A Web Service Recommendation Approach Based on QoS Prediction Using Fuzzy Clustering

Abstract

Web services, as loosely-coupled software systems, are increasingly being published to the web and there are a large number of services with similar functions. Therefore, service users compare the non-functional properties of services, e.g., Quality of Service (QoS), when they make service selection. This paper aims at generating a more comprehensive web service recommendation to users with a novel approach to fulfill more accurate prediction of unknown services' QoS values. We accomplish the QoS prediction by using fuzzy clustering method with calculating the users' similarity. Our approach improves the prediction accuracy and this is confirmed by comparing experiments with other methods. In addition, the quality of web services is considered as a multi-dimensional object, and each dimension is one aspect of the web service's non-functional properties. We also provide an application example to demonstrate how to utilize our approach to rank services by a score function and map multi-dimensional QoS properties into a single dimensional value.