Biblio

Trust is an important topic in today's interconnected world. Breaches of trust in today's systems has had profound effects upon us all, and they are very difficult and costly to fix especially when caused by flaws in the system's architecture. Trust modeling can expose these types of issues, but modeling trust in complex multi-tiered system architectures can be very difficult. Often experts have differing views of trust and how it applies to systems within their domain. This work presents a graph-based modeling methodology that normalizes the application of trust across disparate system domains allowing the modeling of complex intersystem trust relationships. An algorithm is proposed that applies graph theory to model, validate and evaluate trust in system architectures. Also, it provides the means to apply metrics to compare and prioritize the effectiveness of trust management in system and component architectures. The results produced by the algorithm can be used in conjunction with systems engineering processes to ensure both trust and the efficient use of resources.

This paper presents a framework for multiagent systems trust modeling that reasons about both user credibility and user similarity. Through simulation, we are able to show that our approach works well in social networking environments by presenting messages to users with high predicted benefit.