A Practical Approach to Web Service Discovery and Retrieval

1
A Practical Approach to Web Service Discovery and
Retrieval

• Colin Atkinson, Philipp Bostan, Oliver Hummel and
  Dietmar Stoll
• 2007 IEEE International Conference on Web
  Services (ICWS 2007)

Speaker Chun-Hung Wang Date 2009.08.19
2
Outline

• Introduction
• Standard Web Service Brokerage Model
• Web Service Indices
• Lucene/javas jar
• Index Creation
• Index Maintenance
• Suggested Improvements to Web Service Standards
• Conclusion

3
Introduction

• One of the fundamental pillars of the web service
  vision is a brokerage system that enables
  services to be published to a searchable
  repository and later retrieved by potential
  users.
• This paper point out the argument is that
  concepts from component-based development have
  been too naively transferred to web services,
  with the result that many of the related problems
  have been inherited as well.

4
Standard Web Service Brokerage Model

5
Lucene
Web Service Indices
Application Lucene Diagram
6
Lucene

• EX
• Body1Tom lives in Guangzhou,I live in Guangzhou
  too.
• Body2He once lived in Shanghai.
• Body1 Keywordtom live guangzhou i live
  guangzhou
• Body2 Keywordhe live shanghai

7
Lucene
Tab1 General index structure
Tab2 Lucene index structure
8
Index Creation

• One way is through the explicit publication
  efforts of web service developers .
• The other is by means of some kind of crawling
  activity which is focused on finding and
  analyzing web services.
• Crawling the visible web for WSDL descriptions of
  web services presents two basic challenges
• recognition of valid WSDL files and dependencies
• detection of properly working web services

9
Index Maintenance

• Conventional search engines like Google and Yahoo
  keep their content up to date by recrawling the
  web on a periodic basis.
• For search engines that index web services this
  is not possible because the availability of a
  WSDL description does not necessary imply that
  the web service is on-line and working.
• Periodically test the availability of the web
  services in its index, or to test their
  availability before delivering a search result.

10
Suggested Improvements to Web Service Standards

• The first convention propose is that every web
  service should implement a standard liveness
  operation which can be used to check that it is
  still on line and has not been retired.
• They propose to mark them by adding additional
  attributes (such as "sessionID")to the XML schema
  data types definition in a WSDL document.
• Since there will always be a large number of web
  services which are not annotated in this way, at
  least for some time to come, heuristics of the
  kind proposed in the previous section have to be
  used for identifying these management parameters
  and indexing them accordingly.

11
Conclusion

• Using a combination of these techniques, our
  merobase1 search engine has been able to
  assemble a repository of about 3000 existing web
  services.
• 1 http//www.merobase.com/