基于多元索引后继树的序列模式挖掘方法

Method of sequential pattern mining based on Multi-Index Successive Tree

序列模式挖掘是数据挖掘研究领域中的一个课题。本文提出一种新型的序列挖掘模型—多元索引后继树模型,该模型的主要特点是使用索引方法,仅通过对原始序列的一遍扫描而创建描述序列的多元索引后继树,然后利用索引结构使用模式增长的方法生成频繁模式。理论分析与实验表明,该方法简单高效、具有很好的实用价值,通过与其他序列挖掘模型的实验比较,发现基于该模型的挖掘算法在时间和空间两个方面都有较好的性能优势。

Sequential pattern mining was a very important topic in data mining area. In this paper a new model of sequential pattern mining --the Multi-Index Successive Tree （MIST） model and its implementation were put forward. The main characteristic of the model was that the index was used to create MIST through scanning the sequence by one time, the pattern growth was used by index structure to generate frequent patterns. Results of theory analyses and experiments showed that this method was simple, efficient, and had practical application value.