A Gen-Fuzzy Based Strategy (GFBS) for Web Service Classification

Abstract

Service oriented computing makes great growth in developing web services. A web service is a software application with a standardized way of providing interoperability between disparate applications. Web service repository has a main challenging task in how it can help end user and another web service to group and retrieve their service with high performance. With increased number of both unrestricted and distributed services in web that belong to different domains, web services classification become necessary. This paper proposes a new classification strategy for web services called Gen-Fuzzy Based Strategy. The proposed strategy consists of two stages, which are; (i) Outlier Rejection Stage (ORS) and (ii) Classification Stage. In ORS, a proposed fuzzy system has been presented. It used four parameters which are; Distance to Class Center, Cooling Force, Class Belonging Degree and Domain Belonging Degree. In CS, a proposed Distance Based Ensemble Classifiers (DBEC) has been presented. The aim of DBEC is to train each individual classifier independently by using bagging technique. A collective decision is made by their aggregation in several ways. Experimental results have shown the effectiveness of DBEC in terms of accuracy, precision and recall.