Trust Modeling in Virtual Communities Using Social Network Metrics

The paper presents the novel concept of trust evaluation method for autonomous multi-agent systems. The method is based on assumption that multi-agent system constitutes social network and so it is an instantiation of the complex network Several metrics used in description of this type of networks can be used to model trust relation between agents. In the presented model the agent's A trust in an agent's B statements is obtained as the Subjective Logic consensus of modified trust levels of all agents that trust the agent B. The trust level modification is performed in relation to the current position of the agent in the network.