Multi-agent Learning Negotiation Research in Virtual Enterprise Based on Contract Net

With the demand of modern global market competence, the negotiation has been used as a mechanism to solve the task distribution, coordination and conflict in virtual enterprises. The contract net protocol tends to bring communication congestion when used in negotiation. In this paper, the Bayesian decision making is introduced to extend the contract net protocol, and a multi-agent negotiation system architecture is constructed. Agents with incomplete information about their opponent can update their belief when taking into consideration what other agents are thinking about during their interactions. The initial experimental results are given and it is proved that the learning model reduces the information exchange times between agents and increases their utility.