PACT: A New Trust Prediction Method for Multi-agents Recommender Systems

In multi-agents recommender systems, agents interact with each other to provide an efficient recommendation to the end user. Trust is, therefore, important to make these interactions useful. In an open, heterogeneous and dynamic multi-agent environment, it is difficult for agents to assess and establish trusting relationships for cooperation. Thus, modeling and evaluating trust relationships between agents for a better recommendation, is a major challenge. In this paper, we propose a new approach that allows predicting trust relationships between agents. This approach is based on a sound mathematical basis, namely the Fuzzy Formal Concepts Analysis and the Theory of Belief Functions. To validate the efficiency of our work, we carried out a series of experiments using the Advogato dataset.