一种FP-growth的改进算法

An improvement algorithm for FP-growth

FP-growth是关联规则挖掘中一种效率较高的算法,它不产生候选集,但仍需多次遍历结果集L。针对此问题提出了一种基于Hash表的改进算法HFP-growth(Hash FP-growth)。该算法将结果集L的数据以项名称对应项支持度计数的形式存入Hash表,在找某个项的支持度计数时给Hash表传入项名称直接返回对应的支持度计数,改变了以往多次遍历结果集L的方法,从而节省了遍历时间,提高了挖掘效率。实验结果表明,改进后的算法性能明显优于原算法,并将其应用于名智网上招聘系统之中。

In the association rule＇s mining, FP-growth algorithm is one of the high effective algorithm, although it does not produce the candidate sets, still needs traversal result sets L many times. To solve this problem, an improved algorithm HFP-growth （Hash FP-growth） was proposed. HFP-growth stored data in the result sets L into the hash table by its item name which corresponded to its support counts. HFP-growth algorithm outperform FP-growth and can be applied in Online Recruitment System.