Andre Manoel, Renato Vicente
Reputation systems seek to infer which members of a community can be trusted based on ratings they issue about each other. We construct a Bayesian inference model and simulate approximate estimates using belief propagation (BP). The model is then mapped onto a problem of equilibrium statistical mechanics that is analyzed theoretically by employing the replica symmetric cavity approach. The theoretical performance, in terms of estimation error, and robustness of the BP approximation in scenarios involving different topologies, parameter mismatches and targeted attacks are evaluated. Having the fraction of trustful nodes and environment noise level as control parameters, we also calculate theoretical phase diagrams. These diagrams are fully confirmed by simulation and show regions of degraded performance explained by the emergence of a ferromagnetic ordered spin glass phase.
View original:
http://arxiv.org/abs/1211.6462
No comments:
Post a Comment