Web service classification approach with an integrated similarity measure

Service-oriented computing creates a surge of web services. With the increasing number of web services advertised via the Web, service repositories have to face a challenging task, i.e., to automatically group those Web services so as to help themselves or end-users to retrieve those services effectively and efficiently. In this paper, a new web service classification method is proposed. Firstly, a new integrated similarity measure for Web service is developed. It combines a statistic measure (TF-IDF) and a semantic similarity measure based on information content. Then, a couple of popular classifiers, e.g. radial basis function neural network (RBFN) and K-nearest neighborhoods (KNN) are used to group Web services. Experimental results on the OWLS-TC dataset indicate the proposed integrated approach outperforms the semantic method which is popular in Web service discovery.