Download the Full Ebook and Access More Features - ebooknice.

com

(Ebook) Recent Trends in Computer Applications:

Best Studies from the 2017 International
Conference on Computer and Applications, Dubai,
UAE by Jihad Mohamad Alja’am, Abdulmotaleb El
Saddik, Abdul Hamid Sadka ISBN 9783319899138,
https://ebooknice.com/product/recent-trends-in-computer-
9783319899145, 3319899139, 3319899147
applications-best-studies-from-the-2017-international-
conference-on-computer-and-applications-dubai-uae-7322368

OR CLICK HERE

DOWLOAD EBOOK

Download more ebook instantly today at https://ebooknice.com

 Instant digital products (PDF, ePub, MOBI) ready for you
Download now and discover formats that fit your needs...

Start reading on any device today!

(Ebook) Biota Grow 2C gather 2C cook by Loucas, Jason; Viles, James ISBN
9781459699816, 9781743365571, 9781925268492, 1459699815, 1743365578, 1925268497

https://ebooknice.com/product/biota-grow-2c-gather-2c-cook-6661374

ebooknice.com

(Ebook) Matematik 5000+ Kurs 2c Lärobok by Lena Alfredsson, Hans Heikne, Sanna
Bodemyr ISBN 9789127456600, 9127456609

https://ebooknice.com/product/matematik-5000-kurs-2c-larobok-23848312

ebooknice.com

(Ebook) SAT II Success MATH 1C and 2C 2002 (Peterson's SAT II Success) by Peterson's
ISBN 9780768906677, 0768906679

https://ebooknice.com/product/sat-ii-success-math-1c-and-2c-2002-peterson-s-sat-
ii-success-1722018

ebooknice.com

(Ebook) Master SAT II Math 1c and 2c 4th ed (Arco Master the SAT Subject Test: Math
Levels 1 & 2) by Arco ISBN 9780768923049, 0768923042

https://ebooknice.com/product/master-sat-ii-math-1c-and-2c-4th-ed-arco-master-
the-sat-subject-test-math-levels-1-2-2326094

ebooknice.com
(Ebook) Cambridge IGCSE and O Level History Workbook 2C - Depth Study: the United
States, 1919-41 2nd Edition by Benjamin Harrison ISBN 9781398375147, 9781398375048,
1398375144, 1398375047

https://ebooknice.com/product/cambridge-igcse-and-o-level-history-
workbook-2c-depth-study-the-united-states-1919-41-2nd-edition-53538044

ebooknice.com

(Ebook) Advances in Multimedia Information Processing – PCM 2017: 18th Pacific-Rim

Conference on Multimedia, Harbin, China, September 28-29, 2017, Revised Selected
Papers, Part II by Bing Zeng, Qingming Huang, Abdulmotaleb El Saddik, Hongliang Li,
Shuqiang Jiang, Xiaopeng Fan ISBN 9783319773797, 9783319773803, 3319773798,
3319773801
https://ebooknice.com/product/advances-in-multimedia-information-processing-
pcm-2017-18th-pacific-rim-conference-on-multimedia-harbin-china-
september-28-29-2017-revised-selected-papers-part-ii-7149594

ebooknice.com

(Ebook) Computer Science – Theory and Applications: 15th International Computer

Science Symposium in Russia, CSR 2020, Yekaterinburg, Russia, June 29 – July 3,
2020, Proceedings by Henning Fernau ISBN 9783030500252, 9783030500269, 303050025X,
3030500268
https://ebooknice.com/product/computer-science-theory-and-applications-15th-
international-computer-science-symposium-in-russia-csr-2020-yekaterinburg-
russia-june-29-july-3-2020-proceedings-22503900

ebooknice.com

(Ebook) Advances in Mathematical Modeling and Scientific Computing: International

Conference on Recent Developments in Mathematics, Dubai, 2022 – Volume 2 (Trends in
Mathematics) by Firuz Kamalov, R. Sivaraj, Ho Hon Leung ISBN 9783031414190,
3031414195
https://ebooknice.com/product/advances-in-mathematical-modeling-and-scientific-
computing-international-conference-on-recent-developments-in-mathematics-
dubai-2022-volume-2-trends-in-mathematics-55925214

ebooknice.com

(Ebook) Computer Science – Theory and Applications: 17th International Computer

Science Symposium in Russia, CSR 2022, Virtual Event, June 29 – July 1, 2022, ...
(Lecture Notes in Computer Science, 13296) by Alexander S. Kulikov (editor), Sofya
Raskhodnikova (editor) ISBN 9783031095733, 3031095731
https://ebooknice.com/product/computer-science-theory-and-applications-17th-
international-computer-science-symposium-in-russia-csr-2022-virtual-event-
june-29-july-1-2022-lecture-notes-in-computer-science-13296-43838102

ebooknice.com
Jihad Mohamad Alja’am
Abdulmotaleb El Saddik
Abdul Hamid Sadka Editors

Recent Trends
in Computer
Applications
Best Studies from the 2017
International Conference on Computer
and Applications, Dubai, UAE
Recent Trends in Computer Applications
Jihad Mohamad Alja’am •
Abdulmotaleb El Saddik • Abdul Hamid Sadka
Editors

Recent Trends in Computer

Applications
Best Studies from the 2017 International
Conference on Computer and Applications,
Dubai, UAE

123
Editors
Jihad Mohamad Alja’am Abdulmotaleb El Saddik
Computer Science and Engineering Faculty of Engineering
Qatar University University of Ottawa
Doha, Qatar Ottawa
Ontario, Canada

Abdul Hamid Sadka

Electronic and Computer Engineering
Brunel University London
Uxbridge, United Kingdom

ISBN 978-3-319-89913-8 ISBN 978-3-319-89914-5 (eBook)

https://doi.org/10.1007/978-3-319-89914-5

Library of Congress Control Number: 2018958782

© Springer International Publishing AG, part of Springer Nature 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Foreword

By the turn of the second millennium, it became clear that computers (and more
broadly intelligent machines) are becoming the focus of science and technology
for the next few decades to come. This book introduces the reader to the realm of
the most recent trends in the area of computer applications, with a special focus
on sustainable development, marking this important trend during the first decades
of the third millennium. The broad scope of the book is by design as the editors
and authors introduce a wide scope of application fields where modern computing
brought about several paradigm shifts in the way data is analysed, managed and
visualised.
Already a decade ago, the notion of Big Data was introduced, and since then
new scientific and technical challenges were formulated and efficient solutions
have been proposed. Big Data was later cast in the framework of decision-making
environments, in which theories, algorithms, methods and systems have been
developed to efficiently map data into decisions. Data-driven and data-intensive
computer applications have since been developed in a number of areas, including,
but not limited to, media (both audio and visual), healthcare, robotics, security,
web applications and web interfaces. Conceptually, data handling strategy can
conveniently be presented as a three-layered scheme, in which the first layer
interfaces with raw data (computer-generated, time series, sensor data, etc.) and
offers various ways to represent, clean, abstract and possibly augment the source
data. The second layer hosts methods and algorithms for analytics, management
and visualisation of the processed data. And finally, the third layer links the results
of the second layer to a specific application, that is, it interfaces with the real-
world application domain. Most of the contributions in this book cover one or
more of these layers targeting a specific application domain. As the amount of
data keeps increasing exponentially and the demand of split (real-time) decision
is becoming more imminent, the key challenges we are facing today include
scalability, efficiency and real-time performance.
The book is recommended to the tech-savvy managers as well as engineers,
technicians and researchers in various fields of computer applications. It is rather
seldom we come across a reference where such an overwhelming amount of

v
vi Foreword

information regarding diverse application fields has been gathered in a single

volume. The book requires common mathematical and computer science knowledge
acquired by most university college degrees and, therefore, it is easy to read and
grasp the main concepts which are well illustrated throughout the chapters of the
book.

NSF I/UCRC Center for Visual Moncef Gabbouj

and Decision Informatics, TUT-Site
Laboratory of Signal Processing
Tampere University of Technology Tampere, Finland
Preface

This book consists of an agglomeration of know-how and recent research findings

imparted by a broad range of international scientists within the field of information
and communication technologies. The book recognises computers potentially as
data-generating machines and computer applications as platforms for the acqui-
sition, analysis, processing, management and visualisation of data in its multiple
forms, scales (i.e. volumes), complexities and digital representations. While the
editors realise the breadth and diversity of computer applications, and hence the
corresponding data-handling strategies employed therein, the essence of the book
focuses mainly on three key data-driven classes of technologies, namely data
analytics, data management and data visualisation.
In data analytics and processing, the book presents an authoritative set of chapters
addressing various challenges commonly encountered in computer-based applica-
tions and systems, such as segmentation, detection, classification, recognition, etc.,
in the light of vision-based but also multimodal scenarios. In image segmentation,
for instance, one chapter addresses the unsupervised segmentation of images using
graph-based community detection. In particular, an overview of sequential mining
algorithms and their extensions is presented in one chapter, while image/video
classification is addressed using multimodal techniques in another chapter and using
Gabor filters in yet another. An analysis of the current and future directions of
object detection based on convolutional neural networks is also featured. Hand
detection and gesture recognition within the context of human–computer interaction
is addressed in a chapter that focuses on translating recognised hand gestures into
functional ones to enable the real-time manipulation of a 2D image. The book also
looks into the compression aspects of computer applications with multiview video
codecs in perspective. In particular, one chapter addresses the multiview extensions
of two contemporary video coding standards and provides a comparative analysis
of their performance in terms of quality and compression efficiency.
In data management, the authors’ contributions place a particular emphasis
on the security aspects of data in networked computer applications which utilise

vii
viii Preface

cloud computing technologies. One chapter looks at utilising a combination of data

encryption algorithms and a distributed system to improve data confidentiality for an
acceptable overhead performance. Another chapter considers the malware detection
algorithms and argues the complexity and computationally intensive nature of the
process of identifying malicious codes in files or network traffic. It goes further to
propose a novel hybrid solution that leverages the CPU/GPU computing capabilities
for improving the performance and reducing the power consumption of string
matching algorithms on devices such as laptops for instance. Furthermore, the
security aspects of Software-Defined Networking (SDN) are examined from the
standpoint of Distributed Denial of Service (DDoS) attacks. The chapter presents
a controller placement model that helps reduce the impact of DDoS attacks and
hence make SDN more secure and resilient. One interesting computer application
considered in the book is the control of a permanent-magnet DC motor without the
prior knowledge of its parameters. Another chapter dedicates specific attention to
designing a new hardware/software platform that enables the real-time provision of
all the parameters required for the control of the DC motor.
In data access and visualisation, the book presents a series of chapters that are
concerned with providing a user-centric approach to multimedia applications and
services, particularly in e-commerce, future Tactile Internet, search and retrieval as
well as language recognition scenarios. One chapter reviews the state of the art in
user-centric multimodal systems and presents a vision towards the realisation of an
immersive, interactive and collaborative framework for the Internet of Multimodal
Things (IoMT) system. Another chapter considers an automated approach to the
optimisation of Web interfaces for e-commerce. The chapter emphasises primarily
the vital role of User Experience (UX) and Customer Experience (CX) principles
in the provision of any web-based service, application or product. The book
embodies a chapter that features the design, development and evaluation of a web-
based Arabic multimedia search engine that is based on a language transcriber.
In order to enable an efficient and user-friendly human–computer interaction, a
chapter focuses on the review and analysis of specific text-to-picture systems and
approaches to facilitate education. Last but not least, one of the chapters explores
both person-dependent and person-independent Arabic speech recognition systems
and examines how hidden Markov models can be specifically exploited for the
recognition of Arabic, rather than English, words.
This book offers the readers with the unique dual benefit of gaining a meticulous
analysis of current technology trends in computer applications and simultaneously
benefiting from a rich display of recent experimental research findings in a rather
diverse and prolific technological field. While the book is inherently diverse in
its scope and coverage, addressing a broad spectrum of technologies exploited by
computer applications and systems today, the book editors are confident that this
manuscript will put at the disposal of their audience, from both academia and indus-
trial R&D sectors, a useful resource that will not only help expand the beneficiaries’
knowledge base in the relevant fields but will also offer them a supportive guide that
Preface ix

is equipped with a sufficient level of scientific originality, depth and rigour into a
cluster of technological trends and most recent research developments in multimedia
data handling and manipulation, with computer applications in perspective.

