混合属性实例分类与检索的子空间法  

Subspace Method of Mixed Criteria Cases Index

作　　者：钟诗胜[1] 潘启澍[1] 王知行[1] 

机构地区：[1]哈尔滨工业大学机械工程系

出　　处：《哈尔滨工业大学学报》1998年第4期28-32,共5页Journal of Harbin Institute of Technology

基　　金：国家自然科学基金!59675002

摘　　要：在剖析了传统的实例检索策略的不足以后，从向量空间理论出发，提出了一种子空间法实例检索模型，不同类别的实例对应于实例空间的不同的子空间，该模型给出了新问题与各实例类的相似程度的判别准则，从而为异构型分布式实例系统的建造奠定了实例的组织和检索基础。After analysing the existing cases indexing model's disadvantages, a new cases indexing model based on vector space theory, which is called subspace method of mixed criteria cases indexing, is presented. According to this method, different case class corresponds different subspace of case space and the criterion of how to get a similiar degree of new problem with different case class is given. Therefore, isomer and distributed case system's cases organizing and indexing bases are established.

关 键 词：知识工程 分布式系统 子空间法 CBR 检索 

分 类 号：G354.4[文化科学—情报学]