基于案例推理的信用风险控制系统  

CBR-Based Control System of Credit Risk

作　　者：王春震[1] 郭亚军[1] 于兆吉[2] 陈薇[1] 

机构地区：[1]东北大学工商管理学院,辽宁沈阳110004 [2]沈阳工业大学管理学院,辽宁沈阳110023

出　　处：《东北大学学报（自然科学版）》2007年第3期449-453,共5页Journal of Northeastern University(Natural Science)

基　　金：科技部科技兴贸行动计划项目(2003EE550001)

摘　　要：以复杂型信用风险控制问题为应用背景,为了进一步提高案例推理系统的有效性,构建了一种新型的案例推理系统架构.采用了基于框架的案例表达方式,提出了基于改进模糊贴近度的一种案例匹配新方法,提出了统计分析与规划模型相结合的案例调整方式,探讨了几种适用的具体案例维护方法.通过系统实例运行,验证了系统的有效性与准确性.与传统的最邻近法相比,基于模糊贴近度的案例匹配方法提高了匹配精度,同时,新的案例调整方式提高了解的质量.Taking the complex credit risk control as a background, a new frame of CBR （case- based reasoning） system is built to improve further the validity of the existing CBR system. Introducing the frame-based case representation, a novel case adaptation method based on fuzzy proximity is presented for case retrieval. A new way combining statistic analysis with programming model is proposed for case adaptation, and several methods are discussed to maintain properly the definite cases. By the system running of instances, both the validity and veracity are verified. Compared with the conventional nearest neighbor approach, the adaptation method based on fuzzy proximity can improve the retrieval precision and the new way for case adaptation will improve the quality of solutions.

关 键 词：信用风险 风险控制 案例推理 案例匹配 模糊贴近度 规划模型 

分 类 号：F272[经济管理—企业管理]