A Novel Web Service Search System Using WSDL

Dr. Sumathi Pawar Vasanth Nayak

Information Science & Engineering Information Science & Engineering
Canara Engineering College, Canara Engineering College,
BenjanaPadavu, Bantwal, India. BenjanaPadavu, Bantwal, India.
[email protected] [email protected]

GeethaLaxmi Dr. Niranjan N Chiplunkar

Information Science & Engineering Computer Science & Engineering
Canara Engineering College, NMAM Institute of Technology,
BenjanaPadavu, Bantwal, India. Nitte, Karkala, India.
[email protected] [email protected]

Abstract: The proposed system searches Web Services are classified into SOAP
SOAP based Web services. This search WebServices and REST(Representational
system dynamically discovers the Web Sate Transfer) Web Services. In SOAP Web
services and integrates the required Web Services, request and responses are
services to satisfy user requests. Using transferred through SOAP messages.
search engine, WSDL files are retrieved REST Web Services are light weight and
and these files are processed to select request, responses are transferred through
required Web Services. In this system, URL of the browser.
user requested functional words are SOAP messages are heavy in nature
searched using search engine. Horizontal, and requires more bandwidth to request and
vertical, inter-cluster and intra-cluster response messages of Web Services.
searches are performed on these search Therefore it gives delay in communication.
results. All elements of WSDL files are To overcome this problem RESTFul Web
compared with user given functional word Services are used because very minimum
using weight matrix and selected Web required information only transferred. But
services are executed to satisfy the user RESTFul web services are not secured and
request. Performance of different search SOAP Web Services are secured. All
system is analyzed. Due to dynamic nature security factors such as confidentiality and
of Web Services stored data set is used for integrity can be applied to SOAP messages
implementation. Instead WSDL links are if required.
retrieved during run time which is novelty In the proposed research it is surveyed
of the proposed system. that “currently many of the web services do
not contain enough description information
Keywords: WebServices, Simple Object in their WSDL files. Since, enriching
Access Protocol, Service Description descriptions for public web services using
Language, Universal Directory the information outside of WSDL files
Description Interface, eXtended Markup becomes a great challenge”. So, the
Language. proposed research uses description of the
required elements in the WSDL, to mine the
I. INRODUCTION unknown information.
”A Web service is a loose coupled, Distributed Component Object Models
self-described software module identified (DCOM), Remote Method Invocation
by an URI. The WebServices are capable of (RMI), and Common Object Request
bind using interfaces and are being Broker Architecture (CORBA) were the
identified, described and discovered by most often used technologies before Web
XML artifacts[1][2][4]. Using XML based services. But these technologies were not
messages via Internet-based protocols web supported by every platform. Gateways
services supports direct interactions. were required to convert data format and
Standard Internet technologies such as unsupported protocol to target platform.
XML and HTTP (HyperText Transfer Thus, Web services came into
Protocol) are used between communication existence and were platform-independent.
of two WebServices. SOAP messages This means client uses the Web services in
are used for communicating request and the same way, even though .NET or J2EE
responses of Web Services. WSDL is used are used to built the Web Services. Web
for describing Web Services. services use internet technologies for
interoperability and vendor of every
technology supports it.
Many researches are exists in this area. consumer. If the registered Web services are not
Among these researches some of the available then such status information will not
existing systems use semantic-based be available in the UDDI.
discovery systems which are
Ontology-based. But the Ontologies are B. Ontology based Discovery
