Selection of Adaptive Strategies on Main Agent's Attitude Based on Historical Learning

摘要

To solve the problems of uncertainty and variability in the current automated commerce negotiation, the intelligence of agent is applied to the process of the commerce negotiation. We propose the adaptive negotiation strategies based on multi-agent negotiation by historical learning algorithm. During negotiation, the main agent, for example buyer agent, obtains historical information of the opponent, as seller, from the third party agent who stores the information of agents participated in and trade information, and then calculates the negotiation attitude values of the opponents by historical learning algorithm. Considering the information of the dynamic market environment, the main agent presents an appropriate strategy by employing the adaptive concession strategy function and the effectiveness evaluation mechanism. The research achievement of this paper is a foundation for developing a real-life Multi-agent-based commerce negotiation system in the future.