Jihad Mohamad Alja’am

Abdulmotaleb El Saddik
Abdul Hamid Sadka
Acknowledgements

This book would not have been possible without the hard work of all the authors.
I am especially indebted to Prof. Abdul Hamid Sadka and Prof. Abdulmotaleb El
Saddik who worked hard to select the best submitted chapters to be included in
this book. I am grateful to all reviewers who spent a lot of time in the peer-review
process to improve the quality of the accepted chapters. Each reviewer has provided
excellent and professional scientific guidance for the authors to refine their works
to meet the scientific standard. I am indebted to Dr. Nabil Khelifi (Senior Editor—
Springer) who trusted me to lead this project successfully. Many thanks also for
Reyhaneh Majidi (Springer DE) who has always been very supportive. Finally, I
really appreciate the hard work of my assistant Dana Bandock who allocated an
important part of her time in the management for the successful completion of this
book.

Department of Computer Science and Engineering Jihad Mohamad Alja’am

College of Engineering
Qatar University
Doha, Qatar

xi
Visit https://ebooknice.com to
discover a wide range of
eBooks across various genres.
Enjoy exclusive deals and
discounts to enhance your
reading experience. Start your
digital reading journey today!
Contents

Part I Data Analytics and Processing (Including Classification,

Compression, Segmentation, Mining, Detection and
Recognition etc.)
Overview on Sequential Mining Algorithms and Their Extensions . . . . . . . . 3
Carine Bou Rjeily, Georges Badr, Amir Hajjam Al Hassani,
and Emmanuel Andres
Object Detection Based on CNNs: Current and Future Directions . . . . . . . . 17
Long Chen, Abdul Hamid Sadka, Junyu Dong, and Huiyu Zhou
Video Classification Methods: Multimodal Techniques. .. . . . . . . . . . . . . . . . . . . . 33
Amal Dandashi and Jihad Mohamad Alja’am
Proposed Multi-label Image Classification Method Based on Gabor
Filter .. . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 53
Ziad Abdallah, Ali El-Zaart, and Mohamad Oueidat
Vision-Based Approach for Real-Time Hand Detection and Gesture
Recognition .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 67
Rayane El Sibai, Chady Abou Jaoude, and Jacques Demerjian
Unsupervised Image Segmentation via Graph-Based Community
Detection . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 83
Abdelmalik Moujahid, Fadi Dornaika, and Blanca Cases
Multiview Video Coding: A Comparative Study Between MVC
and MV-HEVC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 99
Seif Allah El Mesloul Nasri, Abdul Hamid Sadka, Noureddine Doghmane,
and Khaled Khelil

xiii
xiv Contents

Part II Data Handling and Management (Including Data Security,

Database Handling, Cloud Computing, Hardware and
Software Technologies)
Data Fragmentation Scheme: Improving Database Security in
Cloud Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 115
Amjad Alsirhani, Peter Bodorik, and Srinivas Sampalli
CPU/GPU Hybrid Detection for Malware Signatures for
Battery-Powered Devices Using OpenCL . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 139
Radu Velea, Ştefan Drăgan, and Florina Gurzău
Complete Design of a Hardware and Software Framework for
PWM/Discrete PID-Based Speed Control of a Permanent-Magnet
DC Motor Without Prior Knowledge of the Motor’s Parameters .. . . . . . . . . 153
Chady El Moucary, Abdallah Kassem, Walid Zakhem, Chaybane Ghabach,
Roger El Khoury, and Patrick Rizk
Analysis of DDoS Attack-Aware Software-Defined Networking
Controller Placement in Malaysia .. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 175
Muhammad Reazul Haque, Saw Chin Tan, Ching Kwang Lee,
Zulfadzli Yusoff, Sameer Ali, Ir. Rizaludin Kaspin,
and Salvatore Renato Ziri

Part III Data Access and Visualisation: Tools

and Platforms/Systems
Multimodal Systems, Experiences, and Communications: A Review
Toward the Tactile Internet Vision . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 191
Mohammad Al Jaafreh, Majed Alowaidi, Hussein Al Osman,
and Abdulmotaleb El Saddik
Automating the Optimization of Web Interfaces for E-Commerce .. . . . . . . . 221
Aoun Lutfi and Stefano Fasciani
Arabic Multimedia Search Platform .. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 237
Mohamad Raad, Majida Bayan, Yihya Dalloul, Majd Ghareeb,
and Amin Haj-Ali
MOALEM: An Assistive Platform for Children with Difficulties
in Reading and Writing Arabic.. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 251
Jihad Mohamad Alja’am, Moutaz Saleh, Dominic Massaro,
and Mohamad Eid
Person-Dependent and Person-Independent Arabic
Speech Recognition System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 267
Noor Al-Maadeed and Somaya Al-Maadeed
Information Visualization Techniques for Building Better
Visualization Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 279
Rachael Fernandez and Noora Fetais
 Part I
Data Analytics and Processing (Including
Classification, Compression, Segmentation,
Mining, Detection and Recognition etc.)
Overview on Sequential Mining
Algorithms and Their Extensions

Carine Bou Rjeily, Georges Badr, Amir Hajjam Al Hassani,

and Emmanuel Andres

1 Introduction

Interesting sequential patterns (SPs) in a sequence database are extracted using

Sequential Pattern Mining algorithms. These patterns help in analyzing data and
obtaining interesting and valuable knowledge from large amounts of data. Other
techniques including Sequence Prediction and Sequential Rule Mining are also used
nowadays for decision-making purposes. The main idea is to extract frequent subse-
quences, called patterns, from a massive amount of collected data and understand the
relation(s) between these patterns. Many sectors are interested in these techniques.
For example, analyzing customers’ purchases to improve marketing strategy: Let’s
say a customer buys a camera and a lens. The next time he comes, he buys a tripod.
That information could be used to predict customers’ needs by understanding their
interests. The company may then offer a tripod or a discount when buying a camera
and a lens. Nowadays, Sequential Pattern Mining algorithms play an important
role in the medical domain, for the notion of time is important in analyzing data
related to patients or hospitals. Novel applications are based on sequential mining
for decision-making in the medical field, such as in [1–4]. Sequence Prediction was
also used to predict heart failure in [5, 6].

C. Bou Rjeily · A. H. Al Hassani

Nanomedicine Lab, Université de Bourgogne Franche – Comté, Belfort, France
e-mail: [email protected]; [email protected]
G. Badr ()
TICKET Lab, Antonine University, Baabda, Lebanon
e-mail: [email protected]
E. Andres
Université de Strasbourg, Centre Hospitalier Universitaire, Strasbourg, France
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 3

J. M. Alja’am et al. (eds.), Recent Trends in Computer Applications,
https://doi.org/10.1007/978-3-319-89914-5_1
4 C. Bou Rjeily et al.

The first part of this chapter defines important terms and notations in the
field. The second shows a survey on the most important and recent sequential
mining algorithms according to a clear classification. Lastly, the chapter concludes
with a classification tree showing the main categories of the algorithms and their
extensions. It is important to know that this chapter provides the essential definitions
and functionalities of the algorithms. Knowing the appropriate outputs of the
algorithms will help the user in choosing the most efficient one for his/her studies.

2 Important Terms and Notations

Before presenting the algorithms and their classification, it is important to define

some basic terms used in Sequential Pattern Mining in order to understand the
mining process. These terms are commonly used in data mining processes and
especially in Sequential Pattern Mining.
1. An item is an entity that can have multiple attributes: date, size, color, and so on.
2. I = {i1 , . . . , in } is a nonempty set of items. A k-itemset is an itemset with k items.
3. A sequence “S” is an ordered list of itemsets. An itemset Xy in a sequence, with
1 ≤ y ≤ L, is called a transaction. L denotes the length of the sequence, which
refers to the number of its transactions. S = {(a,b); (b,c); (e,d)}, which means
that the items a and b are occurring together in the same time, while the items
b and c are occurring together although in the same time but after a and b occur
together and so on.
4. A sequential database (SDB) is a list of sequences with a sequence ID (SID) (cf.
Table 1).
5. A sequence β can have a subsequence α, making β a super-sequence of α.
6. A sequential rule r, denoted X → Y, is a relationship between two unordered
itemsets X, Y ⊆ I, where X ∩ Y = ∅. X → Y means that if items of X appear in
a sequence, items of Y will also occur in the same sequence.
7. The support of a rule r in a sequence database SDB is defined as the number
of sequences that contains X ∪ Y divided by the number of sequences in the
database:

|{s; s ∈ SDB ∧ r ∧ s}|

supSDB(r) =
|SDB|

Table 1 A sequence SID Sequence

database
1 {a, b}, {c}, { f, g}, {g}, {e}
2 {a, d}, {c}, {b}, {a, b, e, f }
3 {a}, {b}, { f }, {e}
4 {b}, { f, g}
Overview on Sequential Mining Algorithms and Their Extensions 5

