摘要
使用自动的agent来进行双边多属性协商,是当前人工智能领域的一个重要研究热点.通过引入一个公正的中间agent,提出了一个新颖的封闭多属性协商模型.协商双方agent同时向中间agent提交各自的出价,协商双方最关键的信息(比如每一轮的出价,效用偏好等)是完全私有的,不向对方公布.并进一步通过理论分析,基于期望效用最大化原则,给出了协商双方的最优策略.本文重要的贡献在于将复杂的多属性协商问题简化成单属性协商问题,并能避免交替式出价模式中协商双方都不愿同时让步或揭露更多信息这样一种僵局,在一定程度上能防止欺诈和投机行为.最后,通过一组模拟实验验证了所提出的协商模型是有效的.
Bilateral multi-issue negotiation based on autonomous agents is considered as a significant research issue in artificial intelli- gence. In this paper,we propose a novel sealed-offer multi-issue negotiation model by introducing a nonbiased mediator agent. In our model, both agents simultaneously submit their respective offers to the mediate agent. The most critical information （ e. g. offers of each round, utility preference ） of both agents is kept completely private. Further, we analyze how an agent can exploit the available information in selecting a strategy that maximizes its expected utility and construct an efficient negotiation strategy which can reach an a- greement aiming to maximize their owner＇s utility. An important contribution of this paper is that we simplify complex multi-issue problem into a single-issue negotiation to analyze. This design avoids a deadlock in which both negotiators are unwilling to concede or refuse to disclose more information may arise in an alternating-offer bargaining protocol. And this design can discourage counter-speculation and effectively control fraud and misrepresentation in a certain extent. A group of automated experiments are conducted to evaluate the performance of the proposed automated negotiation procedure. The experimental results demonstrate the efficiency of the proposed negotiation model.
出处
《小型微型计算机系统》
CSCD
北大核心
2015年第8期1842-1846,共5页
Journal of Chinese Computer Systems
基金
国家自然科学青年基金项目(71201050)资助
关键词
智能AGENT
自动协商
多属性
封闭出价
时间约束
autonomous agent
automated negotiation
multi-issue
sealed-offer
time constraint
作者简介
E—mail：misalinlan9904@163．com张林兰，女，1982年生，博士，副教授，研究方向为自动协商协调理论机制研究
刘青,男，1979年生，硕士，工程师，研究方向为协商协调机制研究、产业政策研究；
彭显琪,女，1979年生，硕士，讲师，研究方向为经济金融．
