You are on page 1of 4

1

A Semantic Approach for the Enhancement of


Ranking of Web Services
N.S Gowri Ganesh, Viji Gopal
document from the UDDI registry. The UDDI registry
Abstract— Web services are software components located maintains pointers to the Web Service description and to the
in the web which uses open protocols to communicate. A service. The UDDI allows clients to search this registry, find
web service is described by its WSDL document. Web the intended service and retrieve its details.
services allow clients to invoke procedures, functions, and Th emergence of Web Services developments and
methods on remote objects. There are UDDI registries standards in support of automated business integration has
pointing to the WSDL descriptions of the web services. driven major technological advances in the integration
UDDI allows for the discovery of businesses, their web software space, most notably, the Service Oriented
services and the technical interfaces they make available. Architecture.
A search engine takes input from the user, searches for Paper Outline: The remainder of the paper is organized as
matching services in the repository and returns a set of follows. Section 2 introduces the semantic web, Ontologies
web service references to the user. The semantic and Four Tuple Representation. Section 3 and 4 proposes the
representation of a web service make the search much availability of web services with seven tuple representation.
context specific and meaningful. Ontology is used as a We present the enhancing features in section 4, suggested
mean for increasing the certainty of matching service improvement in section 5 and conclude in section 6.
functionality. WSDL documents are converted to more
generic models and its operations are represented as a II.BACKGROUND
tuple OP = <I,O,C,M>, where I is a set of input messages,
and O is a set of output messages. C is a set of concepts, A.Semantic Web
belonging to a set of ontologies. M is a mapping between Current web which can be assumed to be the biggest
input and output messages to concepts, which assigns a global database lacks the existence of a semantic structure
probability value p є [0, 1] that signifies the certainty of to keep the interdependency of its components and as a
the mapping. We propose to enhance the representation of result the information available on web is mostly human
the semantic web service into a seven-tuple such that it is understandable. Semantic web provides some languages
an aid to improve the ranking of web services when the that express information in a machine process-able format.
result is given to the user by the search engine. We also This implies that we can take more benefit from their
processing power. A huge amount of data are conceptually
suggest some ideas for improving the response time of the
related, but much of these relationships still have to be kept
search engine.
in human memory and not stored in an understandable way
for machines. ultimate goal of Semantic Web is to create
Index Terms—availability, ontology, repository, semantic web
services some smarter content which could be understood by
machines. When the content is understood by machine,
some assertions may come out of the content and new
I.INTRODUCTION pieces of information will be produced.

T he services that are available from the internet using


some standard protocols like HTTP or SMTP to access
are called web services. Remote procedures expose input and
output parameters that a web service must support. The client
queries a UDDI registry for the service either by name,
category, identifier, or specification supported. Once located,
the client obtains information about the location of a WSDL

Manuscript received January 16, 2008.


N.S. Gowri Ganesh, Centrer for development of Advanced Computing, ,


A scientific society under Informaton & Commn. Tech.,Govt.of India, Chennai
(phone: 91-44-2461 0880; fax: 91-44-2461-0898; e-mail: nsgganesh@cdac.in). Figure 1: The architecture of a search engine for web
Viji Gopal is a Master of Engineering Student in R.M.K.Engineering
College, Kavaraipettai, Chennai (e-mail: vijigopalakrishnan@gmail.com). services
2

