A Method for Calculating Trustworthiness in Social Networks

This paper presents a model for computing total trust, essentially a mass function with values for the trust, distrust, and uncertainty (about trust and distrust) an agent (the trustor) should have in a certain other agent (the trustee). There are three components that contribute to total trust: direct trust or distrust, based on the trustor's direct experience with the trustee, indirect trust or distrust, based on referrals the trustor receives from other agents regarding the trustee, and inherent trust or distrust, which exists because of the social context of the agents. Each of these three components provides a mass function with values for trust, distrust and uncertainity, and the Analytic Hierarchal Process (AHP) is used to find weights for combining these functions when there is exactly one referral providing indirect trust. When there are $n > 1$ referrals, there is an indirect-trust mass function for each referral, and the component of total trust provided by indirect trust is the average of these mass functions. As $n$ increases, the weight for indirect trust grows at the expense of the weight for direct trust. The contributions of this paper are the notion of inherent trust, using AHP to provide weights for combining the contributions to total trust, and how we compute the decrease in the contribution of direct trust and increase in indirect trust as the number of referrals increases.