关键词: Agent, 多议题, 协商, 学习, 效用

Abstract: This paper states the algorithm from the respective of the buyer Agent.The negotiation history of the opponent Agent is analyzed firstly，which is obtained from the trade platform.Then it is pre-filtered on the characteristic of data structures.On the basis of the above，it proposes a negotiation rival choosing algorithm and alternation mechanism after analyzing the problems of the present negotiation models.This algorithm can be used to choose negotiation rival and update the initial beliefs of the Agent before negotiation，and plays a guidable role in selecting and implementing the tactics in Agent negotiation，and improve the utility and efficiency in the negotiation.

Key words: Agent, multi-issue, negotiation, study, utility