A Hybrid Approach for Discovery of OWL-S Services Based on Functional and Non-Functional Properties

摘要

Web services are independent software systems designed to offer machine-to-machine interactions over the WWW to achieve well-described operations. Typically, service providers expose their services to the public by providing brief descriptions of the service's operations; the challenge is to discover the right service in response to a specific service request based on rather sparse service descriptions. In this work, we present a hybrid semantic web service discovery framework that offer semantic web service discovery based on both functional (Input, Output, Precondition and Effect - IOPE) and non-functional (text-description) properties of OWL-S (Semantic Markup for Web Services) services. For functional parameter matching, we have used the bipartite graph-based approach for matching the parameters of services. For non-functional parameter matching, we have used natural language processing techniques on the textual-description of a web service. The cumulative similarly measures determine the overall similarity of the advertised service with the service request. We evaluated the performance of our Service Matchmaking framework using the OWLS-TC4 (Test Collection version 4) dataset, and furthermore compared its performance with some existing discovery models. Our results indicate that the proposed web service matchmaking framework offers an improved discovery mechanism with a significant increase in recall rate.