一种有效的多关系决策树分类算法  

ON EFFICIENT MULTI-RELATIONAL DECISION TREE CLASSIFICATION ALGORITHM

作　　者：邓左祥[1,2,3] 李春贵[1,2] DENG Zuoxiang;LI Chungui(School of Computer Science and Communication Engineering,Guangxi University of Science and Technology,Liuzhou 545006,China;Key Laboratory of Intelligent Computing and Distributed Information Processing,Guangxi University of Science and Technology,Liuzhou 545006,China;School of Electronic Information and Electrical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]广西科技大学计算机科学与通信工程学院,柳州545006 [2]广西科技大学智能计算与分布式信息处理重点实验室培育基地,柳州545006 [3]上海交通大学电子信息与电气工程学院,上海200240

出　　处：《内蒙古农业大学学报（自然科学版）》2020年第2期63-68,共6页Journal of Inner Mongolia Agricultural University(Natural Science Edition)

基　　金：国家自然科学基金(61472254,61472255,61420106010,61650203);广西教育厅项目(KY2016YB249,2017KY0352).

摘　　要：多关系数据挖掘,是数据挖掘方向其中一个热门的研究内容,并且是具有挑战的一个问题。在处理多关系时,传统的数据挖掘算法需要进行物理连接,因而存在效率不高的问题。为了解决这个问题,研究多关系数据挖掘的分类,提出一种有效的多关系决策树分类算法,名为EMDT。EMDT的目标是提高分类准确率,并减少运行时间。EMDT利用元组ID传播,构造出一颗决策树,可以直接在多关系中对类标号未知的元组进行分类,不需要进行物理连接。实验表明,EMDT提高分类准确率,并显著减少运行时间。Multi-relational data mining is one of brand-new research directions and a challenging problem in data mining.While dealing with multiple relations,traditional data mining algorithms use physical join,therefore they have the problem of low efficiency.In order to solve this problem,classification in multi-relational data mining was investigated,and an efficient multi-relational decision tree classification algorithm called EMDT was proposed.EMDT intends to increase the accuracy of classification,and decrease running time.By taking advantage of tuple ID propagation,EMDT constructs a decision tree.Without physical join,EMDT can directly classify the tuples of unknown class labels in multiple relations.Performance results demonstrate EMDT increases the accuracy of classification,and significantly decreases running time.

关 键 词：多关系数据挖掘 分类算法 元组ID传播 决策树 

分 类 号：Q5909[生物学—生物化学]