92.5%
usability usability
External links related sites 38 95.0%
Score City Web sites Score City Web sites
Up-to-date information 39 97.5%
11 London 5 Cairo
Sitemap 23 57.5%
11 Hong Kong 5 Beijing
Indication of the navigation
16 40.0% 10 Bangkok 6 Lisbon
path
10 Budapest
10 Sydney
10 Amsterdam
of which are related, in addition, to the subjective
10 Tokyo
perceptions of the user.
The indicators used in the study to measure 10 Buenos Aires

the usability variable are those that provide the 10 Mexico City

navigability for the website: the access menu for


the sections always being visible, the indication
of the navigation path, constant links to the home With respect to the help that the websites offer
page, the existence of a sitemap and an internal their users, 90% of the sites analyzed provided
search engine; the possibility of user help by e-mail contact. In addition, 65% provided a tele-
telephone, e-mail, chat or web call; and external phone helpline. But, however, only one website
links to related sites. offered help by chat and none used web calls (calls
The majority of sites analyzed in the study from marketing staff from the destinations where
showed high levels of structure, which means you request them to call you at a certain time) or
that the websites are, in general, well designed. human clicks (communication in real time of a
At a global level, high percentages were seen for visitor to a website with its administrators).
the variables that make up the structure, usability All in all, the results stated up to now show
and design of the websites. that the majority of websites are well structured
As seen in Table 1, the majority of the sites and designed. Therefore, they have a high degree
analyzed have a list of the site contents that appears of usability.
on all of the pages, along with a link to the home Table 2 shows the rankings of the websites with
page from each page. These two resources give best structure, design and usability, along with
the user his or her location on the web, clarify the those that have the least points in these aspects.
structure of the site and simplify its use. In addition, The structure variable ranges between 5 and 11,
almost all the websites have external links with with sites with 5 having the worst structure and
other related sites. Through these links they can those with 11 the best.
provide complementary information and services. The initial Hong Kong web page (Figure 1),
In some cases connection to the official sites of the one of the best in usability of our study, is simple,
town halls and institutions are provided, in others, graphic and structured. It shows only a big im-
the purchase of services, for example the link on age and the list of different languages the user
Barcelona’s website to ServiCaixa, through which can choose.
you can purchase tickets for entertainment. When you enter in the initial page of the selected

1283
City Brands and their Communication through Web Sites

Figure 1. Hong Kong Web site

language you can observe that it has the content relationships with the marketing organizers of
list in all the pages, visual and graphic links and the cities. They offer the possibility of questions
shows a constant simplicity in the entire site. and complaints from users, opinion surveys, chats
with promotional or sales agents, bulletin boards
Interactivity for users, or the possibility of placing orders and
making reservations on line. Finally, consumer-
The interactivity variable, as explained previously, consumer interactivity is measured using indica-
was analyzed using three aspects: consumer- tors of the relationship between them, such as the
message interactivity, consumer-marketer inter- existence of chats or an email service.
activity and consumer-consumer interactivity.
We measured the concept of interactivity using Consumer-Message Interactivity
the three aforementioned typologies of Cho and
Cheon (2005). The first result we can see is that, in general, the
Consumer-message interactivity is measured majority of websites analyzed use many more
with indicators such as: the existence of search resources that belong to the consumer-message
engines, user help, the option to customize the interaction, than to the consumer-marketer, and
display, displaying virtual reality, multimedia consumer-consumer interaction. Therefore, the
presentations, directional maps, virtual leaflets, sites offer the navigator a greater interactivity
the option of downloads to mobiles, and on-line with the messages that they wish to transmit than
games, among others. However, consumer- with the people that are in charge of marketing the
marketer interactivity is based on indicators of destinations and other consumers. And within the

1284
City Brands and their Communication through Web Sites

resources of the consumer-message interaction, possibility of placing orders or online reservations


the most used in all the websites, that is, those that is offered by 35% of the sites, and 32.5% allow
show the highest percentages, are the interactive the user to track the orders placed. For these two
travel organizers. resources, the prior registration is usually required
So, this shows that the majority of websites of the user, and with this the websites can obtain
analyzed offer services of interactive travel or- personal information about their consumers.
ganizers. We refer to organizers without the pos- In 22.5% of the sites, users can fill in surveys
sibility of purchase. They help users to plan their or opinion polls and 12.5% offer the option of
trips by providing fully personalized information complaints by consumers.
adapted to the needs and interests of the users. The rest of the consumer-marketer interac-
The interactive travel organizers allow the users tive resources are barely used by the websites
to plan their trips in a fully complete and personal- analyzed. For example, only Rome’s site offers
ized way, from their transport to the destination to the users the possibility of asking what they want
their accommodation and other complementary about the city, the services offered or the entity
tourist services. that organizes the marketing of the destinations.
However, there are other interactive consumer- Only the websites of Dublin and Istanbul have an
message resources that are hardly used by the electronic bulletin board available to the users.
websites analyzed. Certainly, 62.5% have direc- And finally, the sites of Madrid, Amsterdam and
tional maps of the cities to orient the users and Montreal are the only ones that allow the consum-
82.5% have geographical markers on the maps ers to sponsor the website. This means that any
with zoom, which offer the possibility of searching organization can pay some money to sponsor the
for and locating elements on the map in detail. But website. All the conditions are very well explained
only three of the sites allow the user to custom- on the site. In exchange, the sponsor can put its
ize the home page, four show virtual excursions logo on the website.
or online games for children, five offer virtual Not one of the websites analyzed offered the
flights of the city from the air or allow users to users the possibility to propose new products or
create their own virtual leaflets in folders, which services, chats with the marketing agents, or “call
they can save with the personalized information me” buttons with time and language selection for
that interests them. those responsible for promoting the city to call the
The informative services using optional down- interested users to provide them with the informa-
loads to mobiles, whether metro maps, informa- tion they want, personally by telephone.
tion on monuments or audio downloads are also
resources that are barely used by the websites Consumer-Consumer Interactivity
analyzed.
All of these interactive resources, in addition to With respect to the consumer-consumer interac-
offering information, provide a certain entertain- tion, there are even fewer resources available. The
ment and distraction to the users of the websites, resource of this type that is most used in the sites
making them more attractive and interesting. analyzed is the on-line postcards service, featur-
ing in 40% of the websites, of which 37.5% do
Consumer-Marketer Interactivity not require registration. This resource, which is
offered more and more by the destination websites
The resources of the consumer-marketer interac- and is mostly used by young users, enables them
tion are still less used by the websites analyzed to send on-line postcards, without the delay or
that those of consumer-message interaction. The costs of sending them.

1285
City Brands and their Communication through Web Sites

Table 3. Resources used in the websites of interactive travel organizers (Source: Authors’ compila-
tion)

Frequency Percentage
Interactive travel organizers (without pos- How to get there 33 82.5
sibility of purchase)
What to do 39 97.5
Attractions/events 38 95.0
Leisure activities 38 95.0
Cultural activities 37 92.5
Where to stay 38 95.0
Transport 39 97.5
Excursions 33 82.5
Rentals 16 40.0

Moreover, only Krakow’s site has a chat for Spanish Mediterranean coast. In it, he highlights
users, and the possibility for consumers to tell the absence of instruments that promote interactiv-
their stories, experiences and summaries of trips is ity in the websites, such as on-line forms, pages
only found on the sites of Hong Kong, Rome and of visitor comments, etc. (Table 3)
Toronto. Finally, none of the websites analyzed The degree of global interactivity of the web-
offer the services of a cyber club of users with sites analyzed, which may range between 4 and
advantages or a cyber community with common 28, can be seen in the ranking in Table 4. These
interests. Cyber clubs which offer some advantages numbers represent the number of interactive tools
of information, discounts and special conditions they use.
to the loyal users. The cyber club of users would Visit Dublin’s web page it’s the most interactive
allow the marketing organizers of the destina- out of all the analyzed sites. If we only observe
tions to create loyalty programs for clients using the first part of the page we can find a searcher
the offer of advantages and discounts. Moreover, and in the graphic frontal there is the option to
the creation of a cyber community would enable click and see a video about the city. The whole
the users of this group to build relationships with page is very interactive.
each other, broaden information in their interest, In the accommodation section the user can
and create a strong position with respect to these search information, but can also reserve and buy
common interests. the products.
So, we can see that there is a great inequality Even the maps of the city and the maps of public
in the use of interactive resources by official web- transport are interactive. They show the routes that
sites of destinations. The most used are those that the user requires in personalized way.
belong to the consumer-message interaction, and
especially, the interactive travel organizers. All in Treatment of City Brands on the Web
all, generally, the percentages of use of interactive
resources are still underused, in particular those of In the analysis of the communication and dis-
the consumer-marketer and consumer-consumer semination of city brands through websites, it
interaction. These results coincide with those is necessary to differentiate the concepts of city
of the study by Anton (2004) about the Internet brand and brand image. As previously explained,
presence of the main tourist destinations of the the city brand is a construct composed of a name,

1286
City Brands and their Communication through Web Sites

Table 4. Ranking of websites with the most and Table 5. Ranking of the best and worst treatment
least points with respect to interactivity (Source: of city brands in websites (Source: Authors’
Authors’ compilation) compilation)

Most interactive websites Least interactive websites Websites that deal with their Websites that deal with city
city brands best brands worst
Score City websites Score City websites
Score City websites Score City websites
28 Dublin 4 Moscow
10 Amsterdam 3 Kuala Lumpur
23 Hong Kong 4 Kiev
8 Hong Kong 5 Madrid
21 Valencia 7 Sydney
8 Athens 5 Zagreb
20 Madrid 8 Cairo
8 Dublin 5 Cairo
18 Berlin 8 Seville
8 Toronto 5 Sydney
18 Istanbul
5 Moscow
18 Rome
5 Lisbon
5 Montreal
5 Rome
a logo, some symbols and values that we try to
5 Mexico City
associate with a city representing its identity, with
5 Kiev
the objective of creating a positioning and a vision
of the city in the minds of the public, which must
be based on a citymarketing plan and projected
through a communication program. On the other must also be taken into account that three of the
hand, the brand image is that which is created in websites analyzed show more than one logo for
the minds of the public as a consequence of the the same city brand, which creates dysfunctional-
communication of the city brand, in conjunction ity and incoherence.
with the perceptions and subjective values of Of the websites, 92.5% offer a brief descrip-
individuals. (Table 5, Figure 2, Figure 3, Figure tion of the destination, but only 5% (two of the
4, Figure 5, Figure 6) sites analyzed) have a minimum explanation of
In the study, the degree of communication and the city brand, of its creation and symbolism.
dissemination of the city brands through the web- Similarly, only 35% of the websites state the
sites was measured using the presence of the brand marketing objectives of the tourist institutions
and the logo, whether there is a description of the of the cities.
city brand, whether the graphic and photographic The graphic images, in general, also are an
images represent the brand, whether the colors underused resource in the dissemination of the city
and the typography of the page are coordinated brands. All the websites transmit the functional
with the logo, and the functional and emotional brand through photographic images, but only half
brand is disseminated. of them (50%) communicate the emotional brand
The results show that city brands are treated through the photographs. On the other hand, the
poorly on the websites analyzed. Their treatment rest of the graphic images are not used to dis-
is even less developed than that of interactivity. seminate the brand at all.
All in all, the vast majority of websites (97.5%) Finally, the graphic and typographic coherence
have the logo on all of the pages, and this is usu- between the brand and the web has not been taken
ally situated in the upper left part. However, it into account either in the majority of websites

1287
City Brands and their Communication through Web Sites

Figure 2. Visit Dublin Web site

Figure 3. Visit Dublin accomodation section

Figure 4. Visit Dublin maps section

1288
City Brands and their Communication through Web Sites

Figure 5. I Amsterdam Web site

analyzed. Only 27.5% of the sites have the pre- The web of Amsterdam, which best dissemi-
dominant colors of the website coordinated with nates the city brand, does not only deal with the
the logo, and 23.5% use the same typography on functional and emotional brand on the web, but
the page and the brand. On the other hand, only in addition it pays attention to many other graphic
one of the sites analyzed includes elements of and visual aspects. For example, the emotional
the advertising campaign transmitted through the brand, which is disseminated through very few
conventional media. websites, in that of Amsterdam it is dealt with both
The degree of global treatment of the city by changing photographs on the page, which show
brands on the websites analyzed, which ranges people of the city, and by an introductory page that
between 3 and 10, can be seen in the ranking in explains in depth what the brand symbolizes. In it,
Table 5. it explains what the brand “Iamsterdam” is, who

Figure 6. I Amsterdam manifesto page

1289
City Brands and their Communication through Web Sites

created it, when and why; and what it symbolizes of the websites may be higher than now without
and means. As can be seen in the home page, the affecting the other variables at all.
logotype is present in the entire site. The brand is As a consequence of the under use of interac-
based on the idea that the value of Amsterdam is tive resources, the possibilities of user records,
their people. It plays with ‘I am sterdam’ and ‘I contained in the interactive resources, also de-
am Amsterdam’. The photographs of the site try crease. This means that the information obtained
to communicate these emotional values through on the part of the website users is minimum. As
different types of people of differing ages, race has been previously confirmed, the possibility
and profession instead of showing monuments or of obtaining more information for more direct
landscapes and scenery. one-to-one marketing that would produce special
But the official site of Amsterdam also pays offers according to the needs of the consumers is
attention to other aspects related to the brand, such being wasted.
as the typography. It uses the same combination of On the other hand, there is even less treatment
colors and type of letters as the logo in the whole of the brands on the websites analyzed. Only the
website, creating a graphical coherence that boosts graphic images, basically the logo, are dissemi-
and constantly reminds of the city brand. nated and in part the functional brand. However,
Finally, this web page contains information on very few communicate the emotional brand.
the city marketing, their objectives and strategies Similarly, few websites show graphic and typo-
and also an explicit explanation of the brand (which logical coherence with the logo, or a relationship
you can read in the I Amsterdam manifesto), with the advertising or the marketing objectives
advertising of the city and the merchandising of of the destinations.
products. These results show that, in general, there is
a certain lack of attention to the communicative
aspects in the construction of the city websites.
CONCLUSION They seem to be made by information technolo-
gists, who pay great attention to usability, and
The study indicates that the majority of websites also, to a lesser extent, interactivity, but do not
analyzed have high usability indexes. Therefore, take into account the communication of the
the official websites of cities pay a lot of atten- brand. Therefore, you can sense the existence of
tion to ease of navigation, so that the user can a certain lack of co-ordination between the IT and
easily navigate and find the information he or communication and marketing departments of the
she wants. entities promoting the cities.
Contrarily, interactivity is much less imple- To effectively create websites that disseminate
mented in the websites analyzed. The interactive city brands and promote the image of the destina-
resources that are most used are those that cor- tions, it is fundamental to be conscious of what
respond to the consumer-message or consumer- an important tool for image, communication and
marketer relationship, despite being generally marketing these websites are for the destinations
underused. and cities.
So, it can be confirmed that the websites ana- As we go further into the Information Era, the
lyzed in this study have a better structure, design role of websites is changing. They are evolving
and usability, than interactivity. The websites are from being merely sources of information, that
more usable and structured than interactive. And, is, intermediaries between tourists and destina-
in addition, there is no correlation between these tions, to being involved in tourist transactions.
two variables. Therefore, the interactive capacity Therefore, tourism and Internet make an ideal

