Struggling with misbehaviours in trust systems

The growing of networks and the success of fully distributed mechanisms and protocols to exchange data - as peer-to-peer networks - emphasise the need of trustworthiness. Usually both trust and reputation are taken into account: the former expresses the direct experience, whereas the latter represents the common opinion of the whole network. Their applicability and usefulness however could become uncertain when some node of the network is fraudulent, i.e. reports false opinion in order to enhance/reduce the reputation of someone else. In this paper we argue an algorithm - that takes inspiration from the secure Eigen-Trust - aiming at reducing the impact of such fraudulent nodes. We report some preliminary results of simulations performed on Advogato and SqueakFoundation datasets.