A search engine handles queries to retrieve web services. set of ontologies. M is a mapping between input and output
(figure 1). It takes as input a set of input parameters of the messages to concepts, which assigns a probability value p є
web services and a set of output parameters of the web [0, 1] that signifies the certainty of the mapping.
service. The search engine contains four parts: a Crawler, The search query terms will be mapped to ontological
an Index, a query interface, and a result interface. The concepts in a similar way. We will define a query as a tuple
retrieved services are ranked and presented to the user via Query = <E,C,M>, where E is a set of structured query
the result interface. The crawler discovers, analyzes and expressions (which are not formally defined in this paper). C
indexes semantic descriptions of Web services. The and M have the same meaning as in the formal model of Web
structure of the index allows the questions above to be services described above. Figure depicts the conceptual
answered, by indexing services according to the concepts models of operations and queries. The small-cap c denotes
they relate to, and according to their relations with other concepts, while the i and o symbolizes the operation’s input
services. and output messages, respectively. The service-retrieval
problem is transformed from a mapping problem between a
query and a set of services to a mapping problem between a
B.Ontologies
set of query concepts and a set of service concepts.
During the last few years many Semantic Web related The matching algorithm analyzes the ontology, using
technologies have emerged or have been elaborated. One of related concepts and the structure of the ontology itself. The
the most important parts of these improvements is the status results will be ranked according to the accumulated certainty
of ontology development languages which looks to be more (over the set of query expressions, concepts and messages) of
stable now. The World Wide Web Consortium (W3C) who the match between the query and the services.
has been working intensively on semantic standards, has
approved the Resource Definition Framework (RDF) and
the OWL Web Ontology Language (OWL) and hence III.SEVEN TUPLE REPRESENTATION
provides a solid base to establish enterprise semantic
applications and has implied a significant leverage of the A.Concept of ‘Fairness’
Semantic Web from a research level to an industry standard As described above, a web service has been transformed
for building next generation applications. to concepts. Based on the affinity it is ranked and displayed
Ontologies are building blocks of Semantic Web based to the user. But the World Wide Web is so dynamic that it
systems. Creating ontologies is not an easy task and changes every moment. The services available today may
obviously there is no unique correct ontology for any not be working tomorrow and those who are down today
domain. The real quality of ontology can be assessed only may be placed back on service tomorrow, after some
by its use in real applications. its use in real applications. renovations or improvements. So the content of the
Web services enable us to access relevant applications, repository is also susceptible to this dynamic nature. The
but the discovery, invocation and composition of web index should show the changes that happen to the web
services still need to be supported by human interaction. services listed in it.
This is the point where Semantic Web comes to play and to We propose that the index should be able to show the
support web services with ontologies as an added value. total time the web service was available to the user. The
Combining the strength of web services and the added value availability of the web service,W during its entire life time
of Semantic Web will result in a concrete base for enterprise till the current time is termed as the fairness, F of the web
applications. service. This is an integer value. Time stamp, T represents
C.Four Tuple Representation the time at which the last periodic check on web service,W
Eran toch et. al proposes a search engine which uses the was performed. LatestStatus, L represents the status of W
above said idea of combining semantic web with ontology. at the time of the last availability check on it. L is a Boolean
This approach relies on ontologies as a mean for increasing value. We propose to add these three parameters along with
the certainty of matching service functionality. By using data the standard parameters of the web service ontological
integration and conceptual model techniques, WSDL representation discussed in section II.C. Now the tuple
documents are analyzed and transformed into a generic becomes <I,O, C, M, F, T, L>.
service model. Following that, the service properties are When the result reaches the user, he is shown the values
mapped to concepts that belong to ontologies in different of F, T and L so that the user can know how reliable the
domains. For instance, a library service will be mapped to web service is and what is its present status at ∆t hours
ontologies that describe book-vendors, members and book- back. ∆t is the difference between the current server system
title concepts. A Web service is described as a set of time and the time stamp, T. The time when the last
operations WS = {OP1,OP2, ...,OPn}. The operations, rather availability verification was performed is not directly shown
than the services themselves, are the elements which will be to the user because of the geographical diversity of the
matched against the query. An operation is defined as a tuple location of users. Different places have different time zones.
OP = <I,O,C,M>, where I is a set of input messages, and O is Showing the number of hours before what time an
a set of output messages. C is a set of concepts, belonging to a availability verification was performed eases the
3

