Computing with Trust: Definition, Properties, and Algorithms

Abstract

Trust is an important facet of relationships in social networks. Whether it is for use in security, determining data access, or recommender systems, the definition of trust guides the development of algorithms to make computations over the trust relationships. In this paper, we present a definition of trust based on sociological foundations. We then describe several properties of trust, including transitivity and composability, that follow from the definition and research in the social sciences, to help guide the methods used for computing with trust, and we discuss the range of values used for expressing trust. We then present a review of algorithms that compute trust in social networks