文献类型：期刊文献

中文题名：压缩FP-Tree的改进搜索算法

英文题名：Improved search algorithm based on compressed FP-Tree

作者：吴倩[1];罗健旭[1]

机构：[1]华东理工大学信息学院,上海200237

年份：2015

卷号：36

期号：7

起止页码：1771

中文期刊名：计算机工程与设计

外文期刊名：Computer Engineering and Design

收录：CSTPCD;;北大核心:【北大核心2014】；

基金：国家自然科学基金项目(61304071);中央高校基本科研业务费专项基金项目

语种：中文

中文关键词：数据挖掘;关联规则;压缩频繁模式树;Apriori算法;频繁模式增长算法

外文关键词：data mining; association rules; compressed frequent pattern tree; Apriori algorithm; FP-Growth algorithm

摘要：为克服Apriori算法候选频繁项集的支持数计算效率过低和频繁模式增长算法FP-Growth多次建立条件模式树时内存耗费大的问题,提出基于压缩频繁模式树(CFP-Tree)的改进搜索算法(MCFP-Tree)。利用Apriori算法候选项集生成的思想和压缩频繁模式树紧凑的数据结构,采用自底向上的搜索策略,快速挖掘压缩频繁模式树及其子树,更快得到候选项集的支持数。实验结果表明,该算法可以高效计算出候选频繁项集出现的频次,挖掘效率明显优于Apriori和FPGrowth算法。
To overcome the disadvantage of Apriori algorithms＇ low efficiency of computing the support of candidate frequent item sets and the defect of the great memory consumption while building conditional pattern trees,MCFP-Tree based on the data structure of compressed FP-Tree was proposed.Both of the method of generating candidate frequent item sets and the compact data structure of compressed frequent pattern tree were used.By digging compressed frequent pattern tree and its sub-trees based on the search strategy of top-down,the results were calculated faster.Experiments results show that MCFP-Tree can effectively count the number of candidate sets,which improves the efficiency enormously.