1290
City Brands and their Communication through Web Sites

combination. The Internet provides the users cities. And only after this first step, will usability
with a way of obtaining much more varied and and interactivity be complementary and effective
detailed information on the destinations and the aspects in the promotion of city brands.
cities than that which existed before. However,
in addition it enables them to, through the same Improvement Proposals
web space, make consultations and reservations
quickly and easily. The creation of the city brand must not be an act
The websites are considered the future of restricted to the activity of citymarketing. Exactly
communication on the Internet, as they offer a the opposite, it must be coherent with the whole
large quantity of information and, in addition, marketing plan and be derived from a competitive
create virtual product experiences (Klein 2003). city strategy.
Websites, on one hand, create a brand image It must start with a prior diagnosis of the cur-
and, on the other, can provoke a direct response rent situation of the city image and the competi-
(Hollis, 2005). As Cho & Cheon (2005) confirm, tion, to later formulate the vision of the city and
websites offer different communication possibili- the positioning that is desirable to achieve in the
ties: public relations, sales promotion, advertis- world urban system.
ing, direct marketing and brand image creation. Only after the two first stages should an identity
As a consequence, this communicative potential program for the city be determined, which must
must be promoted and made use of in the field take into account a historical analysis, and the
of cities. perceptions of the internal and external public.
With respect to how little city brands are dealt Then the visual identity of the city, its logo, colors,
with on the websites analyzed, it is due, firstly, typography should be determined; along with the
to insufficient conceptualization, creation and functional and emotional values of the brand. In
development. As we have already stated, city this whole process it is essential to identify the
brands are a recently created concept that is still internal and external public (citizens, visitors,
at a very incipient stage. For this reason, the web investors, mass media, public institutions, neigh-
deal poorly with them. Firstly, it is necessary to borhood associations), to address them directly
create elaborate city brands, based on a strategy and achieve that they take on the identity and the
of citymarketing, to be communicated later to city brand to be promoted as their own and, if
the public and, if not, disseminated through the possible, this should be by consensus.
websites. Finally, the city brand they must be dissemi-
Despite the studies mentioned above (Spool, nated by a communication plan to each sector of
1996; Chen & Wells, 1999; Chen & al., 2002; Hei- the public through all types of actions above and
jden, 2003) that show the importance of usability below the line, with the intention of continuity
in promoting a more positive brand image among in time, for the values of the brand to catch on
users; and the effect of interactivity, which pro- coherently in the perceptions of all their sectors
duces more favorable attitudes towards the web, of the public and be integrated in the rest of the
the brand and greater intentions of purchase (Sicilia citymarketing actions.
& al., 2005; Ko & al., 2005), we can confirm that But city brands must not only comply with a
high indexes of usability and moderate indexes creation process inserted into the citymarketing
of interactivity do little good in the websites actions, they must also have certain character-
analyzed in the promotion of city brands. First it istics:
is necessary to correctly create the city brand, to
later disseminate it through the websites of the

1291
City Brands and their Communication through Web Sites

1. The city brand must be made up of three and typographic coherence throughout the
basic elements: the graphic brand, the cre- website.
ation of a symbol and a logo; the functional 2. Maximum development of usability through-
brand, based on the real strong and attractive out the page, which facilitates user navigation
points of the city; and the emotional brand, and promotes a positive image increasing
the symbolic and personalizable values the possibilities of recommendation and
that are associated with the city. Without marketing.
one of these three elements, the city brand 3. Maximum creative use of the interactive
is incomplete, losing its identificative and resources, as these improve the brand image
persuasive power. and the users feel drawn to navigate.
2. A single city brand must be created, with
a single strategy, that is applicable to the
diverse sectors of the public, but that at the
same time identifies the brand with unique References
values, which allow the cities to de identi-
fied and differentiated from each other. In Aaker, D. (1991). Managing brand equity. New
addition, this brand must be adopted by all York, NY: Free Press.
the organizations and entities, avoiding the Anton Claver, S. (2004). La presencia en In-
proliferation of several brands that create ternet de los principales destinos turísticos del
dysfunctionality. litoral mediterráneo español. Actas del Congreso
3. Finally, and independently of the municipal TURITEC 2004. Universidad de Málaga, 2004.
political changes, city brands must be created Retrieved June 1, 2008, from http://www.turismo.
to last a long time. If they are agreed upon uma.es/turitec/turitec2004/index.htm
by consensus when created, and therefore,
do not belong to the municipal governments, Baker, B. (2007). Places: The new brand frontier.
they will evolve independently of the politi- Total destination management. Retrieved June 1,
cal channels and have a greater possibility 2008, from www.DestinationBranding.com
of being consolidated in the minds of the
Baloglu, S., & McCleary, K. (1999). A model of
public.
destination image formation. Annals of Tourism
Research, 26, 868–897. doi:10.1016/S0160-
Once the city brand is completely and ad-
7383(99)00030-4
equately created, and starting with the importance
of websites as tools for communication, promo- Biel, A. (1997). Discovering brand magic: The
tion and marketing of the cities, attention must hardness of the softer side of branding. Interna-
be paid to three key aspects in the promotion of tional Journal of Advertising, 16, 199–210.
city brands through the Internet:
Blain, C., Levy, S. E., & Brent Ritchie, J. R.
(2005). Destination branding: Insights and prac-
1. The treatment of the city brand through the
tices from destination management organiza-
constant and unique presence of the logo,
tions. Journal of Travel Research, 43, 328–338.
with an explanation of its symbology and
doi:10.1177/0047287505274646
objectives, with the representation of the
functional and emotional values that are Buhalis, D. (2002). eTourism. Information tech-
desirable to associate both textually and as nology for strategic tourism management. UK:
graphically, and by means of the graphic Prentice Hall.

1292
City Brands and their Communication through Web Sites

Buhalis, D., & Costa, C. (2006). Tourism, manage- Etchner, C. M., & Brent Ritchie, F. (1991). The
ment dynamics. Trends, management and tools. meaning and measurement of destination image.
Oxford: Elsevier Butterworth Heinemann. Journal of Tourism Studies, 2(2), 2–12.
Cai, L. (2002). Cooperative branding for rural Gnoth, J. (1998). Conference reports: Branding
destination. Annals of Tourism Research, 29, tourism destinations. Annals of Tourism Research,
720–742. doi:10.1016/S0160-7383(01)00080-9 25, 750–760.
Chaves, N. (2004, April). La marca destino turís- Hankinson, G. (2004). The brand images of tourism
tico, cinco estrategias gráficas. Paper presented destinations: A study of the saliency of organic im-
at the XIII Simposi Internacional de Turisme ages. Journal of Product and Brand Management,
i Lleure ESADE-Fira de Barcelona. Retrieved 13(1), 6–14. doi:10.1108/10610420410523803
April 30, 2007, from www.esade.es/cedit2004/
Hankinson, G. (2005). Destination brand im-
cat/est_estudios.php
ages: A business tourism perspective. Jour-
Chen, L. D., Gillenson, M. L., & Sherrell, D. L. nal of Services Marketing, 19(1), 24–32.
(2002). Exciting online consumers: an extended doi:10.1108/08876040510579361
technology acceptance perspective. Information
Heijden, H. (2003). Factors influencing the usage
& Management, 39, 705–719. doi:10.1016/S0378-
of web sites: The case of a generic portal in The
7206(01)00127-6
Netherlands. Information & Management, 40,
Chen, L. D., & Wells, W. D. (1999). Attitude 541–549. doi:10.1016/S0378-7206(02)00079-4
toward the site. Journal of Advertising Research,
Hollis, N. (2005, June). Ten years of learning on
39(5), 27–38.
how online advertising builds brands. Journal
Cho, C.-H., & Cheon, H. J. (2005, Summer). of Advertising Research, 255–268. doi:10.1017/
Cross-cultural comparisons of interactivity on S0021849905050270
corporate Web sites. Journal of Advertising,
Hosany, S., Ekinci, Y., & Uysal, M. (2006).
43(2), 99–115.
Destination image and destination personality:
De Chernatony, L., & Dall’Olmo Riley, F. (1997). An application of branding theories to tourism
Modelling the components of the brand. European places. Journal of Business Research, 59, 638–642.
Journal of Marketing, 32(11/12), 1074–1090. doi:10.1016/j.jbusres.2006.01.001
doi:10.1108/03090569810243721
Kapferer, J. N. (1997). Strategic brand manage-
Ekinci, Y., & Hosany, S. (2006). Destination per- ment. London: Kogan Page.
sonality: An application of brand personality to
Keller, K. (1993). Conceptualizing, measuring, and
tourism destinations. Journal of Travel Research,
managing customer-based brand equity. Journal
45, 127–139. doi:10.1177/0047287506291603
of Marketing, 57, 1–22. doi:10.2307/1252054
ESADE, & BDDO Consulting (2004, April). Evo-
Klein, L. R. (2003). Creating virtual product
lución del posicionamiento de los destinos turísti-
experiences: The role of telepresence. Journal of
cos españoles: De lugares de vacación a marcas
Interactive Marketing, 17(1), 41–55. doi:10.1002/
de turismo. Paper presented at the XIII Simposi
dir.10046
Internacional de Turisme i Lleure ESADE-Fira de
Barcelona. Retrieved April 30, 2007, from www.
esade.es/cedit2004/cat/est_estudios.php

1293
City Brands and their Communication through Web Sites

Ko, H., Cho, C.-H., & Roberts, M. S. (2005). McMillan, S., Hwang, J.-S., & Lee, G. (2003,
Internet uses and gratifications. A structural equa- December). Effects of structural and perceptual
tion model of interactive advertising. Journal of factors on attitudes toward the website. Journal
Advertising, 34(2), 57–70. of Advertising Research, 400–409.
Konecnik, M., & Gartner, W. C. (2006). Customer- Morgan, N., Pritchard, A., & Piggott, R. (2002).
based brand equity for a destination. Annals of New Zeland, 100% Pure: The creation of a pow-
Tourism Research, 34(2), 400–421. doi:10.1016/j. erful niche destination brand. Journal of Brand
annals.2006.10.005 Management, 9(4/5), 335–354. doi:10.1057/
palgrave.bm.2540082
Kotler, P., Haider, D. H., & Rein, I. (1993). Mar-
keting places: Attracting investment, industry and Morgan, N., Pritchard, A., & Pride, R. (2004).
tourism to cities, states and nations. New York, Destination branding. Creating the unique des-
NY: Free Press. tination proposition. Oxford: Elsevier.
Lawson, F., & Band-Bovy, M. (1977). Tourism Nielsen, J. (2002). Usabilidad. Diseño de sitios
and recreational development. London: Archi- web. Madrid: Prentice Hall.
tectural Press.
Padín Fabeiro, C. (2004, April). La formación de
Liu, Y. (2003, June). Developing a scale to measure la imagen de un nuevo destino. Paper presented
the interactivity of Web sites. Journal of Advertis- at the XIII Simposi Internacional de Turisme i
ing Research, 207–216. Lleure ESADE-Fira de Barcelona. Paper presented
at the XIII Simposi Internacional de Turisme
Liu, Y., & Shrum, L. J. (2002). What is interactivity
i Lleure ESADE-Fira de Barcelona. Retrieved
and is it always such a good thing? Implications
April 30, 2007, from www.esade.es/cedit2004/
of definition, person and situation for the influ-
cat/est_estudios.php
ence of interactivity on advertising effectiveness.
Journal of Advertising, 31(4), 53–64. Park, C. W., Jaworski, B. J., & MacInnis, D. J.
(1986, October). Strategic brand concept man-
López Lita, R., & Benlloch, M. T. (2005). La
agement . Journal of Marketing, 50, 135–145.
marca territorio. El marketing de ciudad, una
doi:10.2307/1251291
herramienta al sevicio de las marcas territorio.
99% com, 2, 8. Prebensen, N. K. (2007). Exploring tourists’
images of a distant destination. Tourism Man-
Lynch, P. J., & Horton, S. (2004). Manual de estilo
agement, 28, 747–756. doi:10.1016/j.tour-
web. Principios de diseño básico para la creación
man.2006.05.005
de sitios Web. Barcelona: Gustavo Gili, SA.
Riezebos, R. (2003). Brand management. A
Martín Barbero, S., & Sandulli, F. (2005). Mar-
theoretical and practical approach. Harlow, UK:
cating en la Web: BIWAM. Identidad desnuda.
Prentice-Hall.
Madrid: Dossat 2000.
Ritchie, B., & Ritchie, R. (1998). The branding
McMillan, S. (2004). Internet advertising: One
of tourism destination: Past achievements and
face or many? In D. Schumann and E. Thorson
future trends. Paper presented at the 48th Congress,
(Eds.), Internet advertising: Theory and re-
AIEST, St-Gall.
search (2nd edition). New York, NY: Lawrence
Erlbaum.

