基于MapReduce的改进的Apriori算法及其应用研究  

Improved Apriori Algorithm and Its Application Based on MapReduce

作　　者：赵月[1] 任永功[1] 刘洋[1] 

机构地区：[1]辽宁师范大学计算机与信息技术学院,大连116029

出　　处：《计算机科学》2017年第6期250-254,共5页Computer Science

基　　金：国家自然科学基金项目(F020806);辽宁省高等学校优秀人才支持计划项目(LR2015033);辽宁省科技计划项目(2013405003);大连市科技计划项目(2013A16GX116)资助

摘　　要：随着移动通信和互联网技术的迅猛发展,如何高效地分析移动用户的需求并及时推送有用信息成为数据挖掘领域的热点之一。针对上述问题,提出一种基于云计算Hadoop平台的分布式关联规则MRS-Apriori算法。该方法在经典Apriori算法的基础上优化了数据库编码规则,增加了判断标记Judgemark来判断事务项是否频繁,提高了MRS-Apriori算法在连接时扫描数据库的效率。在编码的基础上,采用Hadoop平台下的MapReduce编程框架模型实现并行化处理,提高了迭代时连接步骤的效率,降低了大规模数据样本运算的时间开销。实验结果表明,改进的MRS-Apriori算法可以有效地减少运算时间,在处理大规模数据集上具有较高的准确性。With the rapid development of mobile communications and Internet technology,it becomes one of the hot issues in the field of data mining that how to analyze the requirements of mobile users efficiently and send useful informations in time.In order to recommend the analysis result to users efficiently and timely,a mining method named MRSApriori algorithm based on MapReduce was proposed.This method defines a kind of coding rule to optimize database based on classical Apriori algorithm.A judging mark named Judgemark is added to database to decide whether the transaction database is frequent.This mechanism improves the efficiency of MRS-Apriroi algorithm in connecting database to scan database efficiently.On the basis of encoding rules,the MRS-Apriroi algorithm uses MapReduce programming framework model under Hadoop to achieve parallel processing.It improves the performance of iteration when connecting process and reduces the time in dealing with large-scale data.The experiment results show that MRS-Apriroi algorithm can effectively reduce time and have high accuracy in handling large data sets.

关 键 词：编码规则 关联规则 频繁项集 MAPREDUCE框架 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]