Maintenance-based trust for multi-agent systems

In last years, trust and reputation has been gaining increasing interest in multi-agent systems (MAS). To address this issue, we propose in this paper a maintenance-based trust mechanism for agents operating in multi-agent systems. In the proposed model, a comprehensive trust assessment process is provided to assess the trustworthiness of the participating agents. The main characteristic of this model is the retrospect trust adjustments, which integrate the applicable constraints and modify the involved features with respect to the actual performance of the evaluated agent. Specifically, the retrospect process updates the belief set of the agents in order to adapt them to the social network changes. This paper has two contributions: after describing the architecture of the proposed framework, we provide a theoretical analysis of its assessment and discuss the system implementation, along with simulations comparing it with the broadly known frameworks.