static in nature. Some existing systems It is very tedious and time-consuming to
were used UDDI which are no longer discover the Web Services[7]. In this surveyed
present today. It is difficult to get the system, authors proposed an approach to
location of WSDL due to absence of homogeneous Web services’ clustering. This
UDDI and also it is difficult to get system provided easier way of accessing Web
QoS(Quality of Service) parameters unless Services through RSS/Atom for the consumer.
the Web service is executed. Web Service The updates and changes tracked by service
discovery based on data mining techniques feeds. Authors established Graph-based
also available[9]. algorithm to find similar operations and
composable operations. To find similarity
II. LITERATURE SURVEY
between operations, they constructed the
A. UDDI-based search Service Aggregation Graph by considering
In the olden days, UDDI was storing operation, input and output and evaluated the
repository of WSDL links and all the required efficiency using precision and recall. But they
information related to Web services such as have not discussed about the execution of these
QoS, provider information, business entity Web services. Moreover there is no interaction
information and category related informations. between consumer and system to give opinion
But the public UDDI is not working nowadays. about service composition.
Since January 2006, UDDI Business Registry The authors are worked on the
(UBR) is shutdown and Microsoft, SAP and data-driven services and provided an
IBM have given the intrinsic problems of the approach for discovery of services by
UDDI based service discovery using internet. WSDL files data model extraction[8].
Before the year 2006, Discovery of Web Using ebXML authors prototyped the
services were done with the help of discovery of services. To find the services
UDDI(Universal Directory Description they searched through global Ontology with
Interface) which was used as a overlapping data models that enables the
platform-independent framework for defining data-driven composition.
public available services, to discover required DAML proposed semantic markup of
services, and to share information about the Web Services, which allows the definition
services for interaction in a global registry. of special kinds of services in
Web service registry(UDDI) works like Ontology[13].
"Yellow pages," that provides information of DAML proposed an Ontology
classification [3][5]. Service is identified using which provides special kind of description
name-value pairs based on taxonomy of Yellow of Web services and called semantic
pages. The Yellow pages allow any valid markup of Web services. The first step
identifier to be attached to the business white towards the creation of semantic enabled
pages. The Yellow pages stores and manages the Web Services is DAML+OIL[6].
Web service information based on a geographical Usage of Natural Language processing for
taxonomy. interpreting information and semantic analysis to
A technical model (tModel for short) of retrieve required information from bulk
UDDI defines the type of service. A tModel documents proposed by the Author[14].
defines the technical information of the service
C. Service Discovery Based on WSDL
such as the tModel name, organization name that
offers the service, categories of the services, The WSDL files could not provide
pointer to WSDL specification, service interface sufficient descriptions or details of the Web
definitions, message formats and protocols, QoS services in their WSDL links[5]. Therefore
information such as security information. WSDL information can be enriched by
extracting existing elements and by providing
In UDDI searching of Web Services semantics to them using data processing
takes place to get the link of the Web services. techniques. Authors also proved in their system
They manually access the services. The that matching of <service> name is important
provider of the Web services is required to than other elements. More than one WSDL file
register the Web services and then only these may be resulted for a requested Webservice. It is
services’ information can be retrieved by the necessary to compare other parameters such as