1294
City Brands and their Communication through Web Sites

Sicilia, M., Ruiz, S., & Munuera, J. L. (2005). 3


Blain, Levy and Brent Ritchie (2005) con-
Effects of interactivity in a Web site. Journal of firm that the logos of city brands globally
Advertising, 34(3), 31–45. represent the experience that the visitors
expect of a destination or city. Effectively,
Simon, S. J. (2001). The Impact of culture and
logos stimulate the communication of the
gender on web sites: An empirical study . The
attributes desired among the visitors and
Data Base for Advances in Information Systems,
influence the tourists’ decision to visit a
3(1), 18–37.
place.
Spool, J. M. (1996, January 1). Branding and 4
Hosany, Ekinci and Uysal (2006) demon-
usability. User Interface Engineering. Retrieved strated that there is a relationship between
June 1, 2008, from www.uie.com/articles/brand- the image of a destination and its personal-
ing_usability. ity. They also demonstrated that the image
of a destination’s brand is fundamental to
Tsikriktsis, N. (2002). Does culture influence web
position it effectively.
site quality expectations? An empirical study. 5
Other authors, such as Keller (1993) and
Journal of Service Research, 5(2), 101–112.
Park (1986) add a third category or element
doi:10.1177/109467002237490
to the city brand: experiential attributes,
World Tourism Organisation. (1999). Promoción which refer to the experiences, satisfaction
de destinos turísticos en el Ciberespacio. Madrid: and feelings of the destination’s visitors.
WTO. 6
Spool compared two websites: one more
usable and informative and the other very
World Tourism Organisation. (2001). Comercio
graphically meticulous. He showed that
electrónico y turismo. Madrid: WTO.
the more usable site created a better brand
image, as it satisfied the expectations of
the users to a greater degree. Therefore, the
Endnotes graphic aspects of a website, such as logos
and photographs, have less effect on the
1
According to Nina K. Prebensen (2007), to branding than expected.
build a good city brand, with the most suit- 7
International Organization for Standardiza-
able elements, that create a positive image, tion (ISO). In: http://www.iso.org/iso/home.
the opinion and knowledge of the visitors htm
or future visitors is fundamental.
2
Bill Baker is the founder and President
of Total Destination Management. He is
internationally recognised as an expert in
building brand recognition for destinations
and communities. TDM is a Portland, Oregon
based team of destination branding, tourism
planning and marketing specialists. www.
DestinationBranding.com consulted in July
2007.

1295
City Brands and their Communication through Web Sites

Appendix: List of Websites Analysed

America
Table 6. America

México D.F. www.mexicocity.gob.mx


Los Ángeles www.lacvb.com
Montreal www.tourisme-montreal.org
New York www.nycvisit.com/home/index.cfm
Toronto www.torontotourism.com
Buenos Aires www.buenosaires.gov.ar/areas/turismo/home
Río de Janeiro www.riodejaneiro-turismo.com.br/en/home.php

Africa
Table 7. Africa

El Cairo www.cairotourist.com
Marrakesh www.ilove-marrakesh.com

Asia
Table 8. Asia

Bei-jing english.bjta.gov.cn
Tokyo www.tourism.metro.tokyo.jp/english/index.html
Hong Kong www.discoverhongkong.com
Kuala Lumpur www.kualalumpur.gov.my
Bangkok www.bangkoktourist.com
Macau www.macautourism.gov.mo
Bali www.balitourismauthority.net/home.asp
Delhi delhitourism.nic.in

Oceania
Table 9. Oceania

Sydney www.cityofsydney.nsw.gov.au

Europe
Table 10. Europe

Amsterdam www.iamsterdam.nl
Athens www.cityofathens.gr
Dublin www.visitdublin.com

1296
City Brands and their Communication through Web Sites

Istanbul english.istanbul.com
Florence www.firenzeturismo.it/en_default.asp
Helsinki www.hel2.fi/tourism
Lisbon www.cm-lisboa.pt/turismo
London www.visitlondon.com
Paris www.parisinfo.com
Prague www.visitprague.cz
Roma www.romaturismo.com
Vienne info.wien.at
Berlin www.berlin-tourist-information.de
Kiev www.kmv.gov.ua
Krakow www.krakow.pl
Budapest www.budapestinfo.hu
Moscow www.moscowcity.com
Zagreb www.zagreb-touristinfo.hr

Spain
Table 11. Spain

Barcelona www.bcn.es/turisme/catala/turisme/welcome.htm
Madrid www.esmadrid.com
Sevilla www.turismo.sevilla.org
València www.turisvalencia.es

This work was previously published in Information Communication Technologies and City Marketing: Digital Opportunities
for Cities Around the World, edited by M. Gasco-Hernandez; T. Torres-Coronas, pp. 26-49, copyright 2009 by Information
Science Reference (an imprint of IGI Global).

1297
1298

Chapter 5.6
Assessing the Performance of
Airline Web Sites:
The ARTFLY Case

Elad Harison
University of Groningen, The Netherlands

Albert Boonstra
University of Groningen, The Netherlands

Executive Summary airlines in the industry that deals with the recent
challenges of the intensifying competition, is open
This case takes place in the increasingly competi- for a wide variety of interpretations and should
tive environment of the airline sector.Airline web- be determined due to the firm’s nature of opera-
sites and Internet-based booking systems enable tions and due to its aim to increase the volume
transformation of airline operations and become of its online sales.
strategic weapons for the majority of airlines
worldwide. Established airlines are attempting
to stimulate customers to use the Internet, in re- Organizational Background
sponse to entry of low cost carriers that capture
shares of their market and in order to reduce their ARTFLY, founded in 1919, is a major airline that
sales costs. Nonetheless, the development and operates on a worldwide basis. In 2004 ARTFLY
maintenance of websites and e-commerce plat- merged with the Air Minoli group and became
forms requires substantial investments in capital a division within the joint Air Minoli/ARTFLY
and labor. Therefore, airlines need to assess the group. Air Minoli/ARTFLY has a worldwide
performance of their e-commerce channels in coverage, in terms of destinations, flight routes
terms of profitability, customer appreciation and and marketing units. The group uses two main
volume of sales on a continuous basis. However, European airports as their main hubs and offers air
the design of an assessment model that can serve transport to 128 destinations in 65 countries in 5
the managers of ARTFLY, one of the established continents. To illustrate the volume of the group’s

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Assessing the Performance of Airline Web Sites

activities in 2004-2005 the ARTFLY group trans- by direct sales and using the Internet as a promi-
ported more than 20 million passengers and more nent marketing channel were immediate actions
than 600,000 tons of cargo, generating profits of that were taken by ARTFLY’s management and
255 million Euros ($300 million U.S.). assisted in surviving the intense competition in
In 1995, ARTFLY was one of the first European the market and maintaining its major position in
airlines that introduced fully integrated e-business this rapidly changing business environment.
into its ticketing process, in response to radical In March 1996, the company launched its
changes in the air transport market (Rubin & Joy, first Web site, which was mainly an electronic
2005). In particular, the entry of low cost airlines brochure with information on flights. A year later,
and the price war that followed presented a tangible new functionality was added including real-time
threat to ARTFLY’s market. Low cost carriers information on departure and arrival times, a
succeeded in attracting increasing numbers of reservations module and electronic ordering of
passengers who previously purchased their tickets tickets. From 2001, a complete electronic booking
through travel agents and airline branches and have system was included in ARTFLY’s homepage and,
shifted to booking their flights online via the Web since then, the Web site is continuously maintained
sites of those carriers (see Figure 1). For example, and often face-lifted due to changes in the market
the rise of EasyJet and Ryanair to dominance in and new technological possibilities.
the British market, flying 55 million passengers ARTFLY bases its marketing and sales ac-
from and to the U.K., was achieved in part as a tivities on four main distribution channels that
consequence of online ticket sales (ABTA, 2005). include its own Web site, online travel agencies
ARTFLY replaced its previous procedures with (e.g., Expedia.com and Kayak.com), ARTFLY
online bookings, electronic tickets, and electronic branches and “physical” travel agencies (see Table
check-ins and boarding passes. Reducing its costs 1). The firm has defined its Web site, ARTFLY.

Figure 1. Channels of low cost flight bookings (Source: ABTA, 2005)

100%

90%

80%

70% 64%

60%

50%

40%

30%
21%
20% 17%

10% 5%
3%
0%
Online and By telephone Travel agent Other No response
Internet

1299
Assessing the Performance of Airline Web Sites

com, as the most important channel in its home involve either higher operating expenses (e.g.,
market. The company is keen to increase the share personnel and location costs) or commissions
of online bookings via the Web site (together with paid to independent travel agencies.
other online travel agencies) from 25% up to 40%. The targets of ARTFLY can be compared to
Outside its home market, ARTFLY branches are the industry’s benchmarks. In 2006, the Airline IT
its main sell points. The branches are located Trends Survey (SITA, 2006) measured the relative
at major airports and in the centers of large cit- share of bookings in each distribution channel in
ies worldwide, hence enabling accessibility of 98 major airlines worldwide. It concludes that, on
customers to its booking services. In addition, average, the airlines’ own Web sites (the direct
ARTFLY operates through local travel agencies. online channel) captured 21.5% of the bookings,
However, the firm aims at replacing part of its 8.2% of the tickets were purchased through on-
offline sales by increasing the share of online line travel agencies (the indirect online channel),
bookings, as the physical distribution channels 17.7% of the bookings were made in the branches

Table 1. Four distribution channels of ARTFLY

Direct
Indirect

Online channel
1 www.ARTFLY.com 2 Online travel agencies
Now (2005/06): 12%
(e.g., Expedia.com)
Target (2008/09): 40%
Offline channel 4 Travel agencies with branches
Now (2005/06): 88% 3 ARTFLY branches (e.g., Thomas Cook, but mostly operate on lo-
Target (2008/09): 60% cal, regional or national scales)

Figure 2. Use of online channels for flight bookings (Source: SITA, 2006)

100%
91%
90%
80%
70%
60%
50%
40%
30%
19%
20% 15%
11% 11% 10%
10% 5%

0%
s
te

s
es
te

es
ie

le
sit
si

si
nc

sit
sit

sa
eb

eb
e

eb
eb
ge

in
w

e
w

lin
ir l

w
w
la
n

e
ta
w

on
nc

er
n
ve

io
O

in

th
llia
tra

se
ct
Jo

O
Au
ta

tu
e
in

in

no
nl

Jo
O

Do

1300
Assessing the Performance of Airline Web Sites

of flight companies (the direct offline channel), were approximately 156 million Euros (source:
while the majority of bookings – 52.6% of the Reference Document ARTFLY/Air Minoli, 2006).
flight tickets sold in 2006 – were made through However, the exponential increase in the use of
“physical” travel agencies (the indirect traditional ARTFLY’s Web site led to larger volumes of online
channel). On the supply side, 91% of the firms in traffic and to demands from customers for new
the survey offer their customers the possibility and advanced features, availability and reliability
to book their flights via their own Web sites, and of the Web site.1 Consequently, ARTFLY’s online
only a small proportion of the firms (5%) do not operations required a new assessment model that
have any direct or indirect online channels (see captures the multiple dimensions of its online
Figure 2). operations and their effectiveness in terms of ad-
The firm’s policy, that is, setting a 40% share ditional profits and growth of its customer base.
of sales from online sources, stems from three Currently, the evaluation of ARTFLY’s online
main objectives, as follows: activities and performance is coordinated by the
CEO through a series of monthly meetings that
• Cost-reductions that result from simplify- involve managers of different divisions and units.
ing the booking and payment processes and The managers report the changes in the indicators
reducing the scale of human interventions in for which they are responsible and recommend
the ticketing process by emailing e-tickets on adapting existing online tools and campaigns
to customers. or highlight the needs to develop new systems.
• Better services are provided by a faster Then, the feasibility of developing or modify-
check-in procedure and by enabling online ing new tools, their expected contribution to the
customers to select their seats while book- organization and their costs are discussed within
ing their flight. Doing so, customers do not this forum. However, this process suffers from
have to wait for their boarding pass and can two major shortcomings: First, the configuration
shorten their stay at the airport. of the current performance evaluation process
• Increased market share is achieved by requires significant coordination efforts due to
gaining better performance on the Internet the involvement of different divisions and orga-
than its competitors. This aim involves both nizational units and due to the broad distribution
high degrees of Internet presence and use of of responsibilities among them (see Figure 3).
intelligent strategies and customer relation- Second, the CEO can assess the performance of
ship management (CRM) tools to identify ARTFLY’s Web sites only in a monthly resolution
the preferences and patterns of purchasing and cannot receive continuous measurements of
decisions among online customers. performance to assess it within shorter periods (for
example, on a daily or on a weekly timeline).
Over time, ARTFLY’s Web site has become
increasingly important for the firm in terms of
its growing volume of sales and as an attractor of Setting the Stage
potential customers. To illustrate the growth of its
online channel of distribution, in 1997 ARTFLY’s Recent studies in marketing and in ICT empha-
Web site received approximately 3,000 visitors sized the potential of Web sites and e-commerce
per week, while during 2007 it attracted more tools as new distribution channels that can create
than one million weekly visitors. In 2001, the site new opportunities for cost savings and profits for
reached average monthly revenues of 0.5 million new and existing companies (Buhalis, 2004; Xing
Euros, while the monthly online revenues in 2007 & Grandt, 2006). In particular, early air transport

1301
Assessing the Performance of Airline Web Sites

studies identified how Internet technologies can be via search engine) and via traditional marketing
implemented by airlines to increase the volumes of campaigns and attracting potential customers
their online ticket sales and to expand their mar- that enter it to purchase goods and services
keting channels (Alamdari, 2002; Jarach, 2002; (Chu, 2001). The accessibility to the Web site
Lubbe, 2007; Shon, Chen & Chang, 2003; Yoon, and the knowledge of consumers of its existence
Yoon & Yang, 2006). Recent studies focussed on and the range of information and services that it
assessing the impact of commercial Web sites on provides are major factors in its success, as well
broader dimensions of activities and performance as the choice of advertising channels with high
of firms (see, for example, Agarwal & Venkatesh, potential to attract customers (Monsuwe, Dellaert
2002; Gianforte, 2004; Huizingh, 2002; Otim & & Ruyter, 2004). It is also important to enhance
Grover, 2006; Park & Gretzel, 2007). their presence and the volume of Web traffic to
Those studies seem to be of a major importance them by other means, such as association with
particularly to airline companies, as their market related search terms and Web-based campaigns.
is characterized on the one hand by strong com- Complementary off-line campaigns can also
petition and on the other hand by opportunities reinforce the exposure of Web sites to potential
to sustain customer relations and loyalty (e.g., Web-surfers and customers. Drèze and Zufryden
via frequent flyer programs, Gasson, 2003). In (2004) describe nine categories of traffic building
addition, the entry of low-cost airline companies strategies that should be taken into account while
in recent years intensified price competition and constructing Web-traffic plans: off-line advertis-
put higher pressure on the profit margins of many ing of a Web site, offline news reports, Internet
firms (Klein & Loebecke, 2003; Klein, Klein, advertising/banner ads, links from other Web
Kohne & Oorni, 2004). In this respect, the Internet sites, links from search sites, links from online
assists airlines in developing alternative market- directories, online news reports, reference in an
ing and distribution channels, reducing direct e-mail and discussion/newsgroups. The exposure
costs, reaching new customers and maintaining of Web sites increases not only but including it in
long-term relations with existing customers. search engines and search-word based advertis-
However, the strong competition in the market ing, but also by more “traditional” means, such
and the openness of the Internet as a commu- as banners, TV campaigns and press releases.
nications medium enabled firms to learn and to Attracting large numbers of visitors to the firm’s
apply similar online business practices, Web site Web site is an important step to the successful of
designs and e-commerce tools. In other cases, the Web-based exposure and merchandizing.
rapid development of Internet technologies has Although financial indicators of the opera-
made within a short period Web sites less user tions of the firm (including its online marketing,
friendly or less attractive in comparison to Web distribution and sales)2 are essential for evaluat-
sites of competitors, and sometimes even obsolete ing the performance of Web sites, a different
(Law & Leung, 2000). Consequently, marketing approach that is based on a broader framework
managers in airline firms, as well as in other sorts of the firm’s operations and competitive position
of organizations that base a large part of their is often needed (Toh & Raven, 2003). Financial
activities on electronic commerce, are in need of measurements of the performance of Web sites
an evaluation tool to assess the effectiveness and and online tools mostly reflect short-term measure-
the reach of their online operations and to indicate ments, such as costs and revenues and overlook
what improvements in it should be made. other performance measures of online activities
Internet marketing and e-commerce are largely and long-term effects on the firm.
based on site building, promoting it online (e.g.,

