基于统计特征的属性相似度计算模型  

A similarity calculation strategy based on the statistic of case feature

作　　者：李锋[1] 周凯波[1] 冯珊[1] 

机构地区：[1]华中科技大学系统工程所,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2005年第6期80-82,共3页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金"九五"重大项目资助(79990580);国家自然科学基金资助项目(60174039).

摘　　要：利用统计方法挖掘案例库中案例属性的客观规律,在此基础上建立案例之间相似度计算模型.其实质是通过样本反应出来的属性的分布特征,将传统计算模型中案例属性数值上的距离与空间上几何距离的线性映射关系改进为非线性映射关系,并在此基础上通过传统的最近相邻检索算法计算出更客观的相似度.所提出的计算模型能够更客观地反映属性之间的相似程度,并提高案例检索算法的精度,同时进一步提高了案例推理系统的有效性.In this paper, latent knowledge of cases′ feature was mined from case bases by using statistical methods, and novel knowledge-based model built to calculate the similarity. By using the cases′ feature distribution character to map numeric distance and geometric distance linearity relation in tradition calculate model to non-linear relation. After this preconditioning calculate the degree of similarity among different cases by using traditional KNN arithmetic. The model does not only represent the degree of similarity between two cases′ feature objectively, but also improve the accuracy of case retrieval and the validity of case-based system.

关 键 词：案例检索 基于案例推理 相似度 

分 类 号：TP306.3[自动化与计算机技术—计算机系统结构]