8. The confidence of a rule r in a sequence database SDB is defined as the number

of sequences that contains , divided by the number of sequences that contains X:

|{s; s ∈ SDB ∧ r ∨ s}|

conf SDB(r) =
|SDB|

9. A rule r is a frequent sequential rule iff supSDB(r) ≥ minsup, with min-

sup ∈ [0, 1] being a threshold set by the user.
10. A rule r is a valid sequential rule iff it is frequent and confSDB(r) ≥ minconf,
with minconf ∈ [0, 1] being a threshold set by the user.
11. Apriori-based [7]: many mining algorithms are based on this technique. The
main idea is to create a list of the most frequent items with respect to minsup
and minconf. The list is increased progressively considering the support and the
confidence.
12. Sequential Rule Mining is to find all frequent and valid sequential rules in an
SDB [8].
13. Pattern Growth [9] is a method for extracting frequent sequences by partitioning
the search space and then saving the frequent itemsets using a tree structure.
Extraction is done by concatenating to the processed sequence (called prefix
sequence) frequent items with respect to its prefix sequence. This method
can be seen as depth-first traversal algorithm and eliminates the necessity to
repetitively scan all of the SDB.
14. Searching processes:
– Depth-First Search (DFS) is a searching process that traverses or searches
tree or graph data structures. A node in the graph or tree is considered as
the root where the search begins. In case of graph, some arbitrary nodes are
selected as the root and explored as far as possible along each branch before
backtracking.
– Breadth-First Search (BFS) is a searching process for searching in trees or
graph structures. It starts at the root like (DFS) and explores the neighbor
nodes first, before exploring the next-level neighbors.
Let β = β 1 . . . β n and α = α 1 . . . α m be two sequences where m ≤ n.
15. Sequence α is called the prefix of β iff ∀i ∈ [1 . . . m], α i = β i .
16. Sequence β = β 1 . . . β n is called the projection of some sequence S with
regards to α, iif:
– β s
– α is a prefix of β
– There exists no proper super-sequence β  of β such that β   s and β  also
has a prefix
17. Sequence γ = β m+1 . . . β n is called the suffix of s with regard to α. β is
then the concatenation of α and γ .
Let SDB be a sequence database.
6 C. Bou Rjeily et al.

Table 2 A vertical database for the sequence database of Table 1

A B C D E
SID Itemsets SID Itemsets SID Itemsets SID Itemsets SID Itemsets
1 1 1 1 1 2 1 1 5
2 1,4 2 3,4 2 2 2 1 2 4
3 1 3 2 3 3 3 4
4 4 1 4 4 4
F G
SID Itemsets SID Itemsets
1 3 1 3,4
2 4 2
3 3 3
4 2 4 2

Table 3 Projected database a : projected database

with regards to prefix “a”
{_, b}, {c}, { f, g}, {g}, {e}
{_, d}, {c}, {b}, {a, b, e, f }
{b}, { f }, {e}

18. Horizontal database: each entry in a horizontal database is a sequence as shown

in Table 1.
19. Vertical database: each entry represents an item and indicates the list of
sequences where the item appears and the position(s) where it appears [10]
(cf. Table 2).

20. Projected database: the α-projected database, denoted by SDB|α , is the collec-
tion of suffixes of sequences in SDB with regard to prefix α. Table 3 shows an
example of the projected database considering “a” as prefix.

3 Sequential Mining Algorithms

There exist many data mining techniques such as classification, clustering, associa-
tion rule mining and others. This chapter focuses on sequential mining algorithms.
We present the state of the art of recent algorithms, elaborating a classification
based on their main objectives and principles. Thus, this classification divides
the algorithms into three primary types: SP Mining, Sequential Rule Mining and
Sequence Prediction. Each of these can be split into different criteria and strategies.
Overview on Sequential Mining Algorithms and Their Extensions 7

3.1 Sequential Pattern Mining

3.1.1 Frequent Sequential Pattern Mining

It consists of finding subsequences appearing frequently in a set of sequences called

sequential pattern or frequent subsequence. The frequency of these patterns is no
less than a minimum support threshold minsup specified by the user. The common
frequent Sequential Pattern Mining algorithms are:
The Generalized Sequential Patterns (GSP) algorithm [11] is an Apriori-like
method and was one of the first algorithms that studied SPs after Apriori-All. The
database is scanned multiple times. The first pass determines the support of each
item, which is the number of data sequences that include the item. It simply means
counting the occurrences of singleton transactions (containing one element) in the
given database (one scan of the whole database). After this process, nonfrequent
items are removed, and each transaction consists now of its original frequent items.
This result will be the input of the GSP algorithm. Like Apriori, GSP algorithm
makes multiple database scans. At the first pass, all single items of length 1
sequences (1-sequences) are counted. At the second pass, frequent 1-sequences
are used to define the sets of candidate 2-sequences, and another scan is made to
calculate their support. Same process is used to discover the candidate 3-sequences
but using frequent 2-sequences, and so on until no more frequent sequences are
found. GSP algorithm is composed of two techniques:
1. Candidate Generation: Only candidates with minimum support or above are
conserved until no new candidates are found. This technique generates an
enormous number of candidate sequences and then tests each one with respect of
the user-defined minsup.
After the first scan of the database and obtaining frequent (k−1)-frequent
sequences F(k−1), a joining procedure of F(k−1) with itself is made and any
infrequent sequence is pruned if at least one of its subsequences is not frequent.
2. Support Counting: a hash tree-based search is used. Finally nonmaximal frequent
sequences are removed.
The GSP algorithm also allows frequent sequences discovery with time con-
straints. It can calculate the difference between the end-time of the element just
found and the start-time of the previous element. This time is user defined and
called maximum and minimum gap. Furthermore, it supports the concept of a sliding
window (defines the interval of time between items in the same transaction).
The Sequential PAttern Discovery using Equivalence classes, SPADE [10] is
based on a vertical id-list database format in which each sequence is associated
to a list of items in which it appears: each subsequence is originally associated to its
occurrence list. The frequent sequences can be found by using the intersection on
id-lists. The size of the id-lists is the number of sequences in which an item appears.
SPADE reduces the search space by aggregating SPs into equivalent classes and
8 C. Bou Rjeily et al.

thus reduces the execution time. Thereby, two k-length sequences are in the same
equivalence class if they share the same k-1 length prefix.
In his first step, SPADE computes the support of length 1 sequences, and this
is done in a single database scan. In its second step, SPADE computes the support
of 2-sequences and this is done by transforming the vertical representation into a
horizontal representation in memory. This counting process is done with one scan of
data and uses a bi-dimensional matrix. The idea consists of joining (n−1) sequences
using their id-lists to obtain n-subsequences. If the size of id-list is greater than
minsup, then the sequence is frequent. The algorithm can use a breadth-first or a
depth-first search method for finding new sequences. The algorithm stops when no
more frequent sequences are found.
Sequential PAttern Mining, SPAM [12], is a memory-based algorithm and uses
vector of bytes (bitmap representation) to study the existence (1) or absence (0) of
an item in a sequence after loading the database into the memory. Candidates are
generated in a tree by an S-extension that adds an item in another transaction, and
by an I-extension that appends the item in the same transaction. The candidates are
verified by counting the bytes with a value of one with the defined minsup.
The algorithm is efficient for mining long sequential patterns. Depth-first search
is used to generate candidate sequences, and various I-step pruning and s-Step
pruning are used to reduce the search space.
The transactional data are stored using a vertical bitmap representation, which
allows for efficient support counting as well as significant bitmap compression. One
new feature introduced with SPAM is that it incrementally outputs new frequent
itemsets in an online fashion.
The Prefix-projected Sequential Pattern Mining, known as PrefixSpan [13], is
a pattern-growth-based algorithm that discovers SPs using the idea of projected
database. The algorithm studies the prefix subsequences instead of exploring all
the possible occurrences of frequent subsequences (refer to the definitions 16 and
17). Then, it performs a projection on their corresponding post-fix subsequences.
Frequent sequences will grow by mining only local frequent patterns, showing the
efficiency of this algorithm.
The Last Position Induction algorithm (LAPIN) [14] is used for the extraction
of long sequences and the reduction of the search space. It uses a lexicographical
tree as the search path with DFS strategy. LAPIN-LCI procedure tests each item in
the local candidate list and directly decides whether the item can be added to the
prefix sequence or not. It compares the item’s last position with the prefix border
position. The algorithm assumes that the last position of an item i is helpful to decide
whether this item could be appended to a frequent sequence of length k in order to
get a frequent sequence of k + 1 length.
The CM-SPAM and CM-SPADE [15] are extensions of the two well-known
algorithms SPADE and SPAM to which is added a new structure called Co-
Occurrence MAP (C-MAP). The latter is used to store co-occurrence information
by dividing them into CMAPi and CMAPs substructures. The first stores the items
that succeed each item by i-extension and the second stores the items that succeed
each item by s-extension at least minsup times. Let S be the sequence {I1 , I2 , . . . ,
Overview on Sequential Mining Algorithms and Their Extensions 9

In }. An item k is said to succeed by i-extension to an item j in S, iff j and k ∈ Ix

for an integer x such that 1 ≤ x ≤ n and k >lex j. An item k is said to succeed by
s-extension to an item j in S, iff j ∈ Iv and k ∈ Iw for some integers v and w such that
1 ≤ v < w ≤ n.
The i-extension of pattern P with an item x is considered nonfrequent if there
exists an item i in the last itemset of P such that (i,x) is not in CMAPi. Same for the
pruning of s-extension: The s-extension of a pattern P with an item x is infrequent
if there exists an item i in P such that (i, x) is not in CMAPs.

3.1.2 Closed Sequential Pattern Mining

A Closed Sequential Pattern (CSP) is not necessarily included in another pattern

having the same support. The set of CSPs is much smaller than the set of SPs making
mining more efficient. There exists no super-pattern S of pattern S having the same
support of S. Then S is a closed sequential pattern; in other words, Closed Pattern
Mining means that for the same support the mining process will mine the longest
pattern. Common Sequential Patterns algorithms are given in the following.
The CloSpan algorithm [16] is based on mining frequent closed sequences in
large data sets instead of exploring all frequent sequences and is used to mine long
sequences. Its main advantage is in time and space reduction. The algorithm is
divided into two stages. In the first, it generates a set of all frequent sequences
and eliminates the nonclosed sequences in the second. It represents data with
lexicographical tree or order.
A Lexicographic Sequence Tree (LST) can be constructed as follows:
1. Each node in the tree corresponds to a sequence, and the root is a null sequence.
2. If a parent node corresponds to a sequence S1 , its child is either an itemset-
extension of S1 , or a sequence-extension of S1 .
3. The left sibling is less than the right sibling in sequence lexicographic order.
The BI-Directional Extension (BIDE+) [17] is an extension of the BIDE
algorithm that mines closed SPs and avoids problem of the candidate maintenance-
and-test paradigm used by CloSpan. It works in a DFS manner in order to generate
the frequent closed patterns and consumes less memory compared to the previous
version.
The ClaSP [18] is based on the SPADE algorithm and was the first to mine closed
frequent SPs in vertical databases. ClaSP has two phases: The first one generates
a subset of frequent sequences called Frequent Closed Candidates (FCC), which
is kept in main memory; and the second step executes a post-pruning phase to
eliminate all nonclosed sequences from FCC to finally obtain exactly FCS.
CM-ClaSP [9] is an extension of ClaSP based on the new representation of data
called C-MAP as discussed in CM-SPADE and CM-SPAM.
Visit https://ebooknice.com to
discover a wide range of
eBooks across various genres.
Enjoy exclusive deals and
discounts to enhance your
reading experience. Start your
digital reading journey today!
10 C. Bou Rjeily et al.