complication of time conversion and ambiguities regarding value, the less reliable the web service is. In other words, if
that. the value of F is below 0.2, it indicates that the web service
Each of the parameters F, T and L is assigned an initial was not responding most of the time the verifier performed an
value at the time the crawler enters it in the index. F is set availability verification. If the web service was not responding
to an initial value of 0.500. Its maximum value is 0.999 and whenever the verifier was invoked, its value is 0.001 and does
minimum value is 0.001. Initial value of T is the system not change further.
time of the server at the time the web service is indexed for
the first time. Initial value of L is true.
C.Effect on Ranking
A web service search engine based on ontology ranks the
B.Implementation of Fairness results in the following way. It checks the affinity between the
The fairness of a web service should be calculated and concepts describing the query and the service. The services
stored along with the description, in the index. We use a which has high affinity is given high ranking. But this
verifier to update the fairness of a web service. method can not ensure that a service with a high ranking is
available all the time or most of the time. Sometimes the
service may have been listed, have worked for some time and
shut down after a while.
Our approach ensures that the user is able to know the
responsiveness of a web service when the search engine gives
a result set. From the value of F, user can know how fair the
web service have been during its life time. Also he can know
the latest status of the web service at ∆t hours back when the
availability verifier performed its last check. So even if a web
service was down for a long time and its fairness value is very
low, the user may opt for giving a shot for that web service if
L shows the service is available when it was last checked.

IV.ENHANCING FAIRNESS
There are two chances that a web service does not respond
when the verifier invokes it. It may be shut down or the server
Figure 2: Enhanced Search Engine with Verifier may be busy. In the latter case, actually the web service is
working but not available at present. In this case, we can
Figure 2 shows the structure of an enhanced search engine subtract a value of 0.0005 from the current value of F. This
for web services, with the verifier. The crawler crawls the avoids mistaking a web service as a dead one.
web and populates the index with the references to web
services. The verifier is invoked periodically to check the
current availability of the web services. It takes its input from V.SUGGESTED IMPROVEMENT FOR BETTER PERFORMANCE OF THE
the index and checks each link in the index for liveness. It REPOSITORY
tests the availability of a potential web service by invoking
one of its methods using randomly generated test data.
We plan to improve the response time of the search engine
When the verifier performs an availability verification, if it
by providing a cache effect to its output interface making use
gets a SOAP response from the web service, it indicates that
of the concept of ontology.
the web service is in working condition. Now the verifier adds
0.001 to the current value of F. This indicates the
improvement of fairness. The higher the fairness value, the
VI.CONCLUSION AND FUTURE WORK
more reliable the web service is. In other words, if the value
of F is above 0.900 it indicates that the web service was Even when a rich set of ontologies is built the ontology
available most of the time the verifier performed an engineering process is not terminated, the next problem we
availability verification. If the web service was available should deal with is the issue of ontology aging. Results,
whenever the verifier was invoked, its value is 0.999 and does extracted from an out-of-dated ontology can not be used in a
not change further. totally meaningful way. There should be some mechanism to
When the verifier performs an availability verification, if it detect ontology aging and force the semantic based systems to
gets an HTTP error message, it indicates that the web service evolve ontologies as environmental parameters are changing.
is in not available at that time or it is permanently dead. Now Multilingual support is recognized as one of the most
the verifier subtracts 0.001 from the current value of F. This important challenges of Semantic Web. Nowadays English is
indicates the decrease of fairness. The lower the fairness the predominating language and about 70 percent of Internet
4

content is in English, but only about 44 percent of Internet [18] A Min Tjoa, Amin Andjomshoaa, Ferial Shayeganfar, Roland Wagner.
users are native English speakers. Especially in India this Semantic Web-Challenges and New Requirements. Proceedings of the 16th
International Workshop on Database and Expert Systems Applications
issue is quite essential and the diversity of languages needs to (DEXA’05)
be taken into consideration. One method to handle this
problem is by establishing relevant inter-ontology translators
that map ontologies and content to other languages.
Also we are planning to extend the current research such
that the search engine can identify trusted service providers
because trust is something that is given very high importance
in today’s business world.

