A Trust and Reputation System Model for Open Multi-Agent System

Trust and reputation are important to effective interactions in open multi-agent system. The FIRE model, which was designed for solving problems in open environment, was a most recent one. However, consumer’s personal characteristics were not considered in this model, and so the ratings given by consumers just directly reflected the performance of providers. This has weakened the usability of this model. An extended FIRE model is proposed in this paper, in which some personal characteristics are introduced for the consumers. The personal characteristics include two aspects: the expected service quality assumed by the consumers and the attitudes taken by the consumers. Both can affect the consumers’ ratings of the same service. As a result, the ratings given by consumers do not directly reflect the performance of the providers any more, and this makes the model closer to the reality. At the same time, when a consumer computes the trust ratings of the providers for selection purpose, the computation depends more on the interaction trust of the consumers providing evidences, and this reduces the effort of the communication between agents. The computational results show that the E-FIRE is comparable with the FIRE when fewer interactions are needed, and it is better when more interactions are needed.