Analysis & Generative Model for Trust Networks

Trust Networks are a specific kind of social network where edges in the network have positive and negative signs connoting friendship/trust and antagonism/distrust respectively. While the last few years have seen a rich body of work on generative models for social networks, there hasn’t been much work on understanding in what ways trust networks differ from social networks and to create generative models for them. In this work, we analyze the network structure and properties of two online trust networks – Epinions and Slashdot – using theories of structural balance and social capital from social psychology and propose a generative model for undirected graphs based on these theories that produces synthetic graphs which agree with observed data on key network properties. A generative model for directed graphs is part of ongoing work.