基于D-S证据理论的多Agent辩论谈判策略研究  

Research on argumentation-based negotiation strategy of Multi-Agent based on D-S evidence theory

作　　者：蒋国瑞[1] 胡应兰[1] 

机构地区：[1]北京工业大学经济与管理学院,北京100124

出　　处：《计算机工程与应用》2013年第12期36-40,64,共6页Computer Engineering and Applications

基　　金：国家自然科学基金(No.71071005)

摘　　要：为了使Multi-Agent辩论谈判策略的生成更加科学合理,避免像贝叶斯算法、遗传算法、神经网络算法和其他机器学习方法在谈判领域对先验知识要求过高的缺陷,利用D-S证据理论方法对信息不确定情况下的辩论谈判策略进行研究,提出基于D-S证据理论的辩论谈判策略生成方法。该方法模拟人作决策时的思考和行为方式,综合考虑谈判的内外部环境状况,随时随地获取证据,对证据进行处理,分析证据处理结果作出决策。通过算例验证该方法在谈判领域的适用性和可行性。In order to make the generation of argumentation-based negotiation strategy of Multi-Agent more scientific and rea- sonable, avoid defects like Bayesian algorithms, genetic algorithms and other machine learning methods which need overstretch prior knowledge in the field of negotiation, this paper presents an approach for generating negotiation strategy of Multi-Agent based on D-S evidence theory under uncertainty information condition to simulate the humans＇ thinking and behavior way in decision-making. The evidence can be obtained and processed anywhere by the method, and decisions can be made by the combination of the evidence, according to internal and external environmental situations of the negotiation. A numerical example is designed to illustrate and verify the proposed method in areas of the negotiations is applicable and feasible.

关 键 词：谈判 自适应策略 MULTI-AGENT系统 D—S证据理论 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]