一种基于案例的Agent多议题协商模型

A Case Based Agent Multi-Issue Negotiation Model

不完全信息条件下的Agent协商最优回价策略一般采用间接学习对手偏好的方式;另一方面,Agent一般拥有或多或少的经验和知识,这将帮助它们取得更好的协商结果.这启发了用基于案例的方法直接学习得到最优回价,提出了不完全信息条件下基于案例和对策论的Agent多议题Pareto最优协商模型.所给出的算法计算复杂度为多项式级,且当案例库规模控制在一定范围内时低于Fatima工作的计算复杂度.实验结果显示,采用该算法的Agent能够取得比人类更优的效用和更短的达成一致时间,且优于Lin等人的实验效果.改进了Fatima等人的工作.

Multi-agents multi-issue negotiation under incomplete information is a challenge in open environment. However, until now, the strategy of optimal counter-offer generating under incomplete information is not ideal. Previous work usually use indirect approaches to acquire the preferences of opponents through a variety of data mining of other methods such as the researches of Fatima. On the other hand, agents usually have some experiences and domain knowledge which may help them get better negotiation results. This fact inspires the authors to directly investigate negotiation using case-based method. For this purpose, the authors propose an agent multi-issue negotiation model under incomplete information based on cases and game theory. The Cases are regarded as successful interactions and can be reused in future according to the similarity. A Pareto optimal result is proved in this paper. In particular, the optimal counter-offer can ensure the maximal utility of oneself and the maximal similarity of offer for opponents. The computational complexity of the proposed algorithm is polynomial order and it is commonly lower than that of Fatima as long as the scale of cases base is limited to a bounded quantities. Experimental results indicate that the utility and reaching time of the experiments have an advantage over that of human beings and the method of Lin et al. It improves the work of Fatima.