一种基于可变支持度的缺省规则挖掘算法  

An algorithm of mining eefault decision rules based on variable support measure

作　　者：杨萍[1] 万上海[1] 陈耿[2] 

机构地区：[1]安徽工程科技学院应用数理系,安徽芜湖241000 [2]东南大学计算机科学与工程系,江苏南京210096

出　　处：《安徽工程科技学院学报（自然科学版）》2004年第2期1-5,共5页Journal of Anhui University of Technology and Science

基　　金：国家自然科学基金资助项目(70371015);安徽省自然科学基金资助项目(03042205);安徽省教育厅教学研究资助项目(2003kj029)

摘　　要：Rough集方法提供了一种新的处理不精确、不完全与不相容知识的数学工具.MDRBR算法通过规则支持度进行约束,可有效提高缺省规则的挖掘效率.但MDRBR采用单一的规则支持度约束,使得当规则支持度较小时,挖掘出大量的缺省规则,而当规则支持度较大时,一些重要的小概率分布对象对应的缺省规则被过滤掉.为此,提出了一种基于可变支持度的缺省规则挖掘算法———MDRBVSM,可有效地改进MDRBR等传统算法存在的缺陷.实验结果表明,该算法可有效地过滤噪声、提高规则的挖掘效率.Rough set theory is a new mathematical tool to deal with imprecise, incomplete and inconsistent data. MDRBR algorithm improves efficiency of existing algorithms for mining default rules with support measure constraint. But MDRBR employs the uniform support measure threshold (denoted as sup) such that there exists following shortcoming: A lot of default rules are mined when the sup is very small; conversely, some important default rules that correspond to the sparse objects are filtered when the sup is high. This paper, introduces an algorithm of mining default rules based on variable support measure-MDRBVSM, which can efficiently improves the existing drawbacks of conventional algorithms for mining default rules. MDRBVSM effectively filter out noise and solves the problem of time-consuming of default rules. Experiment results show that the algorithm is practical.

关 键 词：可变支持度 缺省规则 ROUGH集 数据挖掘 粗集 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]