BIDEFCE:一种基于双向扩展的频繁闭情节挖掘算法

BIDEFCE:An Algorithm for Mining Frequently Closed Episodes Based on Bidirectional Extension

在事件序列上挖掘频繁闭情节时,为避免维护频繁情节集,加快挖掘进度,提出基于双向扩展的频繁闭情节挖掘算法BIDEFCE.该算法基于非重叠的最小发生的支持度定义和深度优先搜索策略,在生成新频繁情节的同时,采用向前和向后扩展检查,尽早判断并淘汰非闭情节,将待定情节加入频繁闭情节超集FCE中.然后再对FCE中的情节进行闭合性检查,保留真正的闭情节.该算法避免维护频繁情节集,只需维护频繁闭情节超集,节省存储空间,提高运行效率.实验证实BIDEFCE算法在事件序列上能有效挖掘频繁闭情节.

In order to avoid maintaining frequent episodes set in event sequences while mining closed frequent episodes and to speed up the progress of mining, this paper puts forward the algorithm BIDEFCE for mining frequent closed episodes based on bidirectional extension. Algorithm BIDEFCE discovers all frequently closed episodes by employing the supportive definition of non-overlapping minimal occurrences and the depth-first search strategy. BIDEFCE uses the forward and backward extension check in the generation of frequent episodes,in order to judge and eliminate non-closed episodes as soon as possible,the other episodes are added into frequently closed episodes superset FCE. The true closed episodes will be reserved after the closure check. Moreover,BIDEFCE saves store space,improves operation efficiency, and avoids maintaining the frequent episodes set at the same time. Experiments have proved that BIDEFCE can effectively mine closed frequent episodes in event sequences.