A fuzzy framework for modelling confidence in multi-agent system

摘要

In open multi-agent systems, agents may hide or lie about information regarding some events. In order to avoid such deception, a multi-agent system must be designed to test the confidence of agents. In this paper, we present a new approach to agent-based confidence modeling. This approach integrates two confidence requirements, namely, trust and certainty. We use fuzzy logic in representing the importance of trust and certainty to model an agent's confidence reasoning. The modeling mechanism eliminates the influence of untrusted agents' opinions because certainty level and importance of trust may not be valuable in all circumstances.