Trust Description and Propagation System: Semantics and axiomatization

Abstract

Trust plays a vital role in many successful applications in computer science. However, existing work fails to provide a systematic analysis of trust in a consistent logic framework, resulting in the fact that trust can only be used pragmatically and even be misused. Therefore, in the form of language, semantics and axiomatization, we propose a sound and complete trust logical system named Trust Description and Propagation System (TDPS). TDPS gives a formal language that can explicitly define the various trusts: direct belief trust, direct commitment trust, meta belief trust and meta commitment trust. In terms of possible states, our TDPS semantic model can formally interpret the different meanings of trust and their degrees (trust, distrust and untrust) in logic. Based on the semantic model, we derive the trust calculation mechanism instead of using heuristics. Moreover, we propose and demonstrate a sound and complete trust deduction system to rigorously discern and define the transitivity of trust, which provides a firm theoretic and logical foundation for various trust applications. Through demo examples generalized from real e-business applications, we show how to use the TDPS system to represent, calculate and refer the various trust relationships.