Abstract:
Service oriented architectures provide more effective and dynamic applications. Using Semantic Web services in service oriented architectures improves interoperability and scalability. A very important aspect of using Semantic Web services is the matchmaking process. Semantic matchmaking is used during discovery and composition of Semantic Web services to find valuable service candidates. Among these candidates, best ones are chosen to build up the composition, or for substitution in the case of an execution failure. In this research we enhance Semantic Advanced Matchmaker (SAM) which is based on input and output matching. Our new matchmaking agent supports precondition and effect expressions written in SWRL during matchmaking in addition to input and output annotations. We present a novel approach for assigning matchmaking scores to condition expressions in OWL-S documents written in SWRL language. Proposed matchmaking method utilizes subsumption-based similarity, property-level similarity, similarity distance annotations and WordNet-based similarity. The algorithm uses bipartite graphs in order to find matching parameters of requests and advertisements and then ranks the advertisements according to their semantic similarity to the request. Using classical information retrieval evaluation techniques we show that our proposed agent, which is capable of precondition and effect matching, has significantly higher precision performance with respect to SAM on input and output matching..