REFERENCES
[1] Colin Atkinson, Philipp Bostan, Oliver Hummel and Dietmar Stoll. A
Practical Approach to Web Service Discovery and Retrieval. 2007 IEEE
International Conference on Web Services (ICWS 2007). July 2007.
[2] Birgit Hofreiter, Christian Huemer, Wolfgang Klas. ebXML: Status,
Research Issues, and Obstacles. Proceedings of the 12th Int’l Workshop on
Research Issues in Data Engineering: Engineering e-Commerce/ e-
Business Systems (RIDE’02).
[3] Atkinson, C. and Stoll, D. and Acker, H. and Dadam, P. and Lauer, M. and
Reichert, M.U. (2006) Separating Per-client and Pan-client Views in
Service Specification. In: Proceedings of the 2006 International Workshop
on Service-oriented Software Engineering, 27 - 28 May 2006,
Shanghai, China. pp. 47-53.
[4] Eyhab Al-Masri and Qusay H. Mahmoud. Interoperability among Service
Registry Standards. Published by the IEEE Computer Society. IEEE
INTERNET COMPUTING. MAY-JUNE 2007.
[5] Joseph Chiusano (Booz|Allen|Hamilton). UDDI and ebXML Registry: A
Co-Existence Paradigm. March 2003.
[6] Eran Toch, Iris Reinhartz-Berger, Avigdor Gal, and Dov Dori. OPOSSUM:
Bridging the Gap between Web Services and the Semantic Web.
[7] Natalya F. Noy and Deborah L. McGuinness (2001) “Ontology
Development 101: Guide to Creating Your First Ontology”
http://protege.stanford.edu/publications/ontology_development/
ontology101.html
[8] Farquhar, A. (1997). Ontolingua tutorial. http://ksl-
web.stanford.edu/people/axf /tutorial.pdf
[9] David Martin, Massimo Paolucci, Sheila McIlraith, Mark Burstein,Drew
McDermott, Deborah McGuinness, Bijan Parsia, Terry Payne, Marta
Sabou, Monika Solanki, Naveen Srinivasan, Katia Sycara. Bringing
Semantics to Web Services: The OWL-S Approach.
www.daml.org/services/owl-s/OWL-S-SWSWPC2004-CameraReady.doc
[10] Ankolekar, A., Martin, D. L., Zeng, Hobbs, J. R., Sycara, K., Burstein,
Paolucci, M.,Lassila, O., Mcilraith, S. A., Narayanan, S., and Payne. 2001.
Daml-s: Semantic markup for web services. In Proceedings of the
International Semantic Web Workshop (SWWS). 411–430.
[11] Bechhofer, S., van Harmelen, F., Hendler, J., Horrocks, I., McGuinness,
D., Patel Schneider, P., and Stein, L. 2004. OWL web ontology language
reference. W3c candidate recommendation, W3C.
[12] Bellwood, T., Clement, L., Ehnebuske, D., Hately, A., Hondo, M.,
Husband, Y., Januszewski, K., Lee, S., B., M., Munter, J., and von Riegen,
C. 2002. The UDDI version 3.0 technical report. http://www.uddi.org/.
[13] Christensen, E., F., Meredith, G., and Weerawarana, S. 2001. Web services
description language (wsdl) 1.1. Specification document, W3C. Mar.
[14] Euzenat, J. and Valtchev, P. 2004. Similarity-based ontology alignment in
OWL-lite. Proceedings of ECAI , 333–337.
[15] Schlosser, M., Sintek, M., Decker, S., and Nejdl, W. 2002. Hypercup:
hypercubes, ontologies,and e_cient search on peer-to-peer networks.
Proceedings of the 1st Workshop on Agents and P2P Computing.
[16] Schmidt, C. and Parashar, M. 2004. A peer-to-peer approach to web service
discovery. World Wide Web Journal 7, 2, 211–229.
[17] Toch, E., Gal, A., and Dori, D. 2005. Automatically grounding
semantically-enriched conceptual models to concrete web services. In
Proceedings of the International Conference on Conceptual Modeling
(ER’05). Lecture Notes in Computer Science, vol. 3716. 304–319.

You might also like