一种新的多层频繁模式挖掘算法

A new algorithm for mining multiple-level frequent patterns

在多层频繁模式挖掘时,结合映射和并发技术,改进经典的FP-growth算法,提出了多层映射频繁模式增长算法(ML_MFP_Growth)。首先对事务数据库中的项目编码预处理,随后对编码数据库的每一列进行映射,构造各层映射频繁模式树(MFP-Tree),最后并发挖掘各层MFP-Tree,得到所有频繁模式。实验表明,ML_MFP_Growth算法比传统多层频繁模式挖掘算法性能有所提高。

With the combination of mapping and parallel techniques, the algorithm FP-growth is extended to the discovery of the multiple-level frequent patterns. And then a new algorithm is developed, which is called multiple-level mapping frequent patterns growth algorithm, namely ML_MFP_Growth. In the process of mining, firstly the transaction database is dealt with to get an encoded database; secondly each row of the encoded database is mapped to construct the MFP-Tree at all concept levels; finally each MFP- Tree is mined in parallel to generate all the frequent patterns. Finally a performance test is held, which shows that the ML_MFP_Growth algorithm has better performance than some traditional algorithms.