Solving the Rating Scarcity Issue using an Enhanced Memory-based Collaborative Filtering Method

Abstract

In Service Oriented Architecture (SOA), reputation-oriented web service discovery has gained popularity in finding the optimal service from a pool of services having similar functionality. Almost all reputation-oriented discovery mechanisms make use of the feedback ratings reported by the users in order to assess service reputations. However, there are certain factors which, if not addressed carefully, may affect the process of precise service reputation evaluation. One such factor is the issue regarding rating scarcity. When the percentage of users who rate the web services compared to the percentage of users who avail the web services is low, the issue of missing feedback rating arises which leads to incomplete rating matrix. Since all users, after availing services, may not report their satisfaction levels in the form of feedback ratings, it is obvious that the system will encounter incompleteness in rating information while evaluating service reputations. In this paper, an approach to solve the rating scarcity issue in reputation-oriented service discovery is proposed using an enhanced memory-based collaborative filtering method. Experiments are performed and the results are reported in this paper.