Towards trustworthiness establishment: A D-S evidence theory based scorer reliability tuned method for dishonest feedback filtering

Abstract

Trust is an important character in next generation Internet. Entity's reputation aims at embodying trustworthy interaction information in history and constructing anticipation in future. Spurious reputation threatens to cause latent cheat and leakage in reputation mechanism, which is an exigent problem to be considered in wide distributed, dynamic domain. Based on D-S evidence theory, tuned by scorer reliability, the proposed approach evaluates entities' ratings on multi-facet and filters out dishonest feedbacks and malicious referrers. Importing self-adaptation mechanism, the status of reputation can be adaptively formed, updated, used and evolved in dynamic real-time environment. Compared with other methods, this approach accords with human filtering psychology naturally, is especially instructive in application. © 2008 Springer-Verlag Berlin Heidelberg.