3.1.3 Maximal Sequential Pattern Mining

Sequential Pattern Mining may return too many results, making it difficult for the
user to understand and analyze. Mining maximal SPs may be a solution. A Maximal
SP is a pattern that is not included in another pattern. Maximal Pattern Mining
algorithms are presented in the following.
The MaxSP [19] is inspired by the PrefixSpan algorithm. It is based on a pattern-
growth algorithm that aims to extract maximal SPs without maintaining candidates.
It has an integrated BIDE-like mechanism that checks if a pattern is maximal.
MaxSp reduces the redundancy in SPs that could be time consuming and requires a
lot of storage space.
The Vertical Maximal Sequence Patterns (VMSP) [20] is based on the SPAM
search procedure that generates the pattern and explores candidate patterns having
same prefix in a recursive manner. VMSP integrates three strategies: Efficient
Filtering of Nonmaximal Patterns (EFN), Forward Maximal Extension Checking
(FME) and Candidate Pruning by Co-Occurrence Map (CPC).

3.1.4 Compressing Sequential Pattern Mining

This kind of algorithm is used to reduce redundancy and thus to minimize the size
of mining results.
GoKrimp and SeqKrimp [21] are two compressing SPs mining algorithms, based
on the Krimp algorithm. They explore directly compressing patterns and avoid the
resource-consuming candidate generation. SeqKrimp uses a frequent closed SPs
mining algorithm to generate a set of candidate patterns. It gets the candidate pattern
set and returns a good subset of compressing patterns, then greedily calculates the
benefits of adding/extending a given pattern from the candidates. This procedure
is repeated until no more useful patterns can be added. GoKrimp uses the same
procedures but is an ameliorated version of SeqKrimp. It searches for a set of
sequential patterns that compresses the data most based on the minimum description
length principle; informally, the best model is the one that compresses the data
the most. What differentiates GoKrimp is that it is parameter free. Users are not
supposed to set a minimum support, which is a difficult decision in some cases.
A dependency test is provided to consider only related patterns to extend a given
pattern. This technique aims to avoid the excessive tests of all possible extensions
and makes the GoKrimp faster than SeqKrimp.

3.1.5 Top-K Sequential Pattern Mining

In SP mining algorithm, tuning the minsup parameter to get enough patterns is a

difficult and time-consuming task. To remedy this issue, Top-K Sequential Pattern
mining algorithms were implemented to return k SPs.
Overview on Sequential Mining Algorithms and Their Extensions 11

TSP (Top-K Closed Sequential Patterns) [22] uses the concept of pattern-growth
and projection-based SP mining of PrefixSpan algorithm, and then performs a multi-
pass mining to find and grow patterns. After closed pattern verification phase, the
algorithm applies the minimum length constraint verification, which reduces the
search space.
TKS (Top-K Sequential Patterns) [23] uses a vertical bitmap database repre-
sentation. It adapts the SPAM search procedure to explore the search space of
patterns to transform it to a Top-K algorithm. Then, TSK extends the most promising
patterns, meaning that it finds patterns with high support in an early stage and
discards infrequent items. Finally, the algorithm uses a PMAP (Precedence MAP)
data structure to prune the search space.

3.2 Sequential Rules Mining

3.2.1 Sequential Rules

Mining frequent patterns is not sufficient in decision-making. Sequential rules

mining and sequence prediction (next section) are necessary. A sequential rule
indicates that if some item(s) occur in a sequence, some other item(s) are likely to
occur afterward with a given confidence or probability. Common Sequential Rules
mining algorithms are presented in the following.
CMDeo [24] was first designed to explore rules in a single sequence. It explores
the search space in BFS and extracts all valid rules of size 1*1 respecting minimum
support and confidence. Similarly to Apriori, CMDeo generates a huge amount of
valid rules by applying a left and a right expansion.
RuleGrowth [25] explores sequential rules for several sequences and not only for
one. It is based on the pattern-growth approach in finding the sequential relations
that explores rules between two items and expands them left and right.
CMRules [8] is an alternative of CMDeo. It searches for the association rules
to reduce the search space, then it removes the rules that do not respect minimum
support and confidence. Therefore, it could be used to discover both sequential rules
and association rules at the same time.
The Equivalence class-based sequential Rule Miner (ERMiner) [26] algorithm
uses a vertical representation to avoid database projection. It mines the search
space through equivalence classes to generate rules with the same antecedent or
consequent.

3.2.2 Top-K Sequential Rules Mining

Specifying the number of sequential rules to be found may overcome the dif-
ficulty in fine-tuning sequential rules parameters like minsup and minconf. The
idea of discovering Top-K nonredundant rules comes after the difficulty and the
12 C. Bou Rjeily et al.

time-consuming task to tune the minimum support value by the user. Moreover, the
sequential rule mining algorithms usually return a high level of redundancy. To solve
both problems, the Top-K Sequential Rules Mining algorithms let the user indicate
k, which is the number of rules to be discovered.
The TopSeqRule [27] was the first to address the Top-K sequential rules mining.
It generates rules for several sequences based on the RuleGrowth search strategy
integrated with the general process for mining Top-K patterns. To optimize results,
it first generates the most promising rules and reduces the search space by increasing
minsup.
Top-K Nonredundant Sequential Rules TNS [28] is used to discover the Top-K
non-redundant sequential rules. It adopts the TopSeqRule to mine the Top-K rules
and adapts it to eliminate redundancy. The algorithm gives an approximation and
thus does not guarantee to retrieve the Top-K nonredundant rules. TNS has a positive
integer parameter called delta to increase the result’s exactitude. Results are more
exact with a higher value of delta.

3.2.3 Sequential Rules with Window Size Constraints

This kind of algorithm returns all sequential rules with regard to the specified minsup
and minconf appearing within a window size.
TRuleGrowth [29] is an extension of the RuleGrowth with a sliding window
constraint. It is very useful in the discovery of temporal patterns (patterns that
happen within a maximum time interval). TRuleGrowth allows the user to specify
other optional parameters like the minimum antecedent length and the maximum
consequent length. These parameters define respectively the minimum number of
items appearing in the left side and the maximum number of items in the right
side of a rule, knowing that the left side is the antecedent and the right side is the
consequent.

3.2.4 Sequence Prediction

In many applications, it is very important to predict the next element in a sequence.

Given a set of sequences, the idea is to predict the next element in a sequence
S, based on a set of training sequences. Various applications use the sequence
prediction algorithms. For example, one may need to know the next web page to
be visited by the user, based on his/her, and/or other users’ histories.
The Compact Prediction Tree (CPT) [30] is a lossless sequence prediction model
that uses all information in the sequence for prediction. It consists of two phases:
the training phase and the prediction phase. The first compresses the sequences in a
prediction tree. A given sequence S is predicted by finding all sequences that contain
Overview on Sequential Mining Algorithms and Their Extensions 13

the last x items from S in any order and in any position. CPT is more efficient
than other existent algorithms such as Prediction by Partial Matching (PPM) [31],
Dependency Graph (DG) [32] and All-K-th-Order Markov [33].
CPT+ [34] is an enhanced version of CPT where Frequent Subsequence Com-
pression (FSC), Simple Branch Compression (SBC) and Prediction with improved
Noise Reduction (PNR) strategies were added to improve prediction time and
precision.

4 Conclusion

This chapter summarizes the most recent and common algorithms on the sequential
mining paradigm. It does not aim to give a deep explanation about each algorithm,
but it mentions its purpose and gives an idea about how it works. One should
refer to the related article of each algorithm for additional details. For further
explanation and ease of understanding, this chapter also presents a classification
for the sequential mining algorithms. They are arranged by their usage. This
classification was based on three main axes: frequent sequential pattern mining,
sequential rules mining and sequence prediction. Important terms and notations in
the data mining domain were first introduced. Then, a short definition introduced
each class to let the reader have a quick idea about it. Later, the most important and
recent algorithms in each axis were investigated with a brief description about their
methods and implementations.

5 Discussion

The diagram in Fig. 1 consists of a classification tree containing the most recent
algorithms and their extensions. This tree can help researchers in choosing the
appropriate algorithm according to their needs especially when it comes to sequen-
tial pattern mining. Sequential mining is efficient for applications that are time-
based or take into consideration the order of the event. Sequential mining has proven
its efficiency through time in the economic field starting from GSP that analyzes the
transactions of customers in order to improve the income and marketing strategies.
Sequential Mining started showing its importance in medical field, making it a very
promising field for researchers and programmers.
 14

Sequential Mining
algorithms in a SDB

Sequential
Pattern Sequential Sequence
Mining Rules Mining Predication

Frequent Closed Maximal Compressing Top-K Top-K Based on a set

sequential sequential sequential sequential sequential Sequential sequential Sequential Rules of training
patterns patterns patterns patterns patterns Rules Rules with window sequences
size constraint

CM-SPADE CM-ClaSP SeqKrimp ERMiner

VMSP [2014] TKS [2013] TRuleGrowth CPT+ [2015]
[2014] [2014] [2014] [2014] TNS [2013]
[2012]

CM-SPADE GoKrimp CMRules

ClaSP[2013] MaxSP [2013] TSP [2003] TopSeqRules CPT [2013]
[2014] [2012] [2012]
[2011]

Rule Growth
LAPIN [2007] Bide+ [2007] AKOM [1999]
[2011]

Dependency
PrefixSpan Graph (DG)
CloSpan [2003] CMDeo [2005]
[2001] [1996]

RuleGen
SPAM [2002] TDAG [1994]
[2001]

SPADE [2001]

GSP [1996]

