摘要
现有的事件序列情节模式挖掘算法多是基于滑动窗口和非重叠出现的。目前没有有效算法挖掘基于互异出现的互异情节模式。为此,提出带状态计数的前缀树结构(SCTree)来生成互异情节模式候选集,进行互异计数和裁剪。为减少数据库扫描次数,提出SCTree的主动扩展技术。实验表明了算法的有效性和高效性。
Most of existing mining algorithms for event sequence episodes are sliding window-based and non-overlapped occurrences based.There is no efficient algorithm to mine the distinct episodes that is distinct occurrences-based.This paper introduces a novel state-counted prefix-tree(SCTree) for generating candidate set of distinct episodes and to make distinction counting and pruning.In order to reduce the scanning of database,an eager extension technology of SCTree is proposed.Experimental results show the efficiency and effectiveness of the proposed algorithm.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第3期177-181,共5页
Computer Applications and Software