A Trust Propagation Model for New-Coming in Multi-agent System

In order to cooperate in the open distributed environment, truster agents need to be aware trustworthiness of trustee agents for selecting suitable interaction partners. Most of the current computational trust models make use of the truster itself experience or reputation from community on trustees to compute the trust values. However, in some circumstances, a truster agent may have no experience with a new trustee or may not obtain the information about the reputation of the new trustee agent. And then the truster agent could not utilize the traditional mechanisms based on experience or reputation to infer some trust value for the new trustee. In this paper, we introduce a novel mechanism for estimating trustworthiness in such a situation. Our proposed mechanism is based on the similarity in profiles of the new trustee and ones of well known agents. A weighted combination model is used for integrating experience trust, reputation and similar trust.