1302
Assessing the Performance of Airline Web Sites

Figure 3. ARTFLY’s organizational structure (Divisions and units participating in the evaluation of the
performance of ARTFLY’s Web sites are marked in grey)

Analyses of Internet ventures, including those Case Description


of non-Internet product and service providers,
should be done at the industry level. When an in- By the end of 2006 the Finance Director, Mrs.
dividual tool fails to meet the industry’s standards, Eva van Duinen, started a discussion during a
the firm is advised not to engage in developing weekly board meeting, in which she expressed
online marketing and distribution channels. For her doubts about the performance of the Web
example, if an airline company invests an online site of ARTFLY. She felt that the usual growth of
booking system and the volume of bookings are online sales were stagnating. The CEO, Mr. Jim
inferior to other players in the market, it should Manzetti, agreed with her that it would be wise
either modify the system or abolish it and turn to to give more attention to this theme. The first
its traditional distribution channels (e.g., agents thing to do for a more thorough discussion in the
and call centers, Smyth & Wagner, 2006). management team would be to collect all relevant
data on Web site performance. The members of

1303
Assessing the Performance of Airline Web Sites

Table 2. Comparative measures of ARTFLY’s Web site use in subsequent periods


Measures 2005 Q4 2006 Q4

1 Financial overview:
Revenues from online bookings (in USD) 88,359 84,563
Number of tickets sold via Web site 209,523 206,251
Costs of Web site maintenance (in thousands of USD) 12,450 14,429

2 Customer appreciation of Web site according panel


Ease of finding the Web site (on scale of 1 – 10) 7.3 7.3
Overall appreciation of Web site (on scale of 1 – 10) 6.9 6.5

3 Contact
Number of visitors 5.133.935 4.988.238

the management team agreed that they would look reported from a panel discussion with customers
for this information, especially the information that the overall appreciation of the Web site by the
that compares Web site performance between the panelists was also gone down a little bit. These
4th quarter of 2005 and 2006. panel discussions were held with a group of twenty
One week later, the Web site performance customers on a quarterly basis and covered many
theme was again on the agenda. The Finance direc- issues related to customer services and customer
tor reported that the revenues from online book- satisfaction of ARTFLY.
ings had gone down with nearly 5%, compared The data on Web site performance were
with the last quarter of 2005. She also mentioned grouped by a staff member in Table 2. Based on
that the costs of Web site development and main- these data, a discussion started among the various
tenance had gone up by 12%. John Simons, IT directors. Eva van Duinen, the Finance Director,
director, explained this rise of IT costs by growing felt strongly that a program of cost cutting would
costs of maintenance as a result of the increasing be necessary to align investments in online chan-
number of Web pages and the hiring of external nels with the performance of those channels.
experts. He said that more Web pages and more Director Otto Caroll from Sales argued that the
online traffic automatically lead to higher costs figures reflect a growing and intense competition,
in terms of maintenance. External experts were especially with low-cost airlines, which directly
needed in order to make Web sites more aestheti- compete with ARTFLY in its existing routes, and
cally pleasing. Developing such expertise in-house that a substantial share of ARTFLY’s customers
would be much more expensive. have preferred to book their flights online via
Mr. Otto Caroll, director of the sales depart- their Web sites. He argued that ARTFLY should
ment delivered additional figures that showed that know what they are doing wrong with their Web
the number of tickets sold is also slightly decreas- sites, before cutting costs.
ing. In addition to this, other indicators revealed The IT director contributed to the discussion
that the volume of use of the Web sites by new by saying that a strategy of promoting the online
and returning (“loyal”) customers is lower than channel should be supported by high investments
in the previous years and its popularity among in quality and attractiveness of the Web sites. He
them is decreasing. During the same meeting, the thinks that cost saving programs will jeopardize
public relations manager, Mrs. Anna Jacobsson, such a policy.

1304
Assessing the Performance of Airline Web Sites

All directors agreed that the available infor- also be related to the strategy of ARTFLY, which
mation as reflected in table 1 was not appropriate includes a careful move to online channels, while
for making far reaching decisions on investment maintaining off-line channels and keeping cus-
programs or savings on ARTFLY Web sites. The tomers satisfied who prefer off-line channels.
participants were missing comparisons with the
performance of other airlines as well as figures
about the performance of off-line channels. Others In Need of an Assessment
were asking what the determinants of customer Model: Defining Key
satisfaction of airline Web sites are. Speed? At- Indicators to Assess the
tractiveness? Completeness? It would not be wise Performance of ARTFLY’S
to invest or to save costs without having a thorough Web Site
knowledge on important relevant issues that de-
termine the performance of online channels. Recent studies in marketing and in ICT have ad-
When the discussion continued, it became dressed the potential of Web sites and e-commerce
clear the directors did not completely agree on tools to provide new distribution channels that
the most relevant issues that determine Web site can increase the profitability of flight companies.
performance. The financial director emphasized Online bookings can significantly increase the
the financial justification of investments in Web volume of sales by reaching new customers via
sites. Others agreed that this is important, but Internet presence and campaigns and can reduce
the financial performance will be determined the operation costs of firms by eliminating the
by actual sales. These are determined by issues need to pay commissions to travel agents (Xing
like the easiness to find a Web site, the speed and & Grandt, 2006). Major airlines were aware of
loading time of a Web site, attractiveness and so the need for an assessment model to evaluate the
on. The IT director, John Simons added to this performance of their online activities (notably,
by saying that ARTFLY also pays considerable their Web sites and online ticket sales) and have
amounts of money to search engines like Google adopted various models for this purpose (Alam-
and Yahoo. Costs of clicks from those sites varied dari, 2002; Jarach, 2002; Lubbe, 2007; Shon et
from 0.20 – 0.30 cents. To his opinion, it would al., 2003; Yoon et al., 2006). However, most of
be important to assess whether these costs can these models address only limited aspects of the
be justified by increased online sales. Marketing online activities and Web site operation, such
manager, George Veldman emphasized that Web as the number of entries per period or the user-
sites also realize important cross sales in terms of friendliness of the booking system, and do not
hotel bookings, car rentals and so on. He felt that provide a coherent overview of the indicators
the profits from cross selling should be captured that define in detail the role of the Web sites. For
in the whole decision making process. example, most of these assessment methods over-
The CEO concluded this discussion by saying look the additional added value of the Web sites
that the company needed a Web site performance in terms of their impact on sales, contribution to
monitoring system that includes the issues raised the airlines’ profitability, reputation and presence
by the various directors. Such a system should in the market. A second type of indicators that is
have to lead to frequent reports that show how often omitted from the assessment models that are
the company's Web sites are performing. This described in the literature evaluates the Web sites
would help to make rational decisions with regard in terms of the quality of their technical operation.
to investments and cost saving programs much For example, the proportion of down time and the
easier. He emphasized that the system should number of malfunctions can be included in the

1305
Assessing the Performance of Airline Web Sites

model as indicators for technical performance. • Customer satisfaction: the ability of cus-
The assessment of the technical performance of tomers to locate the Web site of the company,
Web sites is particularly important as the avail- to access it with ease and to use its contents
ability of their online booking systems strongly in ways that meet their needs. This indicator
affects the volume of online sales and the level is measured through customer surveys and
of customer satisfaction. On the other hand, the focus groups.
period in which a Web site cannot be accessed • Accessibility: the ease of finding a Web site
can be directly translated to loss of sales and and the exposure of potential customers to
potential bookings. Even when some of the cus- it. This criterion includes also the technical
tomers decide to complete their order via another aspects of accessing the Web sites with dif-
marketing channel, such as call centers or travel ferent Web browsers, presenting the contents
agencies, it would be less profitable for the firm properly and downloading the Web site in
in comparison to online bookings. a reasonable time when various Internet
The impact of commercial Web sites on the connections are used.
performance of the firm can be assessed through • Traceability: measures the traffic to the
a broad prism that includes both financial key Web site, access to the different contents
performance indicators as well as via qualitative and Web pages within it, its presence in
and quantitative measures. Those measures should online portals and its position (rating) in
reflect the satisfaction of clients from the airline’s search engines. The criterion is measured
Web site and the accessibility to its contents and by online traffic measurement tools.
to the booking system. • Contact: a sales-based criterion that reflects
The financial appraisal of the firm’s opera- the interest of customers in the online ser-
tions is the direct and most important indicator vices of the firm. It measures the volume of
to evaluate its e-commerce channel. Therefore, online visitors that access the Web site in a
online operations, and particularly the construc- given time and the distribution between new
tion and the maintenance of a Web site, can be and returning visitors. The number of visitors
assessed from the financial standpoint by means is especially important as it is used also to
used to evaluate other IT investment decisions: assess the traceability of the Web site and
Profitability is the most important objective and it the accessibility of consumers to it. Contact
reflects the proportion between profits and invest- data are obtained from customers that visit
ments. In addition, the risks associated with the Web sites after viewing advertisements in
development and operation of a Web site and an the media (newspapers, radio and TV), fol-
online booking system should be taken into the low Internet advertisements (banners flash
account of the expected financial gains (Renkema videos and pop-up windows) and adsense
& Berghout, 1997). However, the financial mea- advertisements that link terms used in search
surements present the costs and revenues and do engines to commercial Web sites3. Therefore,
not include other measures that can affect the it is possible to provide a measure for the
performance of the firm, such as service qual- effectiveness of each advertising channel by
ity, reputation and brand loyalty. Therefore, the tracking the volume of traffic that originates
financial measures should be complemented by from it.
qualitative and quantitative indicators. • Sales: indicators measure the revenues and
In addition to those indicators, nonfinancial the profits from online sales in a given pe-
factors that affect the performance of airline riod. The firm can compare between sales
Web sites can be assessed via the following from “off-line” marketing channels, such as
categories:

1306
Assessing the Performance of Airline Web Sites

travel agents and ARTFLY’s offices, and the • After-sales: records data on service quality,
volume of orders that are completed via its the volumes and costs of online transactions
Web site. Changes in the relative share of (in comparison to off-line sales) and the
online bookings reflect changes in consum- satisfaction of customers from their online
ers’ behavior and trust in purchasing services bookings.
via ARTFLY’s Web site and can signal the • Customer support: measures the total num-
need for a proactive online marketing, such ber of visitors that apply for support services,
as Internet-based campaigns and additional such as technical assistance and follow-up on
Web advertising. Sales data by channel can online bookings, in a given period, as well
be obtained from ARTFLY’s financial and as assessing the costs of customer support
accounting systems. services regarding the operation of the Web
site and its various features.

Table 3. Criteria and measures for assessing ARTFLY’s online products and services
Criteria Indicators
Costs
Financial overview Revenues
Direct revenues from bookings
Ease of finding the Web site
Ease of booking a flight online
Speed of downloading the Web site and contents
Customer satisfaction
Speed of operating the Web site
Overall appreciation of Web site
Appreciation of services
Traceability Number of domain names in use
Loading time
Accessibility
Number of languages supported by Web site
Total:
Number of visitors
Number of new visitors

Contact Per channel:


Number of visitors
Costs per visitor
Sales per visitor
Average bookings per visitor
Total revenues
Total sales
Sales per product
Profits per product
Sales Total profits
Volume of bookings
Total conversion
Conversion per product
Ratio of online to offline revenues
Costs per booking
Number of bookings
After-sales
Use in relation to sales
Customer appreciation
Use of the product
Use of the product per visitor
Customer support
Costs of use
Customer appreciation

1307
Assessing the Performance of Airline Web Sites

Since the beginning of 2007, the model, The metrics collected for the assessment model
presented in Table 3, was applied in various provide useful insights that can assist in managing
ways to measure the short-term and long-term the firms’ online strategy and activities. After the
performance of ARTFLY’s Web site. The vari- implementation of the assessment model, ART-
ous indicators have formed a baseline that can FLY’s managers can identify major changes and
provide information on changes in consumers’ market trends with ease and to respond to them.
behaviour (such as growing demand for particular The model is often used to assess current practices
lines or seasonal trends). Further, by following and strategies of the firm, such as its advertising
the indicators of the model over time ARTFLY strategy and the distribution of its advertising
can monitor its online operations and assess budget between online and traditional media.
the effectiveness of new contents and features,
Internet-based promotions, different advertising
channels and online marketing campaigns. Table Current
4 demonstrates the measures of the assessment Challenges/Problems Facing
model after implementing it for two years. How- the Organization
ever, the application of ARTFLY’s assessment
model is only in its initial stages and the firm As mentioned previously, the airline market is
plans to expand its use (such as assessing the use dynamic and it is dominated by intensifying
of the booking system by regular passengers vs. competition, particularly from low-cost carriers.
members of the frequent flier program) and to In this business environment, ARTFLY aims at
broaden the set of measurement indicators. increasing the share of the online bookings ac-

Table 4. Comparative measures of the assessment model in subsequent periods.


