基于模糊概念层的核属性集确定方法及其应用  

An Ascertainment Method of Core Attribute Set Based on Fuzzy Conceptual Hierarchy

作　　者：李茹[1] 张丽芳[1] 褚诚缘[2] 

机构地区：[1]山西大学计算机与信息技术学院,太原030006 [2]中国科学院计算技术研究所,北京100080

出　　处：《计算机工程与应用》2005年第30期162-165,共4页Computer Engineering and Applications

基　　金：国家863高技术研究发展计划项目资助(编号:2004AA115460)

摘　　要：核属性集的确定是从海量数据中发现决策规则的重要步骤。该文在结合粗糙集理论和层次聚类方法的基础上,改进了已有的模糊概念层核属性集确定方法,不仅能运用模糊概念层方法对原始数据进行模糊化处理,排除边缘数据的干扰作用,还可减少噪声数据的干扰作用以进一步简化可区分矩阵,得出核属性集且使得生成的判定规则可信度更高。最后以汽车生产厂商汽车样本各项指标实测数据为例论证了算法的有效性和实用性。The ascertainment of the core attribute set is the important process for discovering decision rules from mass data.With the combination of rough set theory and hierarchical clustering method,this paper improves the ascertainment method of core attribute set presented before.The improved method can not only blur the original data using fuzzy conceptual hierarchy method,eliminates the interference effect of marginal data,but also reduce the disturbance of exceptional data,and then it could make further reduction to the constitution of discernibillty matrix and gets the core attribute set.And the decision rules would be more credible.Finally the effectivity and practicability are proved by analyzing the measured value of cars supplied by some manufacturing plants.

关 键 词：模糊概念层 核属性集 数据挖掘 知识发现 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]