Abstract

There have been always motivations to introduce clustering of entities with similar functionality into groups of redundant services or agents. Communities of Web services are composed by aggregating a number of functionally identical Web services. Many communities with the same type of service can be formed and all aim to increase their reputation level in order to obtain more requests. The problem, however, is that there are no incentives for these communities to act truthfully and not providing fake feedback in support of themselves or against others. In this thesis we propose an incentive and game-theoretic-based mechanism dealing with reputation assessment for communities of Web services. The proposed reputation mechanism is based on after-service feedback provided by the users to a logging system. Given that the communities are free to decide about their actions, the proposed method defines the evaluation metrics involved in reputation assessment of the communities and supervises the logging system by means of controller agent in order to verify the validity and soundness of the feedback. We also define incentives so that the best game-theoretical strategy for communities is to act truthfully. Theoretical analysis of the framework along with empirical results are provided.