PEA: Predicting Expert Agents approach

Abstract

In multi-agent recommender system, the knowledge degree of an agent and its trust degree are two main criteria in the decision-making phase. These criteria are used to recommend the expert agent. Therefore, how to model agent and evaluate its trust is becoming a challenging issue. This problem can affect the whole prediction of expert agents. In this paper, we propose a Predicting Expert Agents approach (PEA). We applied a clustering method, Fuzzy Formal Concepts Analysis, to model agent and evaluate its trust and a probabilistic method, Theory of Belief Functions, to predict the expert agent.