布尔矩阵Apriori算法的MapReduce并行化实现  

Parallel Implementation of the Apriori Algorithm of Boolean Matrix Based on MapReduce Programing Mode

作　　者：陈方健[1,2] 张明新[1,2] 杨昆[3] 

机构地区：[1]苏州大学计算机科学与技术学院,江苏苏州215006 [2]常熟理工学院计算机科学与工程学院,江苏常熟215500 [3]中国矿业大学计算机科学与技术学院,江苏徐州221116

出　　处：《常熟理工学院学报》2014年第2期98-101,106,共5页Journal of Changshu Institute of Technology

摘　　要：提出基于云计算平台(以Hadoop为例)应用布尔矩阵Apriori算法进行大数据关联规则挖掘的MR_B_Apriori算法.将Hadoop平台与布尔矩阵Apriori算法相结合,利用MapReduce框架分块处理布尔矩阵,计算出分块数据的频度,合并融合得到大数据集的频繁项集.分析表明MR_B_Apriori算法能够适用于大数据的频繁项集挖掘.Knowledge discovery is the bottleneck of big data applications in the environment of big data at pres-ent. The authors of this paper propose a MR_B_Apriori algorithm for mining association rules in large data, which is the Apriori algorithm of Boolean matrix based on cloud computing platform （such as Hadoop）. Hadoop platform is combined with Apriori algorithm of Boolean matrix to process blocks of Boolean matrix and to calcu-late the frequency of the block by using MapReduce, and to obtain frequent itemsets of big data by means of combination and integration. The analysis shows that the MR_B_Apriori algorithm can be applied to frequent itemset mining for big data.

关 键 词：大数据 HADOOP 数据挖掘 APRIORI算法 关联规则 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]