Evaluating Feedback Ratings for Measuring Reputation of Web Services

In the field of service computing, reputation of a Web service is usually calculated using feedback ratings provided by service users. However, the existing of malicious ratings and different preferences of different service users often lead to a bias towards positive or negative ratings. In this paper, we propose a novel reputation measure method for Web services. The proposed method employs two phases (i.e., malicious rating detection and rating adjustment) to enhance the reputation measure accuracy. We first detect malicious feedback ratings by the Cumulative Sum Method, and then reduce the affect of different user feedback preferences by using Pearson Correlation Coefficient. Extensive experiments are conducted. Experimental results show that our proposed method is effective and can enhance the reliability of service selection.