Fuzzy Negotiation Agent by Chaotic Evolutionary System

本研究探讨了在企业并购谈判中作为合作伙伴优先公司选择规则的模糊层次分析法的构建，该模糊层次分析法可有效地进行人类价值判断处理。首先，AHP对多个评价标准进行了配对比较。计算了各准则的重要程度。在此评价标准的基础上对多个替代方案进行配对比较。然后计算了备选方案的重要程度。在传统的层次分析法中，很难采取对比较的一致性，这往往成为决策者的一大负担。本文的技术尝试用模糊数来推导出这种一致性问题的负担，以进行改进。作为模糊ARP的一种具体方式，尝试将三角模糊的个数赋值为对偶比较者的感觉，作为修饰符来表示对偶比较结果。其次，在模糊控制中常用等间隔三角隶属函数进行简化。考虑到三角等间隔隶属度函数不足以令人满意地表达决策者，尝试启发式地优化隶属度函数以表达决策者的感觉。该建议技术利用遗传规划这一进化计算技术，利用沃尔什变换和混沌随机特性，对三角隶属函数进行优化，以提高效率和避免进度停滞。结果，似乎找到了合适的隶属函数。

This research examined the construction of the fuzzy AHP system which is effective for human value judgment processing as a choice rule of the partner ahead company in the mergers and acquisitions (M\&A) negotiations. At first, AHP performs pair comparison for plural evaluation standards. The importance degree concerning the criterion is calculated. The pair comparison of plural substitute plans is done based on this each evaluation standard. And then the importance degree concerning alternatives is calculated. In conventional AHP, it is hard to take consistency of the pair comparison, and it often becomes a big burden for the decision-maker. This paper's technique tried that the burden of the issue of this consistency is deduced by using a fuzzy number for improvement. As a concrete manner of the fuzzy ARP, it is tried to assign the number of the triangle fuzzy to be equal to the sense of the person of pair comparison for a modifier to express a pair comparison result. Next, the triangular membership function of equal intervals is often used in the Fuzzy control for the simplification. Considering that the triangular membership function having an equal interval is an insufficient thing for satisfactory expression of the decision-maker, it is tried that membership function is optimized heuristirelly to express a sense of the decision-maker. This suggestion technique can tune up a triangular membership function as a standard, and optimize it utilizing the genetic programming, that is a kind of the evolution computation technique, with the Walsh transformation and the chaotic random characteristics for efficiency and stagnation evasion of progress. As a result, it seemed that the appropriate membership function was found.