Comparative analysis of IR based web service similarity measures using vector space model

Abstract

With the advent of Service Oriented Architecture (SOA), Web Services have gained tremendous popularity. Due to the availability of a large number of Web services, the task of finding relevant services becomes more and more difficult. Moreover, nowadays large enterprises increasingly rely on Web services as methodology for large-scale software development and sharing of services within an organization. . This warrants the need to establish an effective process of discovering similar services. Web service discovery involves three interrelated phases: 1- matching, 2-assessment, and 3- selection. Our work focuses on the phase 1, service discovery matching, in which we are using standard similarity measures i.e. Cosine Similarity and Euclidean distance to retrieve the most relevant services. As there is no predetermined formula or way to identify the parameters on which the similarity between two services can be identified, we develop a Regression Model to verify the significance of parameters used. In the end, we perform the comparative analysis of approaches we used. © 2012 Springer-Verlag.