【摘要】 虽然FP-Growth算法能够有效地从数据库中挖掘频繁模式,但如何由其挖掘出的频繁模式中高效地产生关联规则仍是一个相当复杂的问题。该文提出了用于组织频繁模式的线索频繁模式树(TFPT)和一个从TFPT中挖掘关联规则的高效算法—最短模式优先算法(SPF)。挖掘模式Y的关联规则时,SPF算法应用了两个优化策略,避免了对大量的不可能成为规则XY-X左部的Y的子集的检查,从而获得了很好的性能。实验表明:与类FP-Growth算法结合时,SPF算法运行速度远远快于Apriori算法,并有相当好的可伸缩性。更多还原

【Abstract】 FