一种基于相似度方法的并发协商策略

多Agent协商是目前人工智能、电子商务等领域研究的热点问题。在电子商务活动中为了达到更好的效果,获得更多的利益,买方希望与多个卖方进行并发协商(与多个卖方同时进行协商),从中选择最佳的交易解决方案,这种需求广泛存在。本文利用相似度方法(similarity criteria)发展和评价了一种双边多议题多Agent并发协商策略,一方面使用相似度函数来计算对手的偏好类型,从而选择相应的协商策略;另一方面协商一方采用该策略后能够使得双方达成一致的可能性最大并且收益最大,从而使得一个Agent能够在半竞争、信息不完全和不确定以及存在最大协商时间的情况下,更为有效地完成多Agent并发协商。

A Similarity Criteria Based Strategy for Concurrent Negotiation

Concurrent negotiation is an important form in multi-agent negotiation system.This paper develops and eval- uates a similarity criteria based strategy for concurrent negotiation in trading environments.By using this strategy,an agent can compute the opponent's preference in concurrent multi-issue negotiation in semi-competitive situations in which there exist information uncertainty and deadlines.The strategy enables the negotiation more effective and increa- ses utilities of the system.