Personalized pre-trust reputation management in social P2P network

In this paper, we first show that EigenTrust works poorly when facing sophisticated threat models in which there are disguised peers or spies. We argue that the eigenvector based uniform trust propagation model employed by EigenTrust fails to prevent good peers from propagating their trust to malicious ones, who can always gain positive ratings from good peers by fooling them with some good services from time to time. Then we propose PersonalTrust, a personalized pre-trust reputation management model on the basic of EigenTrust. PersonalTrust utilizes personalized pre-trust and the transitivity of pre-trust. We assume that except for those pre-trusted peers selected by system, every peer can select their own personalized pre-trusted peers based on their own interactions with other peers in the network. By utilizing the transitivity of pre-trust, we can obtain the pre-trust matrix which contains the information to ensure that the trust of peers will only propagate within the circle of trusted friends. In addition, we incorporate a mechanism into PersonalTrust to automatically update the pre-trust matrix. The results of our simulations reveal that our proposal performs better than EigenTrust, and outperforms EigenTrust when threat models consist of not only pure malicious peers but also disguised peers or spies.