Abstract: Based on existed models, this paper proposes a new one that can reduce checking times of the original database, and points out problems in IncSpan+ and defects in interpret mining based on semi-frequent patterns. It constructs a prefix tree to represent the sequence patterns, and continuously maintains the tree structure by using width pruning and depth pruning. Experimental results show the algorithm has a good performance.

Key words: incremental sequence pattern mining, width pruning, depth pruning, prefix-tree

摘要： 在已有模式的基础上，该文挖掘出了新的模式，减少了挖掘原始数据库次数，指出了IncSpan+算法存在的问题，说明了基于半频繁模式的增量挖掘算法的缺陷，提出了一种增量序列模式挖掘算法。该算法构造了前缀树表示序列模式，并用广度剪枝和深度剪枝维护该前缀树的结构。实验表明，该算法具有良好的性能。

关键词: 增量序列模式挖掘, 广度剪枝, 深度剪枝, 前缀树

CLC Number: 

• TP311.13