A time-aware weighted-SVM model for web service QoS prediction

Abstract

With the rapid development of Web services, how to identify services with high Quality of Service (QoS) becomes a hot research topic. Since time-series QoS records are highly nonlinear, complex and uncertain, it is difficult to make accurate predictions through conventional mathematic methods. In order to deal with the challenging issue, this paper proposes a novel personalized QoS prediction approach considering both the temporal dynamics of QoS attributes and the influence of different QoS records. First, slide-window based data grouping is firstly utilized to obtain training dataset for regression model. Then we take the different influence of history QoS records at different time into consideration and eventually propose a weighted-SVM model for QoS prediction. Compared to Auto-Regressive Moving Average Model (ARMA), standard SVM and Collaborative Filtering (CF), the proposed approach in the paper can improve significantly the accuracy in personalized QoS prediction.