一种基于序列末项位置信息的序列模式挖掘算法

摘要

In order to solve the defects of repeatedly scanning projection database looking for local frequent item and produ-cing, mining large number of duplicated project databases in PrefixSpan algorithm, this paper proposed the SPM-LJPT algo-rithm for sequential pattern mining. By connecting the 2-sequence LIFT (last item position table) , the algorithm found the next item of the sequence, realized sequential pattern growth and avoided repeatedly scanning projection database. At the same time, it also could avoid producing and mining large number of duplicated project databases by checking SLIFPT ( same last i-tem first position table) prior to pruning. Experiments show that the algorithm is effective.%针对PrefixSpan算法中反复扫描投影数据库寻找局部频繁项并重复构造挖掘大量重复投影数据库的不足,提出一种基于序列末项位置信息的序列模式挖掘算法SPM-LIPT.通过连接2-序列位置信息表(LIPT)找到序列模式的下一项,实现序列模式增长,避免对投影数据库反复扫描;同时通过检查相同末项序列首位置信息表(SLIFPT)进行前向剪枝;消除大量重复投影的构建.最后通过实验证明了算法的有效性.