Measures 2006 Q4 2007 Q1

105,489 108,549
1 Financial overview: 14,759 15,563
Revenues (in thousands of Euros) 89,467 91,263
Costs (in thousands of Euros)
Direct revenues from online bookings
2 Customer satisfaction 7.1 7.2
Ease of finding the Web site 6.6 7.1
Speed 6.3 6.7
Overall appreciation of Web site 365 389
Appreciation of services
3 Traceability
Number of domain names
6.123669 6.23657
4 Accessibility
Loading time in seconds
Languages supported by Web site
8.36 7.92
5 Contact 11 14
Number of visitors
Visitors from Google
Sales to visitors from Google (millions of Euros) 4.563.236 5.239.246
Costs per click from Google (Euros) 562.346 599.756
Visitors from Yahoo 6.235 6.539
Sales to visitors from Yahoo (millions of Euros) 0.21 0.23
Costs per click from Yahoo (Euros) 123.756 142.153
953 843
0.30 0.29

1308
Assessing the Performance of Airline Web Sites

complished by its customers, as those sustain its well as about removing less relevant elements of
market share. Further, online bookings accom- the model.
modate a tangible potential to lower ARTFLY’s
operational costs. • How frequent should a performance mea-
ARTFLY’s management assesses the perfor- surement system of Web sites be operated
mance of its Web sites and online tools particu- to report to firm managers?
larly on the basis of financial indicators, that is,
the additional revenues and profits gained from Information should be provided to support
those distribution channels in comparison to the decision making that promote the use, the im-
financial investments necessary to construct and provement and the maintenance of Web sites.
to maintain those electronic commerce platforms These decisions also support the outphasing or
and online campaigns. However, the impact of abandonment of Web sites when needed. Man-
ARTFLY’s presence over the Internet goes be- agement teams should determine how often they
yond the direct and immediate financial prospects have to consider these issues.
and has significant positive effects in terms of
customer satisfaction, reputation and the quality • How should the evaluation process of the
of the information and services that it provides. performance of Web sites be organized?
Further, ARTFLY’s management concluded the
way the performance assessment of the Web sites To realize actual use, data have to be collected,
and online booking facilities is carried out could data have to entered into the system, information
be considerably improved. Hence, ARTFLY’s has to produced by the system and the informa-
actual challenges and questions are as follows tion has to be distributed to the relevant people.
and are followed by elaboration of the relevant Management should specify these generic activi-
aspects: ties and the responsibilities of carrying them out
on a continuous basis.
• What are the possible dimensions through
which the performance of Web sites can be • Which organizational units should be re-
determined and assessed? sponsible for the performance evaluation
and take part in it?
The current assessment model can be seen as a
starting point for the evaluation of the performance This issue is related to the question of who is
of Web sites. Firms should address the following primary responsible for the use and for the success
questions: Which dimensions are used in the of Web sites. Depending on the organizational
current model, which aspects are emphasized structure and the roles and responsibilities of
and which are ignored? What are the choices that workers, Web site responsibilities can be divided
are made and how do they reflect the strategy of in different ways. Alternative options are: Web
the company? site managers, business process managers, busi-
ness unit managers, financial managers, division
• How can ARTFLY’s assessment model can managers or CEO’s.
be extended and improved?
• How can a performance measurement sys-
Since the model as reflected in Table 3 and 4 tem be related to the strategies and to the
are an early version of the model management strategic priorities of firms?
should think about further refinements of it, as

1309
Assessing the Performance of Airline Web Sites

Firms have various strategic priorities, for ment models. Journal of Marketing Management,
example, they can follow a low cost strategy, a 18(7-8), 721-747.
differention strategy, a global strategy or a niche
Klein, S., & Loebbecke, C. (2003). Emerging pric-
strategy that can be monitored by the measurement
ing strategies on the web: Lessons from the airline
system. However, those different strategies may
industry. Electronic Markets, 13(1), 46-58.
be evaluated and monitored by different criteria
that assess online processes and functions, as well Klein, S., Kohne, F., & Oorni, A. (2004). Barriers
as the Web sites themselves. to online booking of scheduled airline tickets.
Journal of Travel & Tourism Marketing, 17(2/3),
27-39.
References
Jarach, D. (2002). The digitalisation of market re-
lationships in the airline business: The impact and
Agarwal, R., & Venkatesh, V. (2002). Assessing
prospects of e-business. Journal of Air Transport
a firm’s web presence: A heuristic evaluation
Management, 8(2), 115-120.
procedure for the measurement of usability. In-
formation Systems Research, 13(2), 169. Law, R., & Leung, R. (2000). A study of airlines’
online reservation services on the internet. Journal
Alamdari, F. (2002). Regional development in
of Travel Research, 39(2), 202-213.
airlines and travel agents relationship. Journal of
Air Transport Management, 8(5), 339-348. Monsuwe, T. P., Dellaert, B. G. C., & Ruyter, K.
(2004). What drives consumers to shop online? A
Association of British Travel Agents (2005). Travel
literature review. International Journal of Service
statistics and trends. London: ABTA Press.
Industry Management, 15(1), 102-121.
Buhalis, D. (2004). eAirlines: Strategic and tactical
Lubbe, B. (2007). The effect of Internet appre-
use of ICTs in the airline industry. Information &
hension and website satisfaction on air travellers’
Management, 41(7), 805-825.
adoption of an airline’s website. Journal of Air
Chu, R. (2001). What online Hong Kong travelers Transport Management, 13(2), 75-80.
look for on airline/travel websites? International
Otim, S., & Grover, V. (2006). An empirical study
Journal of Hospitality Management, 20(1), 95-
on Web-based services and customer loyalty.
100.
European Journal of Information Systems, 15,
Drèze, X., & Zufryden, F. (2004). Measurement of 527–541.
online visibility and its impact on Internet traffic.
Park, Y. A., & Gretzel, U. (2007). Success factors
Journal of Interactive Marketing, 18(1), 20-37.
for destination marketing web sites: A qualita-
Gasson, S. (2003). The impact of e-commerce tive meta-analysis. Journal of Travel Research,
technology on the air travel industry. Hershey, 46(1), 46-63.
PA: Idea Group Inc.
Renkema, T. J. W., & Berghout, E. W. (1997).
Gianforte, G. (2004). The world at our fingertips— Methodologies for information system investment
How online travel companies can turn clicks evaluation at the proposal stage: A comparative
into bookings. Journal of Vacation Marketing, view. Information and Software Technology,
10(1), 79-87. 39(1), 1-13.
Huizingh, K. R. E. (2002). Towards successful e- Rubin, R. M., & Joy, J. N. (2005). Where are the
business strategies: A hierarchy of three manage- airlines headed? Implications of airline industry

1310
Assessing the Performance of Airline Web Sites

structure and change for consumers. Journal of Endnotes


Consumer Affairs, 39(1), 215-228.

1
Since the beginning of 2005, the latest ver-
Shon, Z. Y., Chen, F. Y., & Chang, Y. H. (2003). sion of the Web site is available in all the
Airline e-commerce: The revolution in ticketing 65 countries and supports 20 different lan-
channels. Journal of Air Transport Management, guages. ARTFLY’s management demanded
9(5), 325-331. that the Web site would be available online
in 99.99% of the time.
SITA (2006). Airline IT trends survey. Geneva:
SITA Press.
2
On average, airlines expect to save 13%
of their administrative costs by moving to
Smyth, A., & Wagner, C. M. (2006). Business-
online bookings (SITA, 2006).
to-business electronic marketplaces in the airline
industry: Tool for enhanced efficiency in a volatile
3
Google bases its activities and business
business. Transportation Research Record. 1951, model on linking the search terms of users
60-68. to textual adsense advertisements. When
customers click on featured advertisements,
Toh, R. S., & Raven, P. (2003). Perishable asset
they open the promoted Web site. Then
revenue management: Integrated internet mar-
Google charges the advertiser per click.
keting strategies for the airlines. Transportation
Journal, 42(4), 30-43.
Xing, Y., & Grandt, D. B. (2006). Developing a
framework for measuring physical distribution
service quality of multi-channel and “pure player”
internet retailers. International Journal of Retail
and Distribution Management, 34(4), 278-289.
Yoon, M. G., Yoon, D. Y., & Yang, T. W. (2003).
Impact of e-business on air travel markets: Distri-
bution of airline tickets in Korea. Journal of Air
Transport Management, 12(5), 253-260.

This work was previously published in Journal of Cases on Information Technology, Vol. 11, Issue 1, edited by M. Khosrow-
Pour, pp. 47-64, copyright 2009 by IGI Publishing (an imprint of IGI Global).

1311
1312

Chapter 5.7
Aviation-Related Expertise
and Usability:
Implications for the Design of an FAA
E-Government Web Site

Ferne Friedman-Berg
FAA Human Factors Team - Atlantic City, USA

Kenneth Allendoerfer
FAA Human Factors Team - Atlantic City, USA

Shantanu Pai
Engility Corporation, USA

Abstract the findings from this assessment to make design


recommendations for future system enhance-
The Federal Aviation Administration (FAA) Hu- ments that would benefit all users. They discuss
man Factors Team − Atlantic City conducted a why usability assessments are an important part
usability assessment of the www.fly.faa.gov Web of the process of evaluating e-government Web
site to examine user satisfaction and identify site sites and why their usability evaluation process
usability issues. The FAA Air Traffic Control should be applied to the development of other
System Command Center uses this Web site to e-government Web sites.
provide information about airport conditions, such
as arrival and departure delays, to the public and
the aviation industry. The most important aspect INTRODUCTION
of this assessment was its use of quantitative
metrics to evaluate how successfully users with On November 15, 2007, President Bush announced
different levels of aviation-related expertise could actions to address aviation delays during the
complete common tasks, such as determining the Thanksgiving holidays. As part of this announce-
amount of delay at an airport. The researchers used ment, he directed people to visit the Web site fly.

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Aviation-Related Expertise and Usability

faa.gov, which is a Federal Aviation Administra- the Web site (Hung et al., 2006); and personal fac-
tion (FAA) e-government Web site that provides tors such as education level, race, level of current
real time information about airport delays. internet use, and income level (Reddick, 2005).
If a Web site has many functional barriers, such
Fourth, the federal government is using the Inter- as having a poor layout or producing incomplete
net to provide real-time updates on flight delays. search results, customers of the site may not use
People in America have got to know there’s a it (Bertot & Jaeger, 2006).
Web site called Fly.FAA.Gov; that’s where the Early work in e-government has consistently
FAA transmits information on airport backups ignored studying the needs of end users, and there
directly to passengers and their families. If you’re has been little research focusing on the demand
interested in making sure that your plans can -- side of e-government (Reddick, 2005). That is,
aren’t going to be disrupted, you can get on the what are customers looking for when coming to
Web site of Fly.FAA.Gov. As well, if you want to, an e-government Web site? Although there have
you can sign up to receive delay notices on your been many benchmarking surveys conducted on
mobile phones. In other words, part of making e-government Web sites, benchmarking surveys
sure people are not inconvenienced is there to be often do not describe the benefits provided by
-- get transmission of sound, real-time informa- a Web site and only enumerate the number of
tion. (Bush, 2007) services offered by that site (Foley, 2005; Yildiz,
2007). Benchmarks do not evaluate the user’s per-
There has also been a concerted effort by the ception of sites and do not measure real progress
FAA to publicize its Web site by placing adver- in the government’s delivery of e-services. How-
tisements in airports across the United States. ever, governments often chase these benchmarks
Many news outlets now provide airport delay to the exclusion of all other forms of evaluation
information as part of their weather forecasts, (Bannister, 2007).
and this delay information comes, most often, E-government academics emphasize the
directly from the fly.faa.gov Web site. importance of usability testing and highlight the
Because this Web site is the public face of need to focus on Web site functionality, usabil-
a large federal agency, it is important that it ity, and accessibility testing (Barnes & Vigden,
presents the agency in the best light possible. 2006; Bertot & Jaeger, 2006). However, despite
An agency Web site should be a positive public its importance, many organizations still are not
relations vehicle and should not, in itself, create performing usability testing on e-government
any public relations problems. Although use of Web sites. Current work often does not address
e-government Web sites is increasing annually, the needs of different user communities, employ
low user acceptance of e-government Web sites is user-centered design, or use rigorous methods to
a recognized problem (Hung, Chang, & Yu, 2006). test the services being delivered (Bertot & Jaeger;
Many factors affect whether or not someone will Heeks & Bailur, 2007).
use or accept an e-government Web site, including Governments around the world are working
past positive experience with e-government Web to review best practices for e-government evalua-
sites (Carter & Bélanger, 2005; Reddick, 2005); tion methods (Foley, 2005). Because of the social
the ease of use of the Web site (Carter & Bélanger, and economic benefits of providing information
2005; Horst, Kuttschreutter, & Gutteling, 2007); online, it is important that e-government Web site
the perceived trustworthiness of the information designs meet the needs of its targeted users. In
presented on the Web site (Carter & Bélanger; addition, it is important to document the benefits
Horst, et al., 2007); the perceived usefulness of provided by the Web site to increase public support