<operation>, <input>, <output> etc when more Constraints for formation of clustering is
than one service name matches the request. cluster of operations of same Web Service
Hence the proposed research assigned weight to and cluster of operations of different
all these main elements of WSDL. According to similar Web services. Even though
the weightage of the matching, the required Web functional word entered is not matches
services are found. The similarity of the Web exactly to operation element, from online
services is also found using Jaccard’s similarity semantically similar Web services are
and Euclidian distance with WORDNET. retrieved through search engine techniques.
Jaccard and Euclidian combined with These Web services are grouped into one
WORDNET are used to find similarity of Web cluster called as constraint based
services[12]. But most of the text descriptions clustering.
available in the published Web service are
useless information which are not related with B. Horizontal Search
functionality of Web services. The proposed The process of horizontal search is
research extracts only useful information from treating the unknown input parameter’s
WSDL file through WSDL processing. name as functional word and querying it in
The Web service Search engine the search engine. In the retrieved result, if
(Woogle) also uses clustering and association any of the operation’s name resolves the
rule to find the appropriate services according unknown input parameter’s value and if the
to user requests. [10]Authors discussed the input parameter of this resolved operation
problem of encompassing the large number of is unknown to the user then once again the
data sets with heterogeneous data models. intra-cluster search and inter-cluster
The limitations of the Web UDDI based searches are performed to resolve unknown
search is, Public UDDIs are not available input parameter’s value.
nowadays. The limitations of Ontology based
discovery is, Ontologies are dynamic in nature. C. Vertical Search
To overcome both of these methods, this system
uses WSDL based search. If there are no exact matching
operations to resolve the unknown input of
III. METHODOLOGY required operations then these operations
are extracted from synonym matching
The outline of the methodologies used clusters and this type of search is called as
to implement this research is shown in the vertical search.
figure1. Major steps of these methodologies
are given here. D. Intra-Cluster Search
Intra-cluster search indicates
 Functional word required for user is searching for unknown input parameters’
given as input this system. values within the cluster of operations of
 This functional word is searched in the the same Web service. The unknown
search engine. input parameter’s value is compared with
 WSDL links are retrieved as output of all operation’s functions of the same
search results and similar Web Services cluster. If it is not matching exactly, then
WSDL links are retrieved it is compared with output element’s name
 User requested input is compared with of the operations of the same cluster.
input element of WSDL During this comparison, stop words are
 If user required input is matched in needed to be removed. If the unknown
WSDL of many similar Web Services value is not resolved by the intra -cluster
then perform WSDL weight matrix search, then inter-cluster search is
computation to select suitable Web performed.
Services.
 If user required input is not matching E. Inter-Cluster Search
to input element of WSDL then unknown The inter-cluster search indicates
input need to be resolved using horizontal searching for unknown input parameter’s
search, vertical search, inter-cluster search values among different clusters of similar
and intra-cluster search methods. Web Services. The unknown input
 During these searches the resultant parameter’s value is compared with all
WSDL are passed through WSDL weight operations’ functionalities of the different
matrix and selected Web Services are clusters. If it is not matching exactly, then
invoked. it is compared with output elements of
A. Constraint Based Clustering these operations. In this search, if the
 Input Functional Word & known Input
parameters

