DiReCT: Dirichlet-based Reputation and Credential Trust management

Access control is one of the most security requirements that face some challenges in multiagent systems due to agents' autonomy, distribution and communication with strangers. Unlike current access control mechanisms applicable to traditional distributed systems, trust management model provides a more flexible and dynamic access control approach suitable for the nature of multiagent systems, since it depends on agents' attributes rather than identities. This paper presents DiReCT framework; a Dirichlet-based Reputation and Credentials Trust management framework for multiagent systems, that integrates both reputation-based trust management and policy-based trust management models. We argue such integration can enhance significantly trust management systems. DiReCT uses a Bayesian-based approach through employing Dirichlet reputation system for reputation computation. DiReCT also employs SWRL, a semantic-based rule language, for policy representation and reasoning.