AN EFFICIENT APPROACH WITH MEMORY INDEXING FOR DISCOVERING FREQUENT SEQUENTIAL PATTERNS

摘要：

Mining frequent sequential pattern is a fundamental and important problem in data mining area. The previous algorithms for mining frequent sequential patterns need to repeatedly scan sequence database and take a large amount of computation time to find frequent patterns. However, the discovered frequent sequential patterns may become invalid or inappropriate when the memory cannot be used effectively.Thus, how to rationally utilize memory becomes a key to find frequent sequential patterns. In this paper, we propose a memory indexing approach for mining frequent sequential pattern, named FSP_MI. Our approach scans the sequence database only once to load sequence database into memory.Using memory indexing strategy and index table structure, we can recursively discover the frequent items, which can be used to form frequent sequential patterns. Besides, FSP_MI algorithm doesnt generate candidate subsequences in contrast to previous algorithms. The experimental results and analysis show that FSP_MI algorithm is more efficient than GSP,SPADE and PrefixSpan in execution time cost for mining frequent sequential patterns.

关键词： Frequent Sequential Pattern Memory indexing index table

作者: CAO DAN HUI-LI PENG XIAO-JIAN ZHANG XING-ZHENG DU

作者单位: College of Information Science and Engineering, Henan University of Technology, ZhengZhou 450004, Ch College of Information Science and Engineering, YanShan University, Qinhuangdao 066004, China

会议类型: 国际会议

会议名称: 2006 International Conference on Machine Learning and Cybernetics(IEEE第五届机器学习与控制论坛)

会议地点: 大连

会议语种:英文

页码: 1001-1006

在线出版日期: 2006-08-13（万方平台首次上网日期，不代表论文的发表时间）