Perform dynamic discovery using query Functional

_word?WSDL in the search engine

Extract WSDL links from the retrieved results

Cluster the similar operations using constraint based

clustering

Compare Input of matched operations with user requirement

No Yes No
Perform Intra-cluster search for searching Is matches Is all unknown
R
the unknown inputs inputs resolved

Yes
Perform inter-cluster search for searching Perform horizontal search for searching the
the unresolved unknown inputs unresolved unknown inputs Pass matched WSDL through WSDL-Weight
Matrix for selection of suitable Web service

Integrate and Invoke the operations of

matched WebServices dynamically Perform vertical search for searching the remaining
unknown inputs Invoke the selected Web service dynamically and
record Qos parameters

C End
Figure 1 Flow of the System

unknown input is not resolved then once In Table 1 weights are assigned depends
again
Figure 1this unknown input parameter is on importance of elements in WSDL. It is found
searched using search engine to retrieve from the different search methods that, different
community Web services using horizontal searches may give more than one similar Web
search which is mentioned in the above services which are matching to the requested
section. functional word[11]. Therefore it is necessary to
select suitable Web services from these similar
Web services. Returned results from search
methods are passed through the weight matrix of
F. WSDL Weight Matrix WSDL for selection of suitable Web services
from the array of search results.
Matching of
Table 1 WSDL Weight Matrix
Sum of row requested functional word
WSDL Requested <Operation> <Output> <input> <pre (Total Weight to <operation> element is
Record Functional WEL1=0.4 WEL2=0.3 WEL3=0.2 condition> of matched
word WEL =0.1
4 element) more important in WSDL
than other elements.
WSDL Exact Exact Exact Therefore this element is
of Weather 0.4*0.5=0.20 0.3*0.5=0.15 0.2*0.5=0.1 NA 0.36
weather (MFWSDL1,1) (MFWSDL1,2) (MFWSDL1,3) (MFWSDL ) 1,4
given by weight 1. The
service
user requested query
may provide more than
WSDL Exact Synonym Exact
of Currency 0.4*0.5=0.20 0.3*0.2=0.06 0.2*0.5=0.1 NA 0.27 one similar WSDL
currency (MFWSDL2,1) (MFWSDL2,2) (MFWSDL2,3) (MFWSDL ) 2,4 records. It is necessary to
service
compare <output>,
WSDL Partial Synonym Fail <input> and pre-condition
of
USZip
Zip Code 0.4*0.4=0.16
(MFWSDL3,1)
0.3*0.2=0.06
(MFWSDL3,2)
0.2*0.0=0
(MFWSDL3,3)
NA
(MFWSDL )
0.22 also with if <operation>
3,4
service element matches.
In the proposed
research it is focused on matching of with WSDLHelper class. The Web service is
requested term with these different four searched using search engine and JSoup APIs
elements with following weighting factors. are used to process the retrieved results.
Exact> Partial> Synonym > Fail The javax.xml.rpc.ServiceFactory class
As shown in the table 1 in this system, is used to prepare the call method and to
different weights given to different elements invoke the Web services by filling the
of WSDL record, according to their required parameters The
importance. <Operation> element assigned factory.createService method uses API of
by 0.4, <output> element assigned by 0.3, factory object by passing WSDL link,
<input> element assigned by 0.2 and namespace and serviceName as parameters.
<pre-condition> element given by 0.1. ServiceName is already extracted from the
The extent of matching factors and WSDL using getAllServices() of service
weights used for different types of matching definition and getLocalPart() method of
are exact (0.5), partial (0.3), synonym (0.2) QName. The QName class is defined in the
and fail(0) as matching factors (MF). javax.xml.namespace package to indicate the
If user requested function is exactly type of binding.
matches with <operation> element of WSDL
then WSDL matrix value of exact match is “QName portName = new
0.5 i.e.Matching Factor (MF). Weight factor QName(namespace,serviceName+"Soap");”
of <operation> is initialized as 0.4 I.e. weight The above statement initializes type of
of element(WEL). Using equation 3.1, binding to SOAP.
multiplying of weight of exact match and “QName operationName = new
weight of <operation> element gives QName(namespace,Matched_operationName
0.4 x 0.5 = )
0.20 The above statement initializes operation
According to equation (3.1), the name to matched_operation name(The
computed net weight ( ) of matched_operation name is operation
<operation> element in the weight matrix of which matches to the user requested
WSDL is 0.4 x 0.5 = 0.20 functional word). This also requires
In the equation (3.1) I,J is namespace of the service definition.
multiplied with the weight of i th WSDL “Call call =
record and J th matching factor where 1≤I ≤N (Call)service.createCall(portName,
and 1≤ J≤M and N and M are belong to finite Matched_operationName);”
set of natural numbers. The above statement prepares the call method
WELj is weight assigned to J th element as with portName and matched operation name.
shown in the table1 where 1≤ J≤N. “result=call.invoke(PopulatedInputParamete
MF i,j is Matching factor of given rList);”
functional word with j th element in the i th This statement invokes the Web services
WSDL record where MF i,j gives exact or using call.invoke() method with populated
partial or synonym. input parameters.

