A Combination Approach to QoS Prediction of Web Services

Abstract

With a growing number of alternative Web services that provide the same functionality but differ in quality properties, the problem of selecting the best performing candidate service is becoming more and more important. However, users can hardly have invoked all services, meaning that the QoS values of some services are missing. In this paper, we propose a combination approach used to predict such missing QoS values. It employs an adjusted user-based algorithm using Pearson Correlation Coefficient to predict the QoS values of ordinary services. For services with constantly poor performance, however, it employs the average QoS values observed by different service users instead. An extensive performance study based on a real public dataset is finally reported to verify its effectiveness.