Enhanced matching engine for improving the performance of semantic web service discovery

Abstract

Web services are the means to realize the Service Oriented Architecture (SOA) paradigm. One of the key tasks of the Web services is discovery also known as matchmaking. This is the act of locating suitable Web services to fulfill a specific goal and adding semantic descriptions to the Web services is the key to enabling an automated, intelligent discovery process. Current Semantic Web service discovery approaches are primarily classified into logic-based, non-logic-based and hybrid categories. An important challenge yet to be addressed by the current approaches is the use of the available constructs in Web service descriptions to achieve a better performance in matchmaking. Performance is defined in terms of precision and recall as well-known metrics in the information retrieval field. Moreover, when matchmaking a large number of Web services, maintaining a reasonable execution time becomes a crucial challenge. In this research, to address these challenges, a matching engine is proposed. The engine comprises a new logic-based and nonlogic- based matchmaker to improve the performance of Semantic Web service discovery. The proposed logic-based and non-logic-based matchmakers are also combined as a hybrid matchmaker for further improvement of performance. In addition, a pre-matching filter is used in the matching engine to enhance the execution time of matchmaking. The components of the matching engine were developed as prototypes and evaluated by benchmarking the results against data from the standard repository of Web services. The comparative evaluations in terms of performance and execution time highlighted the superiority of the proposed matching engine over the existing and prominent matchmakers. The proposed matching engine has been proven to enhance both the performance and execution time of the Semantic Web service discovery.