(3.2) V. RESULTS
This system is executed in i3 processor with
Computational sum of weight of
th Windows operating system and J2EE platform
WSDL elements Sum i of i WSDL with WSDL helper class library file. The data set
record is computed in the equation(3.2) used is retrieved results during online search
where addition of computed weights because it is dynamic search system and not
of a particular WSDL record. The experimented with stored data set. Due to
WSDL records which are having highest dynamic nature of Web services it is not possible
SumMFWSDL is considered for invocation to work with existing stored data set. For
or integration of Web services and operation example
details of this operation is stored in the Average search speeds of different search
Selected_operation_details array and given as methods of WebServices for searching 10
input parameter to WebService Invocation services are shown in Table2. The average
procedure. search speed of proposed system is scored
second place comparing to other systems.
IV. IMPLEMENTATION General users take more time to search for Web
services because they are not knowing proper
This system is implemented using J2EE way to search for Web Services. The IT
Lunar Eclipse with Windows operating professionals may not be knowing WSDL of
system. WSDL processing is performed
Web Services. Service professionals those [4] Khalid Elgazzar, Ahmed E. Hassan,
having indepth knowledge of WSDL may use Patrick Martin , “Clustering WSDL Documents to
these different systems in very high speed Bootstrap the Discovery of Web Services”, Proc of
Systems. IEEE international conference on Web Services
Table 2 Results of different Web Service Search System –USA, Jul 2010, pp 150-158.
[5] Lijie Wang, Fei Lie,
Liangjie Zhang, Ge Li Bing Xie
, " Enriching Descriptions
for public Web services
using Information Captured
from Related Web pages on the
Internet" , Proc of Fifth IEEE
International Symposium on
Service Oriented System
Engineering – China, Vol 5, Jun
- 2010, pp 141-150.
[6] Thomas Fischer, Johaness
Rubland, Friedrich Schiller,
“Towards Knowledge
Table 2 shows comparison between speed of Discovery in the Semantic
Web” , Proceedings of International conference
search of proposed system and other existing University Jena -MIKWI – 2010, pp 1151-1162.
system. As shown in the table the proposed [7] Xuanzhe Liu, Gang Huang, Mei Hong,
system resulted into high performance “Discovering Homogeneous Web Service Community
comparing to other four existing systems in the User-Centric Web Environment”, IEEE
Transactions on Service Computing, IEEE -2009.
excluding SOAS system. [8] M. Bynens, B. D. Win, W. Joosen , B. Theeten,
"Ontology-Based Discovery of Data-Driven
VI. CONCLUSIONS & FUTURE WORK Services", 2006 Second IEEE International
Symposium on Service-Oriented System Engineering
The proposed system discovers, (SOSE'06), Shanghai, 2006, pp. 175-178.
selects and integrates/composes the [9] Aishwarya Bose, “Effective Web service
required Web services. The literature Discovery using a Combination of Semantic Model
survey is conducted in all these areas of the and Data Mining Technique”, Thesis submitted to
QueensLand University of Technology, Brisbane,
Web services. The drawback of the existing
Queenland, Australia 2008 pp 22-35.
systems, which is indicated in the literature [10] Heath, T., Hepp, M., & Bizer, C, Linked Data,
review of section 2, became the motivation “The story so far. Special Issue on Linked Data”,
for this research. International Journal on Semantic Web and
The proposed system searches for the Information Systems (IJSWIS),
WSDL of Web services using Bingo search http://linkeddata.org/docs/ijswis-special-issue -2011.
engine and not dependent on UDDI [11] Sumathi, Dr. Niranjan N Chiplunkar, “Populating
registry, because public UDDI registries are Parameters of Web Services by Automatic
not available nowadays. Normally the Composition Using Search Precision and WSDL
Weight Matrix”, IJCSE, 1742-7193, InderScience
QoS information is stored in the UDDI and
Publishers – Scopus indexed, (in press:Accepted on
consumer of the Web service should get July 2016).
this QoS information such as response time, (http://www.inderscience.com/info/ingeneral/forthco
security information etc from UDDI. But ming.php?jcode=IJCSE) .
the absence of UDDI hides this information [12] F. Liu, Y. Shi, J. Yu, T. Wang and J. Wu,
from the consumer. Therefore the proposed “Measuring Similarity of Web Services Based on
research invoked the Web services and then WSDL”, Proceedings of IEEE International
records the QoS information for future use. Conference on Web Services, Miami, FL, 2010, pp.
References 155-162.
[1] Web services Activity in World Wide Web [13] OWL-S: Semantic Markup for Web Services,
Consortium. https://www.w3.org/2002/ws/ date of
http://www.w3.org/Submission/2004/SUBM-OWL-S-
access : 01/10/2012.
[2] Thomas Fischer, Johaness Rubland, Friedrich 20041122, last modified Nov-2004, Date of
Schiller, “Towards Knowledge Discovery in the access:10/05/2013.
Semantic Web” , Proceedings of International [14] Joby, P. P. "Expedient Information Retrieval
conference University Jena -MIKWI – 2010, pp System for Web Pages Using the Natural Language
1151-1162. Modeling." Journal of Artificial Intelligence 2, no. 02
[3] “Universal Description Discovery and Integration”
(2020): 100-110.
,https://en.wikipedia.org/wiki/
Universal_Description_Discovery_and_Integration,
last modified on 23 November 2014, at 13:47 date of
access 12/12/2014.