A Semantic Similarity Measure for Semantic Web Services.pdfVIP

A Semantic Similarity Measure for Semantic Web Services Jeffrey Hau William Lee John Darlington London eScience Centre London eScience Centre London eScience Centre Imperial College London Imperial College London Imperial College London 180 Queens Gate, London, UK 180 Queens Gate, London, UK 180 Queens Gate, London, UK jh398@doc.ic.ac.uk wwhl@doc.ic.ac.uk jd@doc.ic.ac.uk ABSTRACT cal approach is often too restrictive. For example, consider Establishing the compatibility of services is an essential pre- an user-oriented service composition environment for com- requisite to service composition. By formally defining the posing complex workflows. In this type of semiautomatic similarity of semantic services, useful information can be composition system, the users would prefer to have a selec- obtained about their compatibility. In this paper we pro- tion of “similar” services that could be potentially composed pose a metric for measuring the similarity of semantic ser- rather than one or two exact logical matches. vices annotated with OWL ontology. Similarity is calculated In this paper we propose a method for calculating the sim- by defining the intrinsic information value of a service de- ilarity between OWL objects (classes and instances) with scription based on the “inferencibility” of each of OWL Lite the view to apply it to match OWL-S annotated services. constructs. We apply this technique to OWL-S, an emerging The similarity between two services is measured by compar- st