1313
Aviation-Related Expertise and Usability

(Foley). Carter and Bélanger (2005) point out that ance with all relevant government regulations,
e-government Web sites should be easy to navigate. and work within government budget cycles and
They note that the organization of information on changing government contractors.
the site should be congruent with citizens’ needs.
When consumers visit an e-government Web site, The FAA and fly.faa.gov
they are most frequently looking for information
(Thomas & Streib, 2003), which they need to be The FAA Air Traffic Control System Command
able to find quickly and easily. If users encounter Center provides information about airport condi-
problems while using a Web site, they may become tions, such as arrival and departure delays, to the
frustrated and be less likely to adopt or utilize public and the aviation community via their Web
e-government services in the future. A positive site, www.fly.faa.gov. This Web site allows users
experience with an e-government Web site will to view airport conditions for specific airports.
be communicated to others (Carter & Bélanger), The Web site has many different functions that
and a usable Web site can play a significant role help the user to search for delay information (see
in engendering trust in the agency itself. Figure 1). Using the Search by Region function,
Most Web usability research focuses on e- users are able to look up airports in different
commerce sites and privately run Web sites (Hung geographic regions, such as the Northeastern
et al., 2006), and people expect e-government Web states and the Southeastern states. When using
sites to be as good or as usable as private sector the Search by Airport function, users are able to
sites (Irani, Love, & Montazemi, 2007). People search for airport delay information by typing in
are more likely to use an e-government Web site the name of a city, airport, or a three-letter airport
if the transactions with that site are compatible code. The View by Major Airport function al-
with previously conducted transactions on similar, lows users to search for delay information using
non-government Web sites (Carter & Bélanger, a drop down list of 40 major airports.
2005). The site is also a repository of information
However, there are clear differences between for use by airlines, pilots, passengers, govern-
e-government and e-commerce Web sites. For ment personnel, academics, individual aircraft
instance, e-government sites must provide uni- operators, and other stakeholders in the aviation
versal accessibility so that all citizens have access community. It provides access to real-time and
to information. Additionally, e-government Web historical advisory information, real-time airport
sites are accountable to the public, whereas com- arrival demand information, current reroutes,
mercial Web sites are only accountable to people and reroute restrictions. It also provides access
who have a financial stake in the Web site. It is to information related to air traffic management
not always clear, however, where the boundary tools, a glossary of aviation terms, a national
between these two types of sites lies (Salem, 2003). routes database, pilot tools for making arrival and
Additionally, there are often challenges faced in departure reservations, a collection of National
producing e-government Web sites that are not Airspace System documents, and many other air
faced by commercial sites (Gil-Garcia & Pardo, traffic tools.
2005). For example, when creating e-government The focus of this assessment was on the evalu-
Web sites, designers need to consider whether the ation of site elements that the general public would
project goals align with the goals or mission of access the most, such as the airport delay informa-
the government agency (Yildiz, 2007). They also tion and the glossary of aviation terms. From the
must make sure that all project stakeholders are user’s point of view, the Web site needs to provide
involved, determine whether they are in compli- accurate information quickly, with minimal effort,

1314
Aviation-Related Expertise and Usability

Figure 1. The www.fly.faa.gov home page, illustrating the View by Region, Search by Airport, View by
Major Airport, and Site Map search methods

while minimizing potential mistakes. The site have little, if any, understanding of aviation or its
should be easy for users to learn and provide an associated jargon) also frequently use the site. The
appealing and satisfying experience. Web site is also being touted (Bush, 2007) as the
We faced some unique issues and challenges first place the public should visit on the Web when
when evaluating the fly.faa.gov Web site. First, looking for travel-related delays in the aviation
the fly.faa.gov Web site presents real-time, up-to- system. Therefore, it was important to evaluate
the-minute data, whereas most e-government Web whether this site is usable by people who do not
sites often present static information or informa- have a background in aviation. In this usability
tion that changes infrequently. It was also clear assessment, we examined how effectively people
that the expectations of site users were likely to with different levels of domain knowledge were
be influenced by the information found on more able to use the site.
commercial aviation sites. Because people have It was difficult to identify a single typology that
preconceived notions about the airlines and the described the Web site. Although the site often
reliability of information provided by airlines, it looks like a Government to Consumer (G2C) site
was possible that this perception could transfer (Hiller & Bélanger, 2001), its original purpose was
to their perception of this Web site. to function as a Government to Business (G2B)
The Web site was also originally designed site or a Government to Employee (G2E) site. The
for use by people associated with the aviation site allows people to perform basic transactions
industry, such as pilots and local airport authori- (Hiller & Bélanger, Stage 3), but it also attempts
ties, who have at least a working knowledge of to be a full-service, one-stop site for many types
various aviation concepts. Because it is accessible of aviation related information (Hiller & Bélanger,
on the internet and other travel sites have links Stage 4). For instance, although this evaluation
to it, members of the traveling public (who may did not focus on the G2B information, airlines

1315
Aviation-Related Expertise and Usability

often use the site to find delay information, and traffic control. We included participants of both
general aviation pilots use the site to make route categories.
reservations. Although the site tries to organize its
content to meet the different needs of these differ- Equipment
ent categories of users (Ho, 2002; Schelin, 2003),
it is not clear how the organizational structure The laptops used in the experiment contained
was determined or whether it is the most optimal fully interactive offline versions of the fly.faa.gov
organization for all types of users. Web site. A User Script asked the participants to
We conducted this formal usability assessment use the Web site to find information to answer
to determine how successfully the Web site meets 17 questions: 12 asked users to search for delay
these usability goals and the needs of its users, information, 3 asked users to find the definitions
including both expert and novice users. The as- for aviation-related terms, and 2 asked users to
sessment employed techniques commonly used in identify the authority to be contacted when try-
usability evaluations (Ahlstrom & Longo, 2003; ing to obtain specific information. The script also
Nielsen, 2003). The participants completed a set asked users to use the Search by Region, Search
of representative tasks using the Web site, while by Airport, and View by Major Airport methods
researchers observed and recorded their actions for specific questions. This allowed us to evaluate
and comments. Users also answered a series of the usability of each function.
questions rating the usability of the site. The
data collected through these activities helped us Procedure
identify a number of problems. After identifying
the final list of usability issues, we used a part of Each session lasted 30 to 45 minutes. After sign-
the heuristic evaluation technique (Nielsen) to ing an informed consent form, the participants
determine the most critical issues. This article completed a Background Questionnaire that
discusses the technique used in this evaluation, collected information about the participants’
highlights some of the most critical issues, and knowledge of computers, Web sites, and aviation
provides suggestions to designers on how to fix terminology.
them. We also discuss the benefits of applying After completing the Background Question-
this formal process to the development of other naire, the participants next completed the User
e-government Web sites. Script. We observed each participant during the
experiment and recorded pertinent actions or
comments. At the end of the experiment, the par-
METHODOLOGY ticipants completed a Post-Session Questionnaire,
where they rated their experience and identified
Participants usability issues.
Because using participants who all had a high
We recruited 32 adult volunteers from the FAA level of aviation-related knowledge could have
William J. Hughes Technical Center to serve as biased the results, we used the data to categorize
participants. Because the participants were FAA the participants into three groups (novices, moder-
employees, many had greater aviation-related ate knowledge users, and experts), based on their
knowledge than the general public. However, aviation-related knowledge. We analyzed the
many FAA employees, such as administrative data by level of expertise to determine whether
assistants and facility support workers, do not aviation-related knowledge had an impact on user
have significant knowledge of aviation or air performance. By analyzing the results in this way,

1316
Aviation-Related Expertise and Usability

we could make recommendations targeted toward User Script Data: Overall Analysis
making the site usable for the different user popu-
lations. When even individuals with a high level Of the 12 questions that asked users to find
of aviation-related expertise had trouble using specific delay information, the participants an-
certain features, this provided strong evidence swered 79.4% correctly. For the subset of five
that those features needed to be redesigned. Even delay questions that allowed the participants to
if novices were the only ones who had a problem use their preferred search method, the partici-
with a feature, we rated that problem as severe if pants answered 71.2% correctly. For the subset
the impact for those users was severe. of four Search by Airport questions, 84.5% of
the participants answered the questions correctly.
For the View by Major Airport question, 90.6%
RESULTS of participants found the correct answer; for the
View by Region question, 81.3% found the cor-
Background Questionnaires rect answer; and for the Site Map question, 87.5%
found the correct answer.
The Background Questionnaire asked the par- Three questions asked the participants to use
ticipants questions regarding their familiarity the site to provide the definition of three aviation
with aviation-related terms and acronyms. For related terms and abbreviations. Although 84.4
example, participants were asked to list three- % of participants answered all three questions
letter abbreviations for airports (e.g., Philadelphia correctly, 6.3% answered one incorrectly, 3.1%
International Airport = PHL), or were given the answered two incorrectly, and 6.3% were not able
three-letter abbreviations and asked to list the to answer any of the questions. By comparing
airports associated with those abbreviations (e.g., the percentage of participants who answered a
MIA = Miami International Airport). Using the question correctly, we determined that all three
correct responses to these and other aviation- questions were equally difficult.
related questions, we categorized the participants Two questions asked the participants to find
as novices (n = 8), moderate knowledge users whom to contact to obtain information about the
(n = 15), and experts (n = 9). The novices were status of an individual flight or why an airport
slightly younger than both the experts and those was closed. For these questions, only 28.1% of
with moderate-knowledge (Mnovice = 41.6 years, the participants answered both questions cor-
Mmoderate = 49.9 years, Mexpert = 49.1 years). More rectly, 56.2% answered one incorrectly, and 15.6%
than 70% of novices and those with moderate answered both incorrectly.
knowledge reported never using the fly.faa.gov
Web site. In contrast, 75% of the experts reported User Script Data: Analysis by Level
using the Web site a few times a year. of Expertise
All the participants had extensive experience
using computers and the Web. Because we found We analyzed the data by level of expertise to
no discernable differences in reported Web and determine whether aviation-related knowledge
computer use among the participants, we were had an impact on user performance. Analyzing
unable to stratify the participants based on these all 17 questions, we found an effect of expertise
factors. on overall task performance, F (2, 29) = 3.54, p =

1317
Aviation-Related Expertise and Usability

.04. Post hoc pairwise contrasts indicated expert three of the questions were significant, 7 of the
participants were able to answer significantly 12 delay questions showed the expected pattern
more questions than novices (85.6% vs. 69.1%, p of results (see Table 1). Therefore, we also tested
= .01), and there was a trend suggesting moderate- the binomial probability that 7 of the 12 delay
level users answered more questions than novices questions would show the expected ordering of
(79.6% vs. 69.1%, p = .07). expert > moderate > novice. We found that it
We performed ordinal (linear) chi-square tests was unlikely that this pattern would occur by
on individual questions to determine whether the chance 7 out of 12 times, p < .001. This suggests
percentage correct increased or decreased across that experts were better able to find information
the user categories (Howell, 2007). Although only on the fly.faa.gov Web site than moderate users,

Table 1. Percentage correct by level of aviation-related expertise


Questions % Correct
Moderate
Novices Experts
Users
1. Los Angeles to Salt Lake City.** 75.0 100.0 100.0
2. Portland to Memphis. 25.0 53.3 33.3
3. Denver to Philadelphia. Search by Air-
87.5 93.3 88.9
port.

4. Houston to Chicago. Search by Airport.** 62.5 73.3 100.0

5. Newark to Burlington. 50.0 73.3 88.9


6. Las Vegas to New York. View by Major
75.0 93.3 100
Airport.
7. Phoenix to Dallas.* 12.5 73.3 77.8
8. Cincinnati/Northern Kentucky to Detroit.
75.0 80.0 88.9
View by Region.
9. Pittsburgh to Washington DC. Site Map. 75.0 86.7 100.0
10. New York to San Jose. Search by
75.0 80.0 100.0
Airport.

11. Orlando to St. Louis. Search by Airport. 87.5 86.7 77.8

12. Houston to Tulsa. 87.5 86.7 100.0


Using information available on the site, pro-
vide the definitions of the following aviation-
related terms or abbreviations:
13. CIGS 87.5 93.3 88.9

14. MULTI-TAXI 87.5 86.7 88.9

15. VOL 75.0 93.3 100.0


Using information available on the site, who
should a visitor contact to obtain information
about the following:
16. Status of an individual flight 100.0 78.6 87.5

17. Why an individual airport was closed 50.0 26.7 44.4


* p < .10, two-tailed. * * p < .05, two-tailed.

1318
Aviation-Related Expertise and Usability

who in turn were better than the novices. We did the information on the site was, a rating of 1
not find the same pattern for the aviation term or indicated too little detail and a 6 indicated too
contact information questions. much detail. For these summaries, we omitted
We grouped the questions to analyze perfor- responses from the participants who chose more
mance on the different subsets of questions. For than one number on the rating scale. The ratings
the 12 questions that asked users to find specific indicated that the participants thought it was
delay information, novices, moderate-level users, fairly easy to find information on the site (M =
and experts answered 65.6%, 81.7%, and 88% of 4.4, SD = .8) and that they understood informa-
the questions correctly, F (2, 29) = 5.04, p = .01. tion once they found it (M = 4.8, SD = 1.0). The
Post hoc pairwise contrasts indicated experts participants also found it fairly easy to navigate
and moderate-level users were better able to find between pages on the site (M = 4.9, SD = 1.2) and
delay information than novices (p = .005 and p = found the design of the site to be consistent (M
.021, respectively). = 4.9, SD = 1.0). They indicated that there was
We further divided the 12 delay questions somewhat too much detail (M = 3.9, SD = 0.8), but
into subcategories based on search method. For that information on the site was fairly readable
the subset of questions that allowed people to (M = 4.8, SD = 1.1). Finally, they indicated that,
find information using their preferred search overall, they were mostly satisfied with the site (M
method, we found an effect of expertise on user = 4.7, SD = 0.8). When we compared satisfaction
performance, F (2, 29) = 9.93, p = .001. Experts ratings to actual performance, it was apparent that
and moderate users performed better than novices participants were not able to accurately estimate
when searching for delay information using their performance, given that they answered an average
preferred search method, answering an average of 20.1% questions incorrectly. However, despite
of 80% and 77.3% of the questions correctly, their performance, the participants still reported
while novices only answered an average of 50% high satisfaction with the site. Given this dissocia-
correctly (p < .001 for both post hoc pairwise tion between performance and satisfaction, it is
comparisons). important that usability experts evaluate not just
For the four delay questions that asked us- user satisfaction, but actual user performance,
ers to specifically use the Search by Airport when evaluating a Web site.
method, novices, moderate users, and experts
answered 78.1%, 83.3%, and 91.7% of them cor- Post-Session Questionnaire:
rectly. Although these results were not statistically Analysis by Level of Expertise
significant, they demonstrated the same trend as
the other sets. We found no significant differences in the rat-
ings between experts, moderate-level users, and
Post-Session Questionnaire: novices. There were, however, some interesting
Overall Analysis trends in the data. The ratings on information
comprehensibility indicated that experts found
The Post-Session Questionnaire asked the partici- the information to be somewhat more compre-
pants to rate their subjective experience with the hensible than moderate-level users, who, in turn,
fly.faa.gov Web site using 6-point scales. Except found the information to be more comprehensible
for the question asking about the level of detail, than novices. In evaluating design and layout
higher ratings indicated positive responses and consistency, the experts were the least satisfied
lower ratings indicated negative responses. For with the design consistency, with novices being
the question that asked the users how detailed the most satisfied, and moderate users falling

