基于前缀路径图的频繁闭项集挖掘算法  

An Efficient Frequent Closed Itemset Mining Approach Based on PrefixpathGraph

作　　者：宋薇[1] 张晓民[1] 郭东恩[1] 

机构地区：[1]南阳理工学院软件学院,南阳473000

出　　处：《计算机与数字工程》2017年第11期2290-2294,2300,共6页Computer & Digital Engineering

基　　金：国家自然科学基金项目"软件系统复杂网络层次化实体挖掘方法及关键技术研究"(编号:61572420)资助

摘　　要：关联规则是数据挖掘的重要方法之一,它主要用来揭示数据库中项或属性之间的相关性。频繁项集是产生关联的基础和核心。频繁闭项集项集数量远远小于频繁项集,而且包含了频繁项集的全部信息。为了有效压缩事务数据库信息,论文提出了前缀路径图结构,该结构可以存储挖掘所需的全部项集信息,减少扫描数据库次数。并且提出了一种基于前缀路径图的频繁闭项集挖掘算法PGraph-FCIMiner。论文的实验均采用Java语言编写,实验结果证明算法具有较好的执行效率和可扩展性。Association rules is one of the important methods of data mining,it is mainly used to reveal the correlation betweenitems or attributes in the database. Frequent itemsets is the basis and core of the generation of the association. The number of fre？quent closed itemsets is much smaller than the frequent itemsets,and it contains all the information of frequent itemsets. A novelstructure termed PrefixpathGraph is defined to store the transaction database compactly,which can store all the item set informationand reduce the number of scanning database. In this paper,a PrefixpathGraph-based algorithm is proposed called PGraph-FCIMin？er. The experiments in this paper are written in Java language,and the experimental results show that the algorithm has a good per？formance and scalability.

关 键 词：数据挖掘 频繁闭项集 前缀路径图 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]