基于概念格的频繁闭项集增量挖掘算法研究

Study on FCI mining algorithm based on concept lattice

近年的研究表明,概念格可以应用于解决频繁闭项集的挖掘问题.针对已有渐进式概念格构造算法中存在的问题,提出了一种基于概念格的频繁闭项集增量挖掘新算法——FIPT-I算法.新算法利用模式树对概念格进行组织,并利用模式树压缩数据库中的事务,在渐进式构造概念格的同时实现了事务的批处理,减少了概念格的调整操作时间.实验结果表明,与其他同类算法相比,FIPT-I算法对于增量挖掘频繁闭项集来说具有更高的效率.

It is found in recent frequent closed item-set minin studies that concept lattice can be used to solve the problem of incremental g. In order to solve the problems existing in the proposed incremental frecremental frequent closed item-set mining is proposed. FIPT-I treats new transactions in batches by utili- zing trees to compress new transactions and represent concept lattice, so it avoids adding new transactions one by one and greatly reduces time of concept lattice reconstructing. The experiment shows that the efficiency of FIPT-I outperforms other similar algorithms when treating incremental frequent closed item-set mining problem.