Trust estimation and fusion of uncertain information by exploiting consistency

Agents may cooperate by communicating their opinions about various phenomena. These opinions are then fused by agents and used for informed decision-making. However, fusing opinions from diverse sources is not trivial - especially in open multiagent systems - where it is not possible to ensure that the sources are honest and their opinions are not misleading. In this paper, we propose a novel approach that exploits consistencies and conflicts between personal observations and shared information to derive trust evidence for information sources. Based on the derived evidence, we describe how opinions from diverse sources can be fused. We have evaluated our approach for trust estimation and opinion fusion using a service selection scenario. Through extensive simulations, we have shown that our approach significantly outperforms the existing trust-based information fusion approaches.