Rating propagation in web services reputation systems: a fast Shapley value approach

A new challenge in Web services reputation systems is to update the reputation of component services. As an emerging solution, rating propagation has received much attention recently. Current rating propagation algorithms either fail to fairly distribute the overall rating to component services or can realize a fair rating distribution at the cost of exponential time complexity. In this paper, we propose a fast Shapley value approach to propagate the overall rating of a composite service to its component services. Our approach ensures the fairness of rating propagation by using the advantage of the Shapley value, but significantly decreases its computational complexity from exponential to quadratic. Its fairness and efficiency are validated by experiments.