Fig. 1 Classification of sequential mining algorithms

C. Bou Rjeily et al.
Overview on Sequential Mining Algorithms and Their Extensions 15

References

1. A.P. Wright, A.T. Wright, A.B. McCoy and D.F. Sittig, “The use of sequential pattern mining to
predict next prescribed medications”. Journal of biomedical informatics, 53, pp.73–80, 2015.
2. G. Bruno and P. Garza, “Temporal pattern mining for medical applications”. Data Mining:
Foundations and Intelligent Paradigms, pp.9–18. 2012
3. K. Uragaki, T. Hosaka, Y. Arahori, M. Kushima, T. Yamazaki, K. Araki and H. Yokota,
“Sequential pattern mining on electronic medical records with handling time intervals and the
efficacy of medicines”. In IEEE Symposium on Computers and Communication (ISCC), (pp.
20–25). IEEE. 2016.
4. K. Choi, S. Chung, H. Rhee and Y. Suh, Classification and sequential pattern analysis for
improving managerial efficiency and providing better medical service in public healthcare
centers. Healthcare informatics research, 16(2), pp.67–76, 2010.
5. C. Bou Rjeily, G. Badr, A. Hajjam El Hassani and E. Andres, “Sequence Prediction Algorithm
for Heart Failure Prediction”, International Conference e-Health, ISBN: 978-989-8533-65-4,
pp.109–116, 2017.
6. C. Bou Rjeily, G. Badr, A. Hajjam El Hassani and E. Andres, “Predicting Heart Failure Class
using a Sequence Prediction Algorithm”, Fourth International Conference on Advances in
Biomedical Engineering (ICABME), 2017
7. R. Agrawal, T. Imieliński, and A. Swami, “Mining association rules between sets of items in
large databases”, In ACM sigmod record(Vol. 22, No. 2, pp. 207–216). ACM, 1993.
8. P. Fournier-Viger, U. Faghihi, R. Nkambou, E. Mephu Nguifo, “CMRules: Mining Sequential
Rules Common to Several Sequences. Knowledge-based Systems”, Elsevier, 25(1): 63–76,
2012.
9. J. Han, J. Pei, Y. Yin, and R. Mao, “Mining frequent patterns without candidate generation: A
frequent-pattern tree approach”, Data mining and knowledge discovery, 8(1), pp.53–87, 2000.
10. M.J. Zaki, “SPADE: An efficient algorithm for mining frequent sequences”, Machine learning,
42(1–2), pp.31–60, 2001.
11. R. Srikant, and R. Agrawal, “Mining sequential patterns: Generalizations and performance
improvements”, In International Conference on Extending Database Technology (pp.1–17).
Springer Berlin Heidelberg, 1996.
12. J. Ayres, J. Flannick, J. Gehrke, J. and T. Yiu, “Sequential pattern mining using a bitmap
representation”, In Proceedings of the eighth ACM SIGKDD international conference on
Knowledge discovery and data mining(pp. 429–435). ACM, 2002.
13. J. Han, J. Pei, B. Mortazavi-Asl, H. Pinto, Q. Chen, U. Dayal, and M.C. Hsu, “Prefixspan:
Mining sequential patterns efficiently by prefix-projected pattern growth”, In proceedings of
the 17th international conference on data engineering, pp. 215–224, 2001.
14. Z. Yang, Y. Wang, and M. Kitsuregawa, M., “LAPIN: effective sequential pattern mining
algorithms by last position induction for dense databases”, In International Conference on
Database systems for advanced applications (pp. 1020–1023). Springer Berlin Heidelberg,
2007.
15. P. Fournier-Viger, A. Gomariz, M. Campos, and R. Thomas, “Fast vertical mining of sequential
patterns using co-occurrence information”, In Pacific-Asia Conference on Knowledge Discov-
ery and Data Mining (pp. 40–52). Springer International Publishing, 2014.
16. X. Yan, J. Han, R. Afshar R., “CloSpan: Mining Closed Sequential Patterns in Large Datasets”,
Proceedings of the 2003 SIAM International Conference on Data Mining, 2003.
17. J. Wang, and J. Han, “BIDE: Efficient mining of frequent closed sequences”, In Data
Engineering, 2004. Proceedings. 20th International Conference on (pp. 79-90). IEEE, 2004.
18. A. Gomariz, M. Campos, R. Marin, and B. Goethals, “Clasp: An efficient algorithm for mining
frequent closed sequences” In Pacific-Asia Conference on Knowledge Discovery and Data
Mining (pp. 50-61). Springer Berlin Heidelberg, 2013.
16 C. Bou Rjeily et al.

19. P. Fournier-Viger, C.W. Wu, and V.S. Tseng, “Mining maximal sequential patterns without can-
didate maintenance”, In International Conference on Advanced Data Mining and Applications
(pp. 169-180). Springer Berlin Heidelberg, 2013.
20. P. Fournier-Viger, C.W. Wu, A. Gomariz, and V.S. Tseng, “VMSP: Efficient vertical mining of
maximal sequential patterns”, In Canadian Conference on Artificial Intelligence (pp. 83-94).
Springer International Publishing, 2014.
21. H.T. Lam, F. Mörchen, D. Fradkin, and T. Calders, “Mining compressing sequential patterns”,
Statistical Analysis and Data Mining, 7(1), pp.34-52, 2014.
22. P. Tzvetkov, X. Yan, and J. Han, “TSP: Mining Top-k Closed Sequential Patterns”, Knowledge
and Information Systems, vol. 7, no. 4, pp. 438-457, 2005.
23. P. Fournier-Viger, A. Gomariz, T. Gueniche, E. Mwamikazi, and R. Thomas, “TKS: efficient
mining of top-k sequential patterns”, In International Conference on Advanced Data Mining
and Applications (pp. 109-120). Springer Berlin Heidelberg, 2013.
24. J. Deogun, and L. Jiang, “Prediction mining–an approach to mining association rules for
prediction”, In International Workshop on Rough Sets, Fuzzy Sets, Data Mining, and Granular-
Soft Computing (pp. 98-108). Springer Berlin Heidelberg, 2005.
25. P. Fournier-Viger, R. Nkambou, and V.S.M. Tseng, “RuleGrowth: mining sequential rules com-
mon to several sequences by pattern-growth”, In Proceedings of the 2011 ACM symposium on
applied computing (pp. 956-961), 2011.
26. P. Fournier-Viger, T. Gueniche, S. Zida, and V.S. Tseng, “ERMiner: sequential rule mining
using equivalence classes”, In International Symposium on Intelligent Data Analysis (pp. 108-
119). Springer International Publishing, 2014.
27. P. Fournier-Viger, and V.S. Tseng, “Mining top-k sequential rules”, In International Conference
on Advanced Data Mining and Applications (pp. 180-194). Springer Berlin Heidelberg, 2011.
28. P. Fournier-Viger, and V. S Tseng, “TNS: mining top-k non-redundant sequential rules”, In
Proceedings of the 28th Annual ACM Symposium on Applied Computing, 2013.
29. P. Fournier-Viger, C.W. Wu, V.S. Tseng, and R. Nkambou, “Mining sequential rules common
to several sequences with the window size constraint”, In Canadian Conference on Artificial
Intelligence (pp. 299-304). Springer Berlin Heidelberg, 2012.
30. T. Gueniche, P. Fournier-Viger, and V.S. Tseng, “Compact prediction tree: A lossless model
for accurate sequence prediction”, In International Conference on Advanced Data Mining and
Applications (pp. 177-188). Springer Berlin Heidelberg, 2013.
31. J. Cleary, I. Witten, “Data compression using adaptive coding and partial string matching”,
IEEE Trans. on Inform. Theory, vol. 24, no. 4, pp. 413-421, 1984.
32. V. N, Padmanabhan, J.C. Mogul, “Using Prefetching to Improve World Wide Web Latency”,
Computer Communications, vol. 16, pp. 358-368, 1998.
33. J. Pitkow, P. Pirolli, “Mining longest repeating subsequence to predict world wide web surfing”,
In: USENIX Symposium on Internet Technologies and Systems, Boulder, CO, pp. 13-25,
1999.
34. T. Gueniche, P. Fournier-Viger, R. Raman, and V.S. Tseng, “CPT+: Decreasing the time/space
complexity of the Compact Prediction Tree”, In Pacific-Asia Conference on Knowledge
Discovery and Data Mining (pp. 625-636). Springer International Publishing, 2015.
Object Detection Based on CNNs:
Current and Future Directions

Long Chen, Abdul Hamid Sadka, Junyu Dong, and Huiyu Zhou

1 Introduction

The goal of object detection is to learn a visual model for concepts such as cars
and use this model to localize these concepts in an image. As shown in Fig. 1,
given an image, object detection aims at predicting the bounding box and the label
of each object from the defined classes in the image. This requires the ability to
robustly model invariants against illumination changes, deformations, occlusions
and other intra-class variations. Among a number of vision tasks, object detection
is one of the fastest moving areas due to its wide applications in surveillance [1, 2]
and autonomous driving [3, 4].

L. Chen · H. Zhou
School of Electronics, Electrical Engineering and Computer Science, Queen’s University Belfast,
Belfast, UK
e-mail: [email protected]; [email protected]
A. H. Sadka ()
Department of Electronic and Computer Engineering, Brunel University, London, UK
e-mail: [email protected]
J. Dong
Department of Computer Science and Technology, Ocean University of China, Qingdao, China
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 17

J. M. Alja’am et al. (eds.), Recent Trends in Computer Applications,
https://doi.org/10.1007/978-3-319-89914-5_2
18 L. Chen et al.

Fig. 1 Bounding boxes and labels with corresponding class probabilities predicted by detectors

2 From Handcrafted Features to Deep CNNs Methods

2.1 Handcrafted Features

Before deep CNNs, convolutional neural networks [5], were introduced, the
progress on various visual recognition tasks had been considerably based on the use
of handcrafted features, such as SIFT [6] and HOG [7]. Handcrafted features can
be broadly divided into three categories:
1. Interest Point Detection. These methods use certain criteria to select pixels, edges
and corners as well-defined local texture features. Among them, Sobel, Prewitt,
Roberts, Canny and LoG (Laplacian of Gaussian) are typical edge detection
operators [8–11], while Harris, FAST (Features from Accelerated Segment Test),
CSS (Curvature Scale Space) and DOG (Difference of Gaussian) are typical
corner detection operators [6, 12, 13]. Interest point detection methods usually
have a certain geometric invariance which can be found at a small computational
cost.
2. Methods based on local features. These methods mainly extract local features,
which are different from global features such as colour histograms, which are
ideal for dealing with partial occlusion of target objects. Commonly used local
features include Scale-Invariant Feature Transform (SIFT) [6], HOG (Histogram
of oriented gradient) [7], Haar-like [14] and Local Binary Pattern [15, 16]. Local
features are informative, unique, with strong invariance and distinguishability.
But the calculation is generally complicated, and local features are further
developed to have better representations in recent years.
3. Methods based on multi-feature combination. A combination of interest point
and local feature extraction methods can be used to handle the deficiency of
using a single feature to represent target objects. DPM (Deformable Part-based
Random documents with unrelated
content Scribd suggests to you:
 Nunc igitur sedeat sapiens et computet actus,
Quam prius adueniat iudicis illa dies.

