Web Service Matchmaking Using a Hybrid of Signature and Specification Matching Methods

Web services are independent software systems designed to offer machine-to-machine interactions over the WWW to achieve well-described operations. Web Service Matchmaking is a process of searching (or discovering) for a web service within a service registry (such as UDDI) to discover a web service that satisfies certain functional requirements as requested by a service consumer. In this paper, we present a hybrid web service matchmaker that analyzes both the signature and specification of a web service. For signature level service matching, we have developed two techniques: (i) logical similarity measures applied to the services' input/output concepts, and (b) non-logical matching based on a Structure Preserving Semantic Matching algorithm. For specification matching, we have introduced a short sentence matching approach applied to a services' description. We evaluated the performance of the matchmaker using the OWLS-TC dataset, and furthermore compared its performance with the similar hybrid matchmaker. Our results indicate a significant improvement in recall and slight improvement in precision.