基于模糊逻辑的Agent社会信用评价模型  

Fuzzy-logic based credit evaluation model for agent society

作　　者：胡斌[1] 梁锡坤[2] 高济[1] 

机构地区：[1]浙江大学计算机科学与技术学院,浙江杭州310027 [2]杭州师范大学信息工程学院,浙江杭州310036

出　　处：《浙江大学学报（工学版）》2008年第5期725-730,784,共7页Journal of Zhejiang University：Engineering Science

基　　金：国家“973”重点基础研究发展规划资助项目(2003CB317000)

摘　　要：针对现有Agent社会中交易信用评价模型过于简单且灵活性不够,而社会、经济领域评价模型计算量过大、难以直接应用等问题,结合评价过程和评价结果模糊性的特点,设计了一种基于模糊逻辑的信用评价模型.模型为每个交易影响因素建立论域以及论域上的模糊集,以交易数据作为输入,通过隶属函数得到各影响因素的模糊评价,在此基础上分别进行一次、多次和综合交易信用评价,从而得到客观的交易信用评价结果.模型根据Agent主观和客观信用评价结果推理得到最终信用评价结果.结合实例说明了该方法的可操作性,并将该模型应用于社区/联邦式的政策导向型Agent社会模型SASA5中,为Agent商务交往提供了关键技术支持.Current credit evaluation models implemented in Agent societies are too simple and lack of ftexi~ bility, while the counterparts in social and economic areas need too much computatiori and are hard tO im- plement. Concerned with the fuzzy character of the evaluation process and result, a fuzzy-logic based credit evaluation model was designed. The model establishes domain and fuzzy set for each factor and＇ accepts deal data as input, then it computes the evaluation with membership function and provides one deal, multi-deal and total credit evaluation which constitute the objective evaluation. The model computed the final credit evaluation from the Subjective and objective credit evaluation. Application＇of this model to the policy-driven agent social institution （SASA5） in the form of social sector/federation provides key technique support for the agent commercial intercourse.

关 键 词：SASA5 Agent社会 模糊逻辑 信用 信用评价 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]