Postquam de gaudiis et penis que bonis et

malis debentur tractauit, consulit vlterius quod
unusquisque ad bonos mores se conuertat, et
de hiis que negligenter omisit, absque
desperacione contritus indulgenciam a deo
confidenter imploret.

Cumque repentinum casum breuis hora

Capm. xx.
minatur,632
Dum tenuem flatum suscitat aura leuis,
Care, memento tui, quis sis, cur, vnde vel ad quid,
1110 Vel cuius factus condicionis eras;
Quod caro sit fragilis, fallax facilisque moueri,
Prona sit ad peius, pessima prompta sequi.
Spiritus hunc mundum spernat speretque futura,
Semper in auctoris fixus amore sui:
Quod caro spiritui subdatur eumque sequatur,
Spiritus auctori seruiat ipse suo;
Quod motus carnis moderetur, commemoranda
Est mors et pena mortis habenda malis.
Non poterit melius hominis caro viua domari,
1120 Quam quod mente gerat mortua qualis erit.
Fletibus assiduis, est dum data gracia flendi,
Penituisse iuuat estque salubre satis:
Nec deus ethereus hec crimina vendicat vlli,
Que confessa dolens non residiua facit.
Qui reus est igitur homo, penam temporis huius
Sustineat, donec diluat omne malum;
Vt sic purgatus, cum iudex venerit, illam
Effugiat penam, que sine fine manet:
Nam qui iussa dei non seruat et vltima vite
1130 Spectat, ad infernum cogitur ille trahi.
Scripture fallunt, aut certe noscere debes
 Quod redit ad veniam vix animalis homo;
Victus enim vicio vicii fit seruus, et in se
Non habet admissum soluere posse iugum:
Ergo perpes ei debetur pena necesse,
Qui sibi peccandi velle perhenne facit.
Parcere nempe deo proprium tamen et misereri
est,
Vnde, licet sero, te reuocare stude.
Figmentum nostrum nouit, set et ipse medetur
1140 Tandem contritum, qui petit eius opem.
Non te desperes, pius est deus, immo deumque
Qui negat esse pium, denegat esse deum:
Hic quasi fons viuus patet omnibus, et vacuari
Vt fons nescit aquis, hic pietate nequit.
Set quia spem nimiam presumpcio sepe fatigat,
Tu tibi spem pone sicut oportet agi:
Vt sapiens speres, tibi sit tua spes moderanda,
Eius habent sancto frena timore regi.
Non timor excedat, quo desperacio mentem
1150 Polluat, immo deum mentis amore time:
Nec spes presumat, set amet commixta timore,633
Sic timor est virtus spes et vterque salus.
Set meditando tamen tua mens de fine remorsa,
Semper amara timens speret habere bona:
Sanccius vt viuas, memorare nouissima semper,
Ledunt nam iacula visa perante minus.
Respice cotidie, mortis quia tempus adesse
Festinat, que simul prospera cuncta ruet.

Hic loquitur quod sunt modo pauci, qui aut

propter celi affectum aut gehenne metum huius
vite voluptatibus renunciant; set quecumque
caro concupiscit, omni postposita racione,
ardencius perficere conantur.
Capm. xxi. Qui sibi commemorans, puto, singula ponderat
eque,
1160 Senciet a fine gaudia vana fore:
Nunc tamen a viciis est quilibet infatuatus,
Quod de fine suo vix memoratur homo.
Quisque suum corpus colit, et de carnis amore
Gaudet, et est anime causa relicta sue:
Gloria nec celi mentes neque pena gehenne
A mundi labe iam reuocare queunt.
Sic caro, sic demon, sic mundus vbique modernos
Deuiat, vt Cristi vix sciat vnus iter:
Est caro que fragilis, demon versutus, iniqus
1170 Mundus, in hoc hominum tempore regna colunt:
Et sic bruta quasi perit humane racionis
Virtus, dum vicium corporis acta regit.
Est homo nunc animal dicam, set non racionis,
Dum viuit bruti condicione pari.
Nescia scripture brutum natura gubernat,
Iudicis arbitrium nec racionis habet:
Est igitur brutis homo peior, quando voluntas
Preter naturam sola gubernat eum.
Corporis, heu! virtus per singula membra
revoluens
1180 Naturam viciis seruit ad acta foris;
Ac anime racio carnis viciata vigore
De virtute nichil interiore sapit.
Morigeri cicius modo sunt derisio plebis,
Et scola peccati iustificabit opus:
Que solet illa viros veteri de more beare,
Iam noua virtuti frena libido mouet.634
Inter eos mundi quibus est donata potestas,
‘Sic volo, sic iubeo,’ sunt quasi iura modo.
Succumbunt iusti clamantes, ‘Ve! quod in orbe
1190 Impia pars hominum singula regna terit.’
Vis prohibet leges, euertunt crimina mores,
Virtus peccati turbine quassa perit:
Mundus turbatur, rerum confunditur ordo,
 Involuitque simul omnia grande chaos.
Squalidus in terra sic stat genitor genitusque,
Quod natura suo vix stat in orbe loco.
Liuor et ambicio, gula, fraus, metuenda libido,
Ira, tumor mentis, scismata, laudis amor,
Ambiciosus honor, amor et sceleratus habendi,
1200 Ipse voluptatis vsus et ecce malus,
Furta, rapina, dolus, metus et periuria, testes
Sunt mundi quod erit ammodo nulla fides.

Hic loquitur de variis vindictis occasione

peccati in hoc seculo iam quasi cotidie635
contingentibus, que absque iustorum virorum
meritis et oracionibus nullatenus sedari
poterunt.

Ecce dies veniunt, predixit quos fore Cristus,

Capm. xxii.
Et patuere diu verba timenda dei.
Precessere fames, pestis, motus quoque terre,
Signaque de celo, stat quoque guerra modo:
Nititur aduersus regnum consurgere regnum,
Gens contra gentem, sic patet omne malum.
Vt pecoris sic est hominis fusus modo sanguis,
1210 Victa iacet pietas, et sinit ista deus:
Est et adhuc vindex extenta manus ferientis
Continuans plagas, nec timet vllus eas.
Longanimis domini sentencia sepe moratur,
Vir bonus inmunis nec malus vllus erit.
Quem deus ille ferit, nullo valet orbe tueri,
Si non contritum culpa relinquat eum.
Mortem peccantis non vult deus, immo misertus
Vult vt vertatur, quo sibi vita datur:
Est pius ipse deus, scripturis sicut habemus,
1220 Pro Sodomis Abrahe dixerat ipse pie:
‘Inter iniquorum tot milia tu populorum
Redde decem iustos, et miserebor eis.
 Est michi nam soli proprium miseris misereri,
Multis pro paucis parcere curo libens.’
O deus, ergo tibi quid dicam, quomodo nostri
Luctus continui sunt tibi nuga quasi?
Nonne decem iusti modo sunt, meritis vt eorum
Stellifer ipse dies curet in orbe malos?
Aut deus oblitus est immemor ad miserandum,
1230 Dormit vel fingit, aut sibi facta latent.
Verius vt dicam, deus est accensus, et ignis
Fulminat inde Iacob, iraque lata furit:
Sic et plasma suum plasmator abhorret, et ipsum
Torquet pro factis que videt ipse malis.
O, qui mentali videt ex oculo mala nostra
Omnibus in gradibus continuare dies,
Dicere tunc poterit quod talia nullus ab euo
Impunita diu crimina vidit homo.
Quis status ille modo, quin sit transgressus, et
ordo,
1240 Quem iustum dicam, deficit vnde sciam.
Hoc nisi gratis emat, dubito prope quod generalis
Decasus nostre prosperitatis adest:
Set quia de summis gradibus mala progrediuntur,
Est qui summus eos corrigat ipse deus.

Hic loquitur sub compendio recapitulando

finaliter de singulis mundi gradibus, qui
singillatim a debito deuiantes ordine virtutes
diminuendo extingunt, et ea que viciorum sunt
augmentando multipliciter exercent.

Dudum prelatus solum diuina gerebat,

Capm. xxiii.
Nunc propter mundum nescit habere deum:
Curatus cure dudum seruiuit, et ipse
Nunc vagus exterius circuit omne genus:
Dudum presbiteri casti, nunc luxuriosi;
1250 Ocia que querunt plurima dampna fouent:
 Ex studio mores dudum didicere scolares,
Nunc tamen econtra stat viciata scola:
Indiuisus amor monachos sibi strinxit vt ardor,
Nunc petit inuidia claustra tenere sua:
Asperitas dudum fratres in carne domabat,
Regula set mollis ammodo parcet eis:636
Dudum milicia fuit et sibi gracia prompta,
Gracia nunc tarda stat, quia vita mala:
Mercator dudum iustum peciit sibi lucrum,
1260 Nunc quoque fraude sua querit habere lucra:
Simplicitas animi fuerat sociata coloni,
Nunc magis indomitum cor gerit ipse ferum:
Lex dudum iusta nulli parcebat amica,
Quam vigor argenti subdit vbique sibi.
Par status imparibus est actibus attenuatus,
Exceditque suum quisque viator iter.
Sic pietas humilis teritur, que superbia regnat;
Liuor adest agilis, torpet et omnis amor:
Permanet ira ferox, et abit paciencia suplex,
1270 Viuit et accidia, sollicitudo perit:
Ebrietas, non sobrietas, tenet ammodo mensas,
Feruet et in viciis crapula plena cibis:
Casta pudicicia dudum precingere lumbos
Affuit, et modo vult soluere luxus eos:
Nuper larga manus inopi sua munera spersit,
Nunc cupit et bursam claudit auara tenax.
Dic modo quot viciis modo sola superbia
mundum
Ad varii sceleris precipitauit opus:637
Dic quot liuor edax acies sua signa sequentes
1280 Subdidit imperio vique metuque suo:
Dic quot auaricie manibus vel mente rapaci
Intendunt populi iura negando dei;
Quot gula deliciis torpet, quot torpor inanes
Carnis adulterio fedat in orbe suo.
Singula nempe vorat anime caro, sic quod vbique
Subdidit inmundam crimine mundus eam:
 Singula fallacis mundi dulcedo subegit,
Nos tamen inmundos mundificare nequit.

Iam in fine libri loquitur magis in speciali de

patria illa in qua natus fuerat, vbi quasi
plangendo conqueritur, qualiter honores et
virtutes veteres a variis ibidem erroribus
superuenientibus, vt dicitur, ad presens
multipliciter eneruantur.

