Classification of Web Services using Fuzzy Classifiers with Feature Selection and Weighted Average Accuracy

Abstract

Web services have become an innovative and accepted means of service delivery over the Internet. In recent years there has been astounding growth in the number of web services provisioned by businesses and corporate houses. In the presence of a plethora of web services, a service consumer faces the real challenge of making a right choice based on certain preferences. Therefore, it becomes necessary to classify a set of web services based on certain quality parameters in order to facilitate user choice of web services under different scenarios. Several classification techniques have been proposed by researchers to classify data sets in different application domains. In this work, we have employed three fuzzy classifiers, namely, Fuzzy Nearest Neighbor, Fuzzy Rough Nearest Neighbor, and Fuzzy Rough Ownership Nearest Neighbor to classify web services. We have used the standard QWS dataset for our experimentation. The accuracy of the classifiers has been computed with and without feature selection. In order to further improve classification accuracy, a Weighted Average Accuracy technique has been applied to the confusion matrix obtained after feature selection.