基于MapReduce的高效用序列模式挖掘算法  

High Utility Sequential Pattern Mining Algorithm Based on Map Reduce

作　　者：程思远[1,2] 马超[1,2] 李聪聪[1,2] 

机构地区：[1]复旦大学计算机科学技术学院,上海201203 [2]上海市数据科学重点实验室(复旦大学),上海201203

出　　处：《计算机系统应用》2015年第12期228-232,共5页Computer Systems & Applications

摘　　要：由于数据规模的快速增长,高效用序列模式挖掘算法效率严重下降.针对这种情况,提出基于Map Reduce的高效用序列模式挖掘算法Hus Ma R.算法基于Map Reduce框架,使用效用矩阵高效地生成候选项;使用随机映射策略均衡计算资源;使用基于领域的剪枝策略来防止组合爆炸.实验结果表明,在大规模数据集下,算法取得了较高的并行效率.Because of the rapid growth of data, the high utility sequential pattern mining algorithms＇ efficiency decreases seriously. In view of this, we propose a high utility sequential pattern mining algorithm based on MapReduce, namely HusMaR. This algorithm is based on MapReduce, which using the utility matrix to generate candidate efficiently, random mapping strategy to balance of computing resources and field-based pruning strategy to prevent an explosion. Experimental results show that in the large scale of data, the algorithm achieves a high parallel efficiency.

关 键 词：序列模式 MAP REDUCE 剪枝策略 高效用序列模式挖掘 随机策略 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]