Web Service Discovery: Adding Semantics through Service Request Expansion and Latent Semantic Indexing

Abstract

Semantic Web technology is a promising first step for automated service discovery. Most current approaches for web service discovery cater to semantic web services, i.e., web services that have semantic tagged descriptions. It is difficult, however, to expect all new services to have semantic descriptions associated. Furthermore, the descriptions of the vast majority of already existing services do not have explicitly associated semantics. There are also severe restrictions on the prospective conversion of existing, non-semantic, descriptions, e.g., Web service description language (WSDL), to corresponding semantic descriptions, e.g., OWL-based Web service ontology (OWL-S). In this paper we present a novel approach for web service discovery that combines ontology linking with latent semantic indexing (LSI). The basic idea is to build the service request vector according to the domain ontology, have the training set of the LSI classifier based on features extracted from selected WSDL files, and finally project the description vectors and the request vector and utilize the cosine measure to determine similarities and to retrieve relevant WSDL service descriptions.