1319
Aviation-Related Expertise and Usability

somewhere in the middle. For the ratings on the 4 = major usability problem: important to fix,
level of detail, experts gave the highest ratings should be given high priority
(i.e., slightly too much detail), with novices giving 5 = usability catastrophe: extremely important
the lowest ratings (i.e., slightly too little detail), to fix, should be given highest priority
and moderate users falling in the middle (i.e., an
appropriate level of detail). After each researcher independently assigned
a severity rating for each issue, we averaged them
Rating of Usability Issues to compute a consolidated severity rating (Nielsen,
2003). These consolidated severity ratings provide
Using comments and questionnaire ratings made a good estimate of additional usability efforts
by the participants, along with our observations needed when developers establish priorities for
of the participants while they completed the User future enhancements. We rank ordered the usabil-
Script, we compiled a consolidated list of usability ity issues from those having the highest severity
issues and rated the severity of each issue (for a rating to those having the lowest.
comprehensive list, see Friedman-Berg, Allen- The following section discusses the eight us-
doerfer, & Pai, 2007). When rating the severity ability issues that had the highest severity rating
of each problem, we considered the following and provides suggestions and design recom-
factors (Nielsen, 2003). mendations regarding how these issues could be
resolved. User interface design standards and
1. Frequency: Is the problem very common best practices drive these suggestions (Ahlstrom
or very rare? & Longo, 2003). In some cases, we developed
2. Impact: How easy is it for the users to over- simple prototypes to demonstrate potential design
come the problem when navigating through concepts that designers could use to remediate
the Web site? some of these issues.
3. Persistence: Can users overcome the prob-
lem once they know about it, or will the Issue 1: User Confusion
problem bother users repeatedly? Regarding Delay Types

The researchers rated each issue as having The primary purpose of fly.faa.gov is to provide
high, medium, or low frequency, impact, and travelers with airport delay information. For
persistence, and then used these three ratings example, a traveler going from Philadelphia to
to determine a severity rating from 0 to 5. The Miami might want to find out about departure
severity rating scale was adapted from Nielsen delays at PHL and arrival delays at MIA. The
(2003). traveler also might have some interest in the causes
of delays, which can include factors like weather,
0 = I don’t agree that this is a usability problem airport construction, and traffic flow programs.
at all However, the difference between delay types
1 = minor/ cosmetic problem only: not necessary was not readily apparent to many participants.
to fix, should be given lowest priority For example, one question asked users to find
2 = usability problem: small benefit from fixing, information about delays at their arrival destina-
should be given low priority tion. The arrival airport had no arrival delays, but
3 = moderate usability problem: moderate ben- did have general departure delays. Because the
efit from fixing, should be given medium instructions indicated that they were arriving at
priority that airport, the participants should have focused

1320
Aviation-Related Expertise and Usability

on the lack of an arrival delay, but only 40.6% of multiple places on the same page. The information
the participants answered this question correctly. was dense, used too much text, and was not well
Those who answered incorrectly seemed to be organized. In many instances, the participants
looking at the departure delay, which indicated had difficulty finding the delays that were rel-
that they did not understand which delays were evant for them. Displaying so much information
relevant for them. This issue received a mean can be especially problematic when users are in
severity rating of 4.3, SD = 0.5. a hurry to find information. Users may scan too
It is important that the site provide users with quickly and get lost. They may read the wrong
the information they want without requiring them line, overlook information they are looking for,
to understand difficult air traffic concepts. We or see a big block of text and give up.
also found that novices had greater difficulty in We recommend simplifying and reorganizing
finding delay information than both moderate level these pages to make it easier for users to find and
users and experts. This was likely due to novice understand information on the page. The page
users not understanding more technical concepts. could use a tabular layout arranged in columns
We recommend that the site not try to present and organized by arrivals and departures (see
difficult concepts to the lay public, but instead Figure 3). Much of the text information is not
present information in a less technical manner. useful, creates clutter, and should therefore be
For instance, instead of referencing ground delay removed. Because the distinction between general
programs as the cause of a delay, the site could departure delays and destination-specific delays
indicate that a delay was due to congestion. For is not clear to users, it should be deemphasized or
users seeking more detailed information, the Web eliminated. Finally, all delay information related
site could provide additional information about to an individual airport should be consolidated.
ground delay programs using links to additional Presenting two sets of delay information for
pages. one airport, especially if the data are inconsistent,
Because the participants were not always able is confusing. The Web site should avoid going into
to identify relevant delays, we recommend that too much technical detail regarding the causes
the site provide users with a capability that gives of delays. It might instead use icons or graphics
them easy access to pertinent delay information. (e.g., clouds with snow, clouds with rain) to depict
For example, the site might provide an interac- weather or other causes of delays. The Web site
tive tool that allows users to input departure and could offer links to additional information for
arrival airports or click on city pairs to generate advanced users.
a single report on relevant delays for air traffic
traveling between a pair of airports. Issue 3: Overuse of Aviation-
Related Acronyms and Jargon
Issue 2: Information Presentation:
Clutter and Redundant Information The site uses too many aviation-specific acronyms
and jargon when providing specific information
The participants’ comments and researchers’ about the causes of delays. This issue received a
observations suggested that there was too much mean severity rating of 4.0, SD = 0.0. Aviation-
information on the typical search results page specific acronyms, abbreviations, and jargon are
(see Figure 2). This issue received a mean sever- difficult for the general public to understand, and
ity rating of 4.3, SD = 0.5. The site sometimes the glossary is difficult to find. The average user
presented information for a single airport in of the Web site may never be aware that it exists.

1321
Aviation-Related Expertise and Usability

Figure 2. Crowded Airport Status Information page

Figure 3. Airport Status Information in a redesigned format

When the participants had to find the definition Issue 4: User Confusion with
of three aviation-related terms, 16% were unable Using the View by Region Maps
to find the definition for at least one of them.
Therefore, we recommend eliminating the use The fly.faa.gov Web site provides users with a
of these terms when they are not essential. This View by Region search function that allows us-
would eliminate unnecessary detail, simplify the ers to look up airports by searching in different
site, and make it easier to use and understand. geographic regions. These regions include the
Northeast, North Central, Northwest, Southeast,

1322
Aviation-Related Expertise and Usability

South Central, and Southwest regions, along Issue 5: Lack of User Knowledge
with Alaska and Hawaii. When a user uses the Regarding Three-Letter Airport
View by Region function, they are taken to a Identifiers
map that contains only states that are part of a
region. However, it is not easy for someone with All commercial airports have three-letter identi-
little knowledge of geography to determine the fiers, and using them is an efficient way to obtain
region for a particular state. The participants got delay information about an airport. The site
lost when looking for airports that were not on provides a function that allows users to type a
the main U.S. map because they were unable to three-letter identifier directly into the Search by
determine the relationship of regional maps to Airport text box, which will take the user to the
the main U.S. map. This was especially difficult details page for that airport. It also provides cues
for states such as Ohio that lie at the edge of a to site users by labeling airports on the main U.S.
region. These issues make the View by Region map with their three-letter identifiers (see Figure
method difficult for the general public to use and 1). However, many participants did not know the
the participants found the View by Region maps to correct three-letter identifiers for airports and did
be confusing. This issue received a mean severity not use the cues on the main map to determine
rating of 4.0, SD = 0.0. the correct identifier. This issue received a mean
One question asked the participants to find severity rating of 3.3, SD = 0.6.
delay information for an airport that was not avail- The site should emphasize that the Search by
able on the main map or on the View by Major Airport text box accepts regular airport names
Airport menu. Only 71.9% of the participants and city names in addition to three-letter identi-
found the correct answer for this question, indi- fiers. Although the Search by Airport text box
cating that the participants had some difficulty does have a label indicating that users can enter
finding information when they needed to drill city, airport code, or airport name information in
down on the maps. this field, we recommend that the Web site provide
There are several recommendations that could the user with specific examples to highlight and
alleviate some of the issues related to the use of better explain the different search options.
the View by Region method. First, the site could
place an outline around the different regions or Issue 6: The Search by Airport
use color coding to highlight the different regions Function Returns Redundant and
on the U.S. map. This would help users identify Irrelevant Results
which states belong in which region. The site
could display split portions of the main U.S. City name searches using the Search by Airport
map on the same page to better orient users to function generate an intermediate results page
the different regions. To familiarize people with that lists multiple airports. These listings often
relevant geographic information, the site could contain redundant and irrelevant results. This is-
label states, both on the main U.S. map and on the sue received a mean severity rating of 3.3, SD =
smaller regional maps. The site could also offer 0.6. For example, a search for Chicago generates
users a drop-down menu that listed the various a search results page listing two airports: Midway
airports by state. and O’Hare International. The site lists each re-
sult twice, once under City Name Matches and
once under Airport Name Matches (see Figure
4). This format is confusing and users may not

1323
Aviation-Related Expertise and Usability

Figure 4. The www.fly.faa.gov results page for a Search by Airport search for Chicago

realize that both links take them to the same when users make spelling mistakes, and should
information. Some participants questioned why ignore spacing errors.
the site listed an airport twice. We recommend
that the Airport Lookup Search Results page Issue 8: Inconsistent Use of
consolidate search results and list airports only Pop-up Windows
once in any search results list.
The fly.faa.gov Web site is inconsistent in its use
Issue 7: User Spelling and of pop-up windows. When users access informa-
Misspellings and Their Impact on the tion using the Search by Airport method or when
Search by Airport Function they click on the color-coded dots on the main site
map, the Web site displays the search results in
User spellings and misspellings can have a seri- a pop-up window. However, when users access
ous impact on the Search by Airport function. information using the View by Major Airport
In some instances, the correct spelling does not method, the site displays the same information in
work, but a misspelling does. For example, typing the current browser window rather than in a pop-
O’Hare does not return any results, but Ohare up window. This issue received a mean severity
does. Typing LaGuardia returns no results, but La rating of 3.0, SD = 0.0.
Guardia does. In addition, common misspellings During the assessment, some participants
do not produce any results at all, even when the accidentally closed the browser by clicking the
system could provide reasonable guesses about Close button when search results appeared in
what the user intended. For example, Newyork the main browser window. These participants
does not produce any search results at all. This had become accustomed to results appearing in
issue received a mean severity rating of 3.3, SD a pop-up window. When search results appeared
= 0.6. The participants quickly became frustrated in the main browser window, they still reacted as
and confused when the site did not return any if they were in a pop-up window and accidentally
search results for correct spellings or reasonable closed down the site, along with the browser.
misspellings. The search function should always We recommend that the site be more consistent
result in a hit when the correct spelling is used, in how it returns search results and Airport Status
should provide “best guess” search result even Information pages. Users become confused when
the site responds differently to similar actions.
If the standard convention of the site is to bring
up search results in pop-up windows, then the

1324
Aviation-Related Expertise and Usability

site should bring up all search results in pop-up conjunction with user comments and researcher
windows. observations, we were able to identify a number
of human factors issues with the Web site that
we would not have identified by relying solely
DISCUSSION on subjective data.
After identifying issues, we rated each one in
The level of aviation-related expertise had an terms of its impact on site usability, discussed each
impact on many aspects of user performance. issue in detail, identified supporting data when
Experts were more likely than novices and appropriate, and provided recommendations for
moderate-level users to have had some prior improving the usability of the Web site. Many
interaction with the fly.faa.gov Web site. They of the suggested improvements should be easy
were also better at finding delay information on to implement and should further increase user
the Web site. Experts appeared to have a better satisfaction and site usability.
conceptual understanding of the different types
of airport delays than both novices and moderate
users. Finally, experts indicated that they found CONCLUSION
the information on the Web site to be slightly more
comprehensible than both novices and moderate One of the primary lessons that we learned from
level users. Although we realize that there may this usability evaluation is that developers should
be some performance decrement for people who not simply rely on subjective reports of usability
have no affiliation with the FAA, we expect that when evaluating e-government Web sites. It is
their performance and their issues should be most just as important to observe users interacting
similar to our novice users. with a Web site and collect objective performance
On the basis of performance differences, we data to better identify usability issues. By having
recommend that the primary goal of site design- people use the Web site to find different types of
ers should be to make the site more usable for information, we were better able to identify those
people who do not have an aviation background. If areas of the site that caused problems for users.
people in the general public visit this site without To encourage organizations to perform usability
an aviation-related background, we would expect evaluations on e-government sites, we should
them to have substantial difficulty (a) understand- ensure that they provide value by identifying
ing which delays were relevant for them, (b) important usability issues that can be remedied
understanding how airport delays differ from through redesign. As we saw in this evaluation,
airline delays, and (c) interpreting much of the subjective reports often fail to identify these issues.
jargon used by aviation experts. Although both If research on Web site usability fails to identify
experts and novices use the site, simplifying the significant usability issues, it is likely that such
Web site should help all users, not just novices. evaluations will not be used.
Links to additional information can be provided We also found that having researchers rate the
for expert users. severity of usability issues improved our evalua-
Subjective reports indicated that the partici- tion. Future e-government usability assessments
pants were generally satisfied with the fly.faa.gov could reap benefits by using this technique. Many
Web site, and objective data revealed that they times, when a usability assessment is performed,
could successfully complete most tasks using the output of the assessment is a laundry list of is-
the site. By evaluating user performance data in sues that usability experts present to site designers.
If guidance is given on issue severity or criticality,

