Handling Sequences of Belief Change in a Multi-agent Context

This paper focuses on the features of belief change in a multiagent context. Agents have to consider information received from the other agents and the ability of how to change beliefs after receiving a message is a prerequisite for many multi-agents problems. The agents are embedded in an environment where they are communicating and they have to prevent potential internal conflicts in their beliefs. We study the belief change operators in that context. Our approach is to consider that agents’ belief state is a set of pairs 〈belief, origin of the belief〉 combined with a preference relation over the agents embedded in the multi-agent system. The belief revision procedure for handling received messages is a syntactic approach which aims at selecting the minimal subsets of the belief base in conflict with the received information and, according to the reliability of the sources of the conflicting beliefs, removing the less reliable beliefs in order to handle the received message. The proposed belief change operators ensure that an agent makes as few changes as possible but also handle iterated belief changes by preserving the order on the beliefs.