Research On A Feature Vector-based Web Service Discovery Algorithm

Web services is a developing rapidly distributed computing model in recent years. It can mask differences between application programming written by different programming languages and software development platforms, further realize the interoperability between various component. Service-Oriented Architecture involves Web Services providers, Web Services consumers, Web Services register center, Web Services publication, Web Services discovery and Web Services invoking. The basic process of Web service discovery is like this: first carry on description about Web services, then make use of the matching algorithm to find proper Web services in deployed Web services according Web services request information, finally filter, sort search results and submit the final results to users. Web service discovery is one of web service key technologies. Web service discovery aims to meet the requirements of the service from a large number of web services. Currently, web service discovery based on syntax take use of classification and keyword matching. Although the way is use of query efficiency, the precision of query is too low to meet users. The space vector model and clustering algorithm provide solutions for the-mentioned problems, using vector and clustering to classify web services can effectively improve the precision of web service. At present, based-syntax web service discovery has two shortcomings:disdependent classification and low precision.In this text, we deeply study technologies of web service discovery at home and abroad, the research follows below:Firstly, we deeply analyze Web Services Description Language. Because web service discovery mainly use portType, Message, Operation, Binding in WSDL, we deeply research these elements. portType and Operation contain input information, output information and approach rules. Message contains information type. Secondly, we study Concept Lattice theory and Vector-based text information retrieval technology in-depth and introduce these theories into web service discovery. Furthermore, the way improve the precision increasingly.Finally, on the basis of current web service discovery, we propose feature vector-based web service discovery algorithm. The method takes the most information from web service description as key feature items which accurately and effectively represent web service, then we use Web Service clustering algorithm to classify web services. Reducing the amount of retrieved web services, the way can improve the precision and recall. Besides, on the base of algorithm, we design a feature vector-based web service discovery system, and describe web service library module, Web service matching module and web service vector module in detail. In the end. we establish experiment and prove properties of the algorithm.