1325
Aviation-Related Expertise and Usability

it is usually ad hoc and is not derived using any An assessment of the validity and value of e-
formal methodology. By requiring evaluators to government comparisons, International Review
explicitly rate each item on frequency, impact, of Administrative Services, 73, 171-188.
and severity, they are required to think about how
Barne, S. J., & Vidgen, R. T. (2006). Data tri-
and in what ways the problem will affect the user.
angulation and Web quality metrics:  A case
This user-centric focus is the key element of this
study in e-government source. Information and
methodology. It allows site evaluators to provide
Management, 4, 767-777.
designers with a roadmap of how they can best
focus their effort to provide a more optimal user Bertot, J. C., & Jaeger, P. T. (2006). User-centered
experience. Additionally, we recommend that us- e-government: Challenges and benefits for gov-
ability assessments use more than one evaluator ernment Web sites. Government Information
to make severity ratings. We found that different Quarterly, 23, 163-168.
evaluators might have different priorities, but by
Bush, G.W. (2007, November 15). President Bush
using combined severity ratings from three or
discusses aviation congestion. Office of the Press
more evaluators, you can increase the reliability
Secretary [Press release]. Retrieved February 22,
of the ratings (Nielsen, 2003).
2008, from http://www.whitehouse.gov/news/
By employing an evaluation processes like the
releases/2007/11/20071115-6.html
one used in this study to evaluate e-government
sites, whether they are G2B sites, G2C sites, Carter, L., & Bélanger, F. (2005). The utilization
or G2E sites, designers and system developers of e-government services: Citizen trust, innova-
can better allocate limited resources during the tion and acceptance factors. Information Systems
design process. In general, it is important that Journal, 15, 5-25.
e-government Web site designers take into con-
Foley, P. (2005). The real benefits, beneficiaries
sideration the demographics of those who will use
and value of e-government. Public Money &
their Web site or application. If an e-government
Management, 25, 4-6.
Web site or application, initially targeted to us-
ers with a specific area of expertise, is going to Allendoerfer, K., Friedman-Berg, F., & Pai, S.
be redesigned for use by the general public, the (2007). Usability assessment of the fly.faa.gov Web
site must be evaluated for usability. Based on site (DOT/FAA/TC-07/10). Atlantic City Interna-
the results of such an evaluation, changes need tional Airport, NJ: Federal Aviation Administra-
to be made to ensure that the site is usable by the tion William J. Hughes Technical Center.
broadest possible audience.
Gil-García, J. R., & Pardo, T. A. (2005). E-
government success factors: Mapping practical
tools to theoretical foundations. Government
REFERENCES
Information Quarterly, 22, 187-216.
Ahlstrom, V., & Longo, K. (Eds.). (2003). Hu- Heeks, R., & Bailur, S. (2007). Analysing eGov-
man factors design standard for acquisition ernment research. Government Information
of commercial-off-the-shelf subsystems, non- Quarterly, 22, 243-265.
developmental items, and developmental systems
Hiller, J., & Belanger, F. (2001). Privacy strategies
(DOT/FAA/CT-03/05/HF-STD-001). Atlantic
for electronic government. E-Government series.
City International Airport, NJ: FAA William J.
Arlington, VA: Pricewaterhouse Coopers Endow-
Hughes Technical Center.
ment for the Business of Government.
Bannister, F. (2007). The curse of the benchmark:

1326
Aviation-Related Expertise and Usability

Ho, A. (2002). Reinventing local government and Hershey, PA: Idea Group Publishing.
the e-government initiative. Public Administration
Thomas, J. C., & Streib, G. (2003). The new face
Review, 62, 434-444.
of government: Citizen-initiated contacts in the
Horst, M., Kuttschreutter, M., & Gutteling, J. M. era of e-government. Journal of Public Adminis-
(2007). Perceived usefulness, personal experi- tration Research and Theory, 13, 83-102.
ences, risk perception, and trust as determinants
Yildiz, M. (2007). E-government research:
of adoption of e-government services in the
Reviewing the literature, limitations, and ways
Netherlands. Computers in Human Behavior,
forward. Government Information Quarterly,
23, 1838-1852.
24, 646-665.
Howell, D. C. (2007). Chi-square with ordinal
data. Retrieved January 22, 2007, from http://
www.uvm.edu/~dhowell/StatPages/More_Stuff/
OrdinalChisq/OrdinalChiSq.html
Hung, S. Y., Chang, C. M., & Yu, T. J. (2006).
Determinants of user acceptance of the e-gov-
ernment services: The case of online tax filing
and payment system, Government Information
Quarterly, 23, 97-122.
Irani, Z., Love, P. E. D, & Montazemi, A. (2007). E-
Government: Past, present, and future. European
Journal of Information Systems, 16, 103-105.
Nielsen, J. (2003). Severity ratings for usability
problems. Retrieved April 15, 2006, from http://
www.useit.com/papers/heuristic/severityrating.
html
Reddick, C.G. (2005). Citizen interaction with
e-government: From the streets to servers? Gov-
ernment Information Quarterly, 22, 38-57.
Salem J. A. (2003). Public and private sector
interests in e-government: A look at the DOE’s
PubSCIENCE. Government Information Quar-
terly, 20, 13-27.
Schelin, S. H. (2003). E-government: An overview.
In G. D. Garson (Ed.), Public information technol-
ogy: Policy and management issues (pp. 120-137).

This work was previously published in International Journal of Electronic Government Research, Vol. 5, Issue 1, edited by V.
Weerakkody, pp. 64-79, copyright 2009 by IGI Publishing (an imprint of IGI Global).

1327
1328

Chapter 5.8
Quality Enhancing
the Continued Use of
E-Government Web Sites:
Evidence from E-Citizens of Thailand

Sivaporn Wangpipatwong
Bangkok University, Thailand

Wichian Chutimaskul
King Mongkut’s University of Technology Thonburi, Thailand

Borworn Papasratorn
King Mongkut’s University of Technology Thonburi, Thailand

ABSTRACT INTRODUCTION

This study empirically examines Web site quality Electronic government, so called e-government,
toward the enhancement of the continued use of has been broadly defined as the use of informa-
e-government Web sites by citizens. The web site tion and communication technology (ICT) to
quality under examination includes three main transform government by making it more acces-
aspects, which are information quality, system sible, effective, and accountable (infoDev & CDT,
quality, and service quality. The participants were 2002). The Internet is indeed the most powerful
614 country-wide e-citizens of Thailand. The data and popular means of delivering e-government.
were collected by means of a web-based survey Hence, Web sites have been employed as a plat-
and analyzed by using multiple regression analysis. form for delivering a wide range of government
The findings revealed that the three quality aspects services electronically.
enhanced the continued use of e-government Web By using e-government Web sites, citizens
sites, with system quality providing the greatest can conveniently access government information
enhancement, followed by service quality and and services and gain greater opportunities to
information quality. participate in the democratic process (Fang, 2002).

Copyright © 2010, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Quality Enhancing the Continued Use of E-Government Web Sites

Citizens can access government information and service quality to examine the continued use in
services anywhere and anytime. Thus, the time the context of e-government Web sites.
spent in traveling and waiting is reduced. From the This study therefore aims to examine the Web
government’s point of view, the more citizens that site quality toward enhancement of the continued
use e-government Web sites, the more operation use of e-government Web sites by citizens. The
and management costs are reduced. population of interest for this study is e-citizens of
To obtain these benefits, the initial adoption Thailand, a group of citizens who has experienced
and subsequent continued use of e-government Thailand’s e-government Web sites. The reason
Web sites by citizens are required. In general, an that makes Thailand an ideal place to study is that
information system indicated that its eventual e-government is considered a new innovation to
success depends on its continued use rather than Thai citizens and is conceived as a fundamental
first-time use (Bhattacherjee, 2001; Limayem, element to encourage the country development.
Hirt, & Cheung, 2003). Likewise, initial use of In the next section, the background of this study
e-government Web sites is an important indicator is briefly reviewed. Thereafter, the research model
of e-government success. However, it does not and hypotheses development, research methodol-
necessarily lead to the desired outcome unless a ogy, and data analysis are presented. Finally, the
significant number of citizens move beyond the discussion, limitations, and suggestions for future
initial adoption and use e-government Web sites research are given.
on a continual basis. To enhance the continued use,
this study proposes that quality of e-government
Web sites is one significant factor. BACKGROUD OF STUDY
According to DeLone and McLean (2002), the
three quality aspects, information quality, system DeLone and McLean’s Information
quality, and service quality, are the determinants System Success Model
that effect user’s intention to use an information
system. In practice, these three aspects have been In order to ascertain the success of an information
employed to study the initial intention to use the system, DeLone and McLean (1992) proposed the
information system and to evaluate the quality of Information System Success Model (referred here-
information system (e.g., Lee & Kozar, 2006; Ne- after as the ‘D&M IS Success Model’) as shown in
gash, Ryan, & Igbaria, 2003; Wilkin & Castleman, Figure 1. The model asserts that system quality and
2002). However, there is a lack of prior research information quality are the determinants of system
that uses information quality, system quality, and

Figure 1. Original DeLone and McLean’s Information System Success Model (DeLone & McLean,
1992, p.87)

System Use
Quality
Individual Organizational
Impact Impact

Information User
Quality Satisfaction

1329
Quality Enhancing the Continued Use of E-Government Web Sites

use and user satisfaction which effect individual Information Technology Committee (NITC).
and organizational impact respectively. Several programs, such as computer training for
However, Pitt, Watson, and Kavan (1995) mid-level officers, specifying minimum require-
noticed that commonly used measures of informa- ments of information technology equipment for
tion system effectiveness focus on the products, government agencies, and the establishment of
rather than the services. They then proposed that a Chief Information Officer (CIO) in the public
service quality needs to be considered as an ad- sector, have been imposed to support and promote
ditional measure of the D&M IS Success Model. this initiative (Chamlertwat, 2001). The School-
DeLone and McLean (2002) therefore reformu- Net Thailand (a national school informatization
lated the D&M IS Success Model by including program to empower all schools to access a large
service quality as an additional determinant that pool of information resources using the Internet),
effects the use and user satisfaction as shown in Government Information Network or GINet
Figure 2. (a government backbone network to facilitate
D&M IS Success Model has become popular intra- and inter- agencies communication and
for the specification and justification of the mea- information exchanges), and the development of
surement of the dependent variable in an infor- legal infrastructure to support the application of
mation system research. In the summer of 2002, information technology have also been initiated
a citation search yielded 285 refereed papers in under the first National IT Policy Framework
journals and proceedings that referenced D&M for the year 1996−2000 (IT 2000) (Ateetanan,
IS Success Model during the period 1993 to mid- 2001).
2002 (DeLone & McLean, 2003). In practice, a In March 2001, the two-year period e-govern-
number of empirical studies (e.g., Iivari, 2005; ment project was established by NITC to establish
Molla & Licker, 2001; Seddon & Kiew, 1994) gave a framework for building up e-government and to
support for the associations among the measures implement some pilot projects. In March 2002,
identified in the D&M IS Success Model. the National IT Policy Framework for the year
2001−2010 (IT 2010) was approved by the cabinet
E-Government in Thailand and e-government was a manifest flagship, in ad-
dition to e-industry, e-commerce, e-society, and
Like many countries, Thailand has been fully e-education. Subsequently, in September 2002,
cognizant of both the potentials and the benefits the cabinet further endorsed the first National ICT
of e-government. Since 1996, the e-government Master Plan for the year 2002−2006. The master
development has been driven by the National

Figure 2. Reformulated DeLone and McLean’s Information System Success Model (DeLone & McLean,
2002, p.2974)
Information
Quality Use
(Intention to Use)
System Net Benefits
Quality

User Satisfaction
Service
Quality

1330
Quality Enhancing the Continued Use of E-Government Web Sites

plan devises seven key strategies. One of which the meaning of all constructs and the theories
is e-government (NECTEC, 2003). supporting the relationships are presented.
As a result, e-government has been develop-
ing rapidly in Thailand. In early 2004, NECTEC Information Quality
initiated the first on-line survey of government
e-services. The survey revealed that all 267 According to DeLone and McLean (1992), infor-
government agencies have Web sites to provide mation quality is concerned with the measure of
information to the public (NECTEC, 2005a). the information that the system produces and deliv-
With accordance to the global ranking, the E- ers. When applied to this study, the information
government Readiness Survey of United Nations quality focuses on characteristics of information
(United Nations, 2003; United Nations, 2004; produced by e-government Web sites.
United Nations, 2005) reported that Thailand Quality of information is believed to be
owned an E-government Readiness Ranking at the most salient factor for predicting customer
56 from 191 global countries in 2003, moved up decision-making behavior (Jeong & Lambert,
to the 50th rank in 2004, and finally at the 46th 2001) and user intention to use a particular system
rank in 2005. (DeLone & McLean, 1992, 2002; Molla & Licker,
In terms of overall usage, there is not much 2001). Furthermore, information quality has long
usage of e-government Web sites compared been found associated with customer or user sat-
to other Web site categories. According to the isfaction in previous empirical studies (Seddon &
Truehits 2005 Award (Truehits, 2006) govern- Kiew, 1994; Spreng, MacKenzie, & Olshavsky,
ment Web sites occupied the 15th rank from 19 1996; Szymanski & Hise, 2000; Negash, Ryan,
Web site categories and had proportion of usage & Igbaria, 2003; Iivari, 2005).
only 1.64%, which went down 0.13% from the Concerning to the case of e-government, the
previous year. quality of information on the Web sites is very
At this stage it is not clear if Thai citizens significant since most citizens use e-government
will continue using Thailand’s e-government Web sites for informational purposes (Accenture,
Web sites. Hence, better understanding of the 2004), and the first phase of e-government imple-
factors that enhance citizens to continue using mentation is to publish government information
e-government Web sites can create greater value (infoDev & CDT, 2002). If e-government Web
for Thailand’s government and also other govern- sites contain low information quality, they are
ments all over the world. useless. Furthermore, high quality information
encourages citizens to use the Web sites (Cullen
& Hernon, 2004). Hence, the following hypothesis
RESEARCH MODEL AND is proposed:
HYPOTHESES DEVELOPMENT
H1: Information quality of e-government Web
Based on the review of the aforementioned lit- sites enhances the continued use of e-government
erature, the conceptual research model used to Web sites.
guide this study is proposed as shown in Figure
3. The model is based on the three quality aspects According to a review of related literature
of the D&M IS Success Model, adapting to the (Bailey & Pearson, 1983; Doll & Torkzadeh,
e-government Web site context. In the following, 1988; Wang & Strong, 1996), the fundamental

1331
Quality Enhancing the Continued Use of E-Government Web Sites

Figure 3. Conceptual research model


Quality of Websites

Information Quality
• Accuracy
• Timeliness H1
• Relevance
• Understandability
• Completeness

System Quality
• Functionality
H2
• Dependability Continued Use
• Ease of Use
• Usefulness

Service Quality
• Tangibles
• Reliability H3
• Empathy
• Responsiveness
• Assurance

dimensions of information quality is composed of & McLean, 1992, 2002). The more satisfied the
five dimensions: accuracy, timeliness, relevance, user is with the system the more he or she will
understandability, and completeness. This study be inclined to use it. Conversely, if system use
thereby uses these five dimensions to measure does not meet the user’s needs, satisfaction will
citizens’ perceptions toward information quality not increase and further use will be avoided
of e-government Web sites. Table 1 shows a brief (Baroudi, Olson, & Ives, 1986). Therefore, this
definition of each dimension. study postulates that:

System Quality H2: System quality of e-government Web sites


enhances the continued use of e-government
According to DeLone and McLean (1992), sys- Web sites.
tem quality is concerned with the measure of the
actual system which produces the output. The Based on a review of related literature (Bailey
system quality in this study therefore focuses & Pearson, 1983; Doll & Tork

You might also like