高级检索

    一种劝说式多Agent多议题协商方法

    Persuasive Multi-Agent Multi-Issue Negotiation

    • 摘要: 多Agent系统中的协商问题往往由许多议题组成,导致问题空间十分庞大.传统的协商方法通过对问题空间进行穷尽搜索来找到最优解,并不适合多议题协商.而且,传统的方法不考虑协商偏好变化的情况,使得Agent在不完全及不正确环境下找到的最优解并不合理.提出一种劝说式多Agent多议题协商方法.借助信念修正这一有效的推理工具,协商Agent能够在协商过程中接受协商对手的劝说,考虑对手对协商议题的偏好,并根据一种基于辩论的信念修正方法调整自身的偏好.这样就能够使协商Agent对变化的协商环境具备适应性,从而提高协商的效率及正确率,快速准确地达成协议.

       

      Abstract: The negotiation in MAS is usually comprised of multiple issues, which causes the extremely huge problem space. Classical negotiation methods based on game theory obtain the optimal resolution through thorough search of the space, which are consequently not suitable for multi-issue negotiation. In addition, those methods consider little about the change of the agents' preference, so that with the incomplete and incorrect information, the optimal resolution is not rational. A persuasive multi-agent multi-issue negotiation method is illustrated, which is underpinned by belief revision logic. These belief-based negotiation agents are able to persuade their opponents to change their position. Preliminary experiments show that the belief-based negotiation agents can adapt to the changing environment and reach the agreement more quickly and correctly than those based on the classical negotiation model under time pressure.

       

    /

    返回文章
    返回