Similarity-based Web service matchmaking

With the increasing growth in popularity of Web services, matchmaking of relevant Web services becomes a significant challenge. Commonly, Web service is described by WSDL and published on UDDI registers. UDDI provides limited search facilities allowing only a keyword-based search of businesses, services, and the so called tModels based on names and identifiers. This category-based keyword-browsing method is clearly insufficient. Semantic Web service uses DAML-S instead of WSDL to represent capabilities of Web services. This improvement enables software agents or search engines to automatically find appropriate Web services via ontologies and reasoning algorithm enriched methods. However, the high cost of formally defining to the heavy and complicated services makes this improvement widespread adoption unlikely. To cope with these limitations, we have developed a suite of methods which assesses the similarity of Web services to achieve matchmaking. In particular, we present a conceptual model which classifies properties of Web services into four categories. For each category, a similarity assessment method has been given. In Web service matchmaking process, these similarity assessment methods can be used together or individually. Experiments highlight complementary contributions that our work makes to facilitate Web service matchmaking.