Web Service Tags Optimization Based On Semantic Similarity And Quantity Of Information

With the rapid development of cloud computing, as one of its base technology, Web service are widely used, the number of Web services on the Internet growth rapidly, how to locate Web services quickly and accurately in order to be used for service discovery and composition becomes necessary and difficult.Nowadays, most of the Web services published on the Web are descripted by WSDL, so how to make full use of WSDL document to find Web services effectively is very important. However, the WSDL document lack of semantic description, Web service similarity matching based on WSDL document has the low accuracy problem. What is more, the structure of some WSDL documents are unstandardized, the existing feature extraction proposal cannot meet the demands. Web service tags are terms annotated by users to describe the functionality or attributes of Web services. They can provide additional information to Web services in order to offset the problem of WSDL documents which lack of sufficient information, which are used to improve the accuracy of Web service similarity matching, thereby improving service discovery, service composition and service clustering, etc. However the present proportion of imprecise and even false tags is high and affects the quality of service similarity matching.The unstandardized structure of WSDL and high proportion of invalid tags will affect the web service similarity matching. In this paper, we propose a WS-TOM model, which contains similarity calculation and tag optimization module, to handle these problems. In the similarity calculation module, we firstly present a feature extraction approach, which considered the programming style and the naming rules, and then give a Web services similarity calculation method. In the tag optimization module, we rank tags by employing both semantic similarity and the information quantity, and then filter some tags according to the Power-law distribution. Experimental results show that the similarity calculation method can not only improve the accuracy of the similarity matching of Web services but also perform well when the WSDL structure is unstandardized. The tag optimization proposal can filter some imprecise tags, thereby improving the accuracy of Web service similarity matching.