Towards an automatic enrichment of semantic web services descriptions

Abstract

Web service discovery consists in identifying suitable existing services that satisfy specific goals or user requirements. Web service discovery can be a hard task because of the lack of explicit and semantic information for apprehending what services really do. In this article, we propose a service description enrichment approach based on I/O relations for improving Web service discovery. Our process is divided into two parallel steps. The first step consists in extracting existing relations between I/O of the services from the underlying ontologies using SPARQL, while the second step concerns the extraction of services’ I/O relations from the text descriptions of services using NLP techniques. Matching the I/O relations extracted by the two steps is applied in order to enrich the initial service description, allowing a more accurate automatic service discovery. This article presents our approach that uses dependency grammar and word2vec as well as our experimental results on OWLS-TC.