Singula que dominus statuit sibi regna per

Capm. xxiiii.
orbem,
1290 Que magis in Cristi nomine signa gerunt,
Diligo, set propriam super omnia diligo terram,
In qua principium duxit origo meum.
Quicquid agant alie terre, non subruor inde,
Dum tamen ipse foris sisto remotus eis;
Patria set iuuenem que me suscepit alumpnum,
Partibus in cuius semper adhero manens,
Hec si quid patitur, mea viscera compaciuntur,
Nec sine me dampna ferre valebit ea:
Eius in aduersis de pondere sum quasi versus;
1300 Si perstet, persto, si cadat illa, cado.
Que magis ergo grauant presenti tempore, saltem
Vt dicunt alii, scismata plango michi.
Vna meo sensu res est, que pessima cunctis
Iam poterit dici fons et origo mali.
Heu! quia iusticia procul abcessit fugitiua,
Cessit et est alibi pax sociata sibi:
Pax, que iusticie dudum solet oscula ferre,
Nunc fugit a terra, ius perit ecce quia.
Plures iam nocui sumunt sibi regna magistri,
1310 Vis iubet et velle, iura nec vlla videt:
Nunc vbi se vertit magnas, sine iure sequntur
Leges, set populus inde subibit onus:
Corpore sicque meo non tantum torqueor, immo
 Sunt michi pro minimo res quibus vtar ego.
Non est de modicis quod adulterium modo ledit;
Que caro deposcit omnia namque licent.
In terris aliis Venus et si predominetur,
Exsoluunt meritis hoc aliunde suis;
Est ibi nam posita lex, que communis ad omnes
1320 Iudicat, et causas terminat absque dolo:
Non status aut sexus, non dona, preces, timor aut
quid
Possunt a minimo tollere iura viro:
Et sic iusticia redimit quodammodo culpam
Carnis, que fragili condicione cadit.
Set nos in patria non solum vincimur ista
Ex carnis stimulo, quo stimulatur homo;
Immo suas metas lex transit nescia iuris,
Sicque per obliquas patria nostra vias
Deuiat in tanto, quod, dicunt, amplius ordo
1330 Non erit in nostris partibus: vnde deus
Visitat has partes vindicta, qualis ab euo
In nullo mundi tempore visa fuit.
Non tamen est terra que gaudet in omnibus vna,
Set magis in nostra fit modo virga fera:
Clamor vbique, vide, non solus conqueror ipse;
Culpas tam patulas est reticere nephas:
Sic fleo cum flente, lex fallit, fallor et ipse;
Stat mea nam grauibus patria plena malis.
Nos, quibus assueuit numquam crudeliter vti
1340 Fatum, iam pressos sternit vbique reos.
Que fuerat tellus omni preciosa metallo,
Iam nequit ex plumbo pondus habere suum;638
Dignior argento, fuluo quoque dignior auro,
Nobile que genuit, vix valet esse quadrans.
Nuper dixerunt quicumque venire solebant,
‘Venimus ad portus, vbera terra, tuos.’
Nunc tamen vt sterilis reputaris et es, quia mores
Nunc neque diuicie sunt aliquando tue.
Quo ferar, vnde petam mestis solacia rebus?
1350 Anchora iam nostram non tenet vlla ratem.
Sic mea, que stabilis fuit, infirmatur iniquis
Patria iudiciis, iura negando viris:
Sic gentis domina, quasi iam viduata, tributa
Reddit peccato, statque remota deo.
Sic que morigera fuerat, nunc est viciosa;
Dudum legifera, nunc sine lege fera:
Sic ea que larga fuerat, nunc tollit egena;
Que fuerat sancta, fit Venus ipsa dea:
Est sale iam spersa, fuerat que fructibus ampla,
1360 Et velut vrtica, que solet esse rosa:
Que fuerat pulcra, quasi monstrum stat reputata;
Fit caput in caudam, sic terit omnis eam.
Scandala feda parit nouiter transgressa nouerca,
Omnis que laudis mater et hospes erat:
Que fuit angelica nuper, nunc angulus extat,
Languet et in tenebris sorde repressa magis.
Patria, quam famam dicunt habuisse sororem,
Est magis infamis omnibus ipsa locis:
Que fuerat digne super omnes celsior orbe,
1370 Nunc deus est alibi, subditur ipsa quasi:
Ordine retrogrado quicquid sibi laudis habebat
Cedit, et instabilis vndique spreta iacet.
Firma mouet, ruit alta, terit modo forcia discors
Error, et innumera spergit vbique mala:
Torpescunt proceres, clerus dissoluitur, vrbes
Discordant, leges sunt sine iure graues:
Murmurat indomitus vulgus, concrescit abvsus
Peccati solitus; sic dolet omnis humus.
Hinc puto quod seuit pes terreus in caput auri,
1380 Et lupus agnorum cornua vana timet.639
In meritis hominum solum deus aspicit orbem,
Et sua de facto tempora causat homo.

O sterilis terra morum, sani viduata

Consilii, lesa nec medicamen habens,
Dic vbi fortuna latitat modo, qua reputabas
 Nuper in orbe tuum non habuisse parem.
Si Lachesis sortem tibi contulit esse dolosam,
Iam venit ipsa tui r e d d e r e pacta doli:640
Nunc palletque tuis nigris Aurora venenis,
1390 Cuius lux aliis fulsit in orbe magis;
Nuncque iuuentutis flos que tibi creuit habunde
Aret, et a viciis inveterata peris;
Fedaque nunc volucris, venturi nuncia luctus,
Concinit in fatis bubo propheta tuis.
Scit deus hanc causam specialius esse notandam,
Qua locus iste modo distat honore suo:
Hoc scio, quod cunctis locus in prouerbia crescit,
Et quasi nunc speculum denotat omnis eum.
Talia per terras fatali lege geruntur,
1400 Vt reputant, set ego non ita stare puto:
Non est fortuna, que talia causat habenda,
Nec sors, set merita nostra per acta mala.
Qui tamen hanc stare modo credit et hanc
reuocare
Vult, purget crimen, sic reuocabit eam:
Gracia prompta dei querentibus inuenietur,
Nam sibi conuersis vertitur ipse deus.
Dum pia pro pace cecinit processio terre,
Firmaque iusticia fecerat acta sua,
Dumque fides steterat et amor s i n e l a b e
m a n e b a t,641
Tunc, quia pax viguit, sors bona
1410 c u n c t a t u l i t.
N u n c igitur nostra s i t v i t a d e o renouata,642
Ne sors fortuita plus queat esse mala:
Vota vetusta precum redeant domino dominorum,
Vt redeat dominus cum pietate suis;
Per quem pax et honor et tempora sana redibunt,
Que pro peccato sunt fugitiua modo.
Prospera qui veteris vult temporis esse renata,
Reddat et emendet facta priora nouis.
Est deus ipse piis pius et seuerus iniquis,
 1420 Sic valet ob meritum quisquis habere deum.
Nos igitur, domine, tua gracia, que solet olim
Ferre reis veniam, te miserante iuuet:
Anticipet pietas tua nos, ne dicat eorum
Gens, ‘Vbi sit dominus, qui solet esse pius?’
Da, precor, accessum lacrimis, mitissime, nostris,
Nam sine te nullum scis quod habemus opem:
Nunc tua pro lapsis nitatur gracia rebus,
Nostra nec anterior sit tibi culpa memor:
Numquam pigra fuit causis tua gracia nostris:
1430 Est vbi nunc illa, que solet esse salus?
Nos peccatores sumus, et tu plus miserator,
Scit bonitasque tua nos opus esse tuum:
Si plus peccaret vir, plura remittere posses,
Materiam venie sors tibi nostra dedit.
Si quociens homines peccant, tua fulmina mittas,
Exiguo presens tempore mundus erit:
A te pendentem sic cum circumspicis orbem,
Auctor, pacificum fac opus esse tuum.
Nos, deus alme, tui serui, quamuis modo tardi,
1440 Te, non fortunam, credimus esse deum:
Scimus te solum super omnes esse colendum;
Sic nostri solus tu miserere, deus!

Hic loquitur qualiter ea que in hoc presenti

libello quasi sompniando de mundi scripsit
erroribus, non ex se tantum, set ex plebis voce
communi concepit. Consulit tamen finaliter
quod, si quis inde se culpabilem senciat,
priusquam nobis peiora succedant tempora,
suam ex humili corde culpam penitens
emendet.

Capm. xxv. Hos ego compegi versus, quos fuderat in me

Spiritus in sompnis: nox erat illa grauis.
Hec set vt auctor ego non scripsi metra libello,
 Que tamen audiui trado legenda tibi:
Non tumor ex capite proprio me scribere fecit
Ista, set vt voces plebis in aure dabant.
Quem sua mens mordet, de voce sit ille remorsus,
1450 Curet vt in melius que tulit egra prius:
Qui tamen inmunem se sentit, ab inde quietus643
Transeat, et meritis sic stet vterque suis.
Quem non culpa grauat mea non sentencia culpat,
L e d i t u r h i n c n u l l u s, sit nisi forte reus:644
N e g r a u e t e r g o t i b i, gibbosus namque
panelli645
Et non sanus equs ferre recusat onus.
Non tamen in specie quemquam de pondere culpe
Accuso, set eo se probet intus homo:
Non ego mordaci distrinxi crimine quemquam,
1460 Nec meus vllius crimina versus habet.
Que sompno cepi, vigilans mea scripta peregi,
Sint bona dicta bonis, et mala linquo malis:
Omnis enim mundum gemit esse dei laceratum
Vindicta nostri pro grauitate mali.
Ergo suam culpam contrito corde, priusquam
Consumpti simus, corrigat ipse malus.
Corrigit hic mundum, qui cor retinet sibi
mundum:
Cor magis vnde regat, hec sibi scripta legat.
Quod scripsi plebis vox est, s e t e t i s t a
v i d e b i s,646
Quo clamat populus, est ibi sepe
1470 d e u s.
Qui bonus est audit bona, set peruersus obaudit,
Ad bona set pronus audiat ista bonus.
Hec ita scripta sciat malus, vt bonus ammodo fiat,
Et bonus hec querat, vt meliora gerat.
Mundus non ledit iustum, bene dummodo credit,
Quando set excedit, mundus ad arma redit:
Mundus erit talis, fuerit viuens homo qualis;
 Obstet vitalis quilibet ergo malis.
Culpa quidem lata, qua virtus stat
v i c i a t a ,647
Cum non purgata fuerit set
1480 continuata,
Que meruit fata sunt sibi fine data.

648Explicit libellus qui intitulatur Vox

Clamantis, editus precipue super articulo primi
infortunii, quod infortunato Ricardo secundo in
primordiis regni sui, vt audistis, quasi ex dei
virga649 notabiliter in Anglia contingebat. Et
nunc vlterius, quia ipse non inde remorsus,
immo magis ad modum tiranni induratus,
regnum suum assiduis oppressionibus
incessanter flagellare non desistit, diuine
vindicte flagellum vsque in sue deposicionis
exterminium non inmerito assecutus est. Tres
namque tunc regni nobiles super hoc specialius
moti, scilicet Thomas Dux Glouernie, qui
vulgariter dictus est Cignus, Ricardus Comes
Arundellie, qui dicitur Equs, Thomas Comes de
Warrewyk,650 cuius nomen Vrsus, hii vero
vnanimes cum quibusdam aliis proceribus sibi
adherentibus, vt regie malicie fautores delerent,
ad dei laudem regnique commodum in manu
forti iusto animo viriliter insurrexerunt, prout in
hac consequenti cronica, que tripertita est,
scriptor manifestius declarare intendit.

FOOTNOTES:
590 9 magnanimi CEHDL magnatum S
591 27 Consilium CEHDL
592 43 descessit SHG descescit L decessit CED
593 87 da pinguibus CE
594 125 circumprecordia SG
595 160 Est S Et CEHDL
596 167 grossantur S grassantur HDLT crassantur CEG
597 182 dolum SCGDL suum EHT
598 184 scelus] dolus EH
599 187 rutilans albedo set SCGDL albus paries tamen EHT
600 189f.

Sic foris ex auro tumulus splendescit, et intro

Fetet putredo, vermibus esca caro. EHT

DL have both this couplet and that in the text

601 192 falco SCGDL nisus EHT
602 218 pulchra C
603 237 ingenii H (corr.) Ingenuum L assimulari CE
604 256 ipse SL ille CEHD
605 269 Ordinary paragr. in CDL, no paragr. E
606 291 Ordinary paragr. CEDL
607 350 credet CEHGDL credit S
608 409 seruus CE
609 470 estatis C (ras.)
610 490 Passebatque S
611 535 set] et CEL
612 562 sunt] sint CE
613 599 sitis] satis D
614 611 tradidit] contulit CE
615 684 queris SGL mauis (mavis) CEHD
616 691 regno peccato EHL Regni peccato D
617 707 eum CEHDH₂ enim SGLT
618 721 perinterius GLT
619 767 spernens famam C famam serpens L
620 816 iuuat S iuuant EGDLTH₂
621 Cap. xiii. Heading 1 mortui corporis CH
622 843 redolens dudum CEHD
623 903 nisi] sibi C
624 918 et] est S
625 921 nostre D nostri SCEHGL
626 931 imparat C
627 961 fit S sit CEHGDL
628 997 et om. S
629 Cap. xix. Heading 1 duplici CEGL dupplici SHT duplice D
630 1067 Tethis D
631 1095 si SGL sit CEHD
632 1107 Cumque ST Dumque CEHGDL
633 1151 comixta H
634 1186 mouet] tenet EDLT
635 Cap. xxii. Heading 2 quotidie CED
636 1256 parcit CE
637 1278 opes S
638 1342 suum SG suo CEHDL
639 1380 Et lupus SHDL Pastor et CG (ras.) Lupus et E
640 1388 reddere SEHG soluere CDLT
641 1409 f. commune regebat Perstitit in nobis tunc honor atque
salus EDLTH₂ (gerebat E)
642 1411 Sint igitur nostra bona facta deo renouata EDLTH₂
(reuocata L)
643 1451 inmunen S
644 1454 Text SCEG Sic precor vt nullus DLT
645 1455 Text SCEG Detrahet inde michi DL Se trahit inde michi
T
646 1469 f. Text SCEHG (in ista E) per scripta cauebis Que mala
sunt, ideo te dabis atque deo DLTH₂ (perscripta D)
647 1479-81 DLTH₂ have two lines only, as follows:—

Omnibus ipse tamen peior sum, sed

releuamen
Det michi per flamen conditor orbis.
Amen.1480*

648 Explicit, &c. Explicit libellus qui intitulatur Vox Clamantis

(omitting the rest) EDTH₂ Explicit liber intitulatus Vox
clamantis (omitting the rest) L
649 4 virga dei CHG
650 12 Warwyk CH
CRONICA TRIPERTITA651
Ista tripertita, sequitur que, Opus
mente perita humanum est
Cronica seruetur; nam pars que inquirere
pacem et
prima videtur
persequi eam.
Est opus humanum, pars illa Hoc enim
secunda prophanum fecerunt hii tres
Est opus inferni, pars tercia iure proceres de
superni quibus infra fit
Est opus in Cristo. Vir qui bene mencio, vbi
sentit in isto fides interfuit.
Scire potest mira, quid amor sit, Opus inferni
quid sit et ira: est pacem
Est tamen hoc clamor, ‘Omnia turbare,
iustosque regni
vincit Amor.’ interficere. Hoc
enim Ricardus
c a p i t o s u s65
2 dolosa
circumvencione
facere non
timuit.
Opus in
Cristo est
deponere
superbos de
sede et
exalt a r e
humiles.
Hoc enim
deus fecit;
o d i o sum
R i c a rdum de
Solio suo
proiecit, et
 pium Henricum
omni dileccione
gratissimum
cum gloria
sublimari
constituit.

Tolle caput mundi, C ter et sex Hic in prima

lustra fer illi, parte653 cronice
Et decies quinque cum septem comp o s i t o r
post super adde: tempora
Tempus tale nota, qui tunc fuit distinguens,654
causas vnde
Anglia mota. regnum fuit in
Dum stat commotus se diuisum,
Ricardus amore postmodum per
remotus, singula
Principio Regis oritur tractabit.
transgressio legis,
Quo fortuna cadit et humus
retrogreda vadit.
Quomodo surrexit populus, quem
non bene rexit,
Te m p u s a d h u c p l a n g i t s u p e r h o c ,
quod cronica tangit.
Libro testante, stat cronica scripta
per ante;
Est alibi dicta, transit nec ab aure
10 relicta:
Audistis mira, vulgaris que tulit
ira:
Omnibus in villis timuit vir iustus
ab illis.
Rex induratum cor semper habet, Qualiter
neque fatum infortunatus rex
Tale remordebat ipsum, qui iure Ricardus,
virgam dei non
carebat:655 metuens, de
malo in peius
 Stultorum vile sibi consilium suam semper
iuuenile maliciam
continuauit.
Legerat, et sectam senium dedit
esse reiectam:
Consilio iuuenum spirauerat ille venenum,
Quo bona predaret procerum, quos mortificaret:
Sic malus ipse malis adhesit, eisque sodalis
20 Efficitur, tota regis pietate remota.
Tunc accusare quosdam presumpsit auare,
Vnde catallorum gazas spoliaret eorum.
Tres sunt antiqui proceres, quos regis iniqui
Ira magis nouit, et eos occidere vouit:
Et sic qui cati pellem cupit excoriati,
Fingebat causas fallaci pectore Nota de
clausas. iudicibus illis,
Caucius vt factum sibi possit qui vt regis
errorem
habere subactum,
precipue contra
Leges conduxit, pro parte suaque illos tres
reduxit: proceres quos
Munere corrupti suadente occidere vellet
timoreque rupti iustificarent,
Legis in errorem regi tribuere literas sub
30 fauorem: eorum sigillis
scriptas
Hii tunc legiste, quicquid rex dixerat erronice
iste, composuerunt.
Federa componunt, que sigilla sub
ordine ponunt.
Tunc rex letatur, super hoc quod fortificatur,
Quo magis ad plenum diffundat ille venenum:656
Tunc aderant tales iuuenes, qui sunt speciales,
Laudantes regem, quia vertit sic sibi legem.
Hoc concernentes alii, que dolos Qualiter tres
metuentes, proceres
Ad defendendum statuunt cito quid predicti de
regis malicia
sit agendum.
secrecius
Tunc rex festinat, et ad hoc sua premuniti in sui
iussa propinat,
 Vt tres querantur vbi sunt, et ibi defensionem
40 capiantur. roborati sunt.
Tunc tres, qui iusti fuerant et ad arma robusti,
Factum disponunt et ad hoc sua robora ponunt.
Qui fuerant isti proceres, in nomine Cristi
Expedit vt dicam referens, et eis benedicam.
Si non directe procerum cognomina recte,
Hec tamen obscura referam, latitante figura:
Scribere que tendo si mistica verba legendo
Auribus apportant, verum tamen illa reportant.
Sunt Olor, Vrsus, Equs, stat Nota de
eorum quilibet equs, nominibus trium
Non hii diuisi, set in vnum sunt procerum
predictorum
50 quasi visi:
sub figura.
Penna coronata tribus hiis fuit Comes
associata: Marescallus.
Qui gerit S tandem turmam Strenuissimus
comitatur eandem, Comes
Nobilis ille quidem probus et Derbeie.
iuuenis fuit idem,
Sic quasi de celis interfuit ille fidelis:
Hac sub fortuna presens aquilonica Comes
luna657 Northumbrie,
cuius Signum
Non fuit ad sortem, sequitur set
fuit luna
mente cohortem.658 crescens.
Qui solem gessit tenebrosus Qualiter rex,
lumina nescit, cuius Signum
In Troie metas dum vendicat ipse Sol erat, ciues
dietas. Londonienses
Troia fuit prima, per quem sol tendit pro auxilio ab
eis contra
ad yma;
dictos tres
Pallet in eclipsi populus quia non proceres
fauet ipsi: optinendo
Obsistunt turbe Phebo, ne scandat requisiuit; set illi
60 in vrbe, regis maliciam
Dumque suis alis Cignus fuit perplectentes
imperialis. eidem
 Fraus tamen obliquas nubes nullatenus
commouit iniquas, consensierunt.
Extera dum rebus temptauit lumina Qualiter rex
Phebus: Comitem
Cestria surrexit, Aper in qua lumina Oxonie, qui per
aprum
rexit, designatur, vt
Regis vexillum fatue signauerat ipse contra tres
illum. proceres
Set conspiranti deus obstat et antedictos
insidianti, gentes
Quo dolus exosos inuoluit fine bellatrices
secum duceret,
dolosos: in partes
Auxilio Cigni, regis pro parte Cestrie vna
maligni 659 cum regio
Si vis queratur, contraria vis vexillo
70 operatur. destinauit.
Querit Aper latebras, fraudes mortisque
tenebras,660
Quo regnum periat regisque superbia fiat;
Cignus et expresse super hiis que cernit adesse
Prouidet, et curam regni colit ipse futuram:
Ducit Aper gentes, quas concitat arma gerentes,
Liber vt hiis pergat proceresque per omnia
spergat.
Cignus vt hoc sciuit, venientibus Qualiter
obuius iuit, quodam die
Belliger et purgat regnum, quo vita Veneris Comes
Oxonie cum
resurgat:
suis
Cum Venus incepit lucem, sors sequentibus in
bella recepit. conspectu
Stat Tetis a parte, cecidit dum ducis
80 Cestria Marte; 661 Glouernie, qui
Thamisie fluctus capiunt de tunc vulpis
caudam in
sanguine luctus: lancea gessit,
Vicit Olor pennis, sit ei quo vita prope villam
perhennis. Oxonie in
fugam se vertit,
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebooknice.com