多Agent谈判中议题相关性及权重度量研究  

Research on Issues Relativity and Weight Measurement in Multi-Agent Negotiation

作　　者：甘早斌[1] 朱春喜[1] 肖国强[1] 

机构地区：[1]华中科技大学计算机科学与技术学院,武汉430074

出　　处：《计算机工程》2011年第18期167-170,共4页Computer Engineering

基　　金：国家自然科学基金资助项目(70672041);中央高校基本科研业务费基金资助项目(2010MS112);湖北省自然科学基金资助项目(2007ABA307)

摘　　要：已有的多议题谈判研究局限于议题相互独立、议题权重固定等前提假设。针对该问题,提出一种议题分组算法。采用移动平均法和事例推理法使用户能更灵活地设定权重,根据单个议题权重初值的确定函数以及议题组的权重求解规则,给出分组变化时的权重动态变更算法。分析结果表明,该算法能解决多议题谈判中的议题相关性问题和议题权重的动态度量问题,使多议题谈判更加科学合理。Existing researches on multi-agent multi-issue negotiation are based on the assumption that the relationship among issues is independent and weights are given fixed.To solve these problems,an issue grouping algorithm is presented in this paper.The function of computing the initial value of a single issue＇s weight and the rule of calculating the weight of issue group are given by means of moving average method and case-based reasoning.In addition,a dynamic weight algorithm for group changing is studied.Compared with similar operations in existing multi-issue negotiation models,the interdependency of mutli-issue can be solved well and the dynamic weight evaluation can be implemented,the negotiation result is more scientific and more reasonable.

关 键 词：多Agent谈判 议题分组 动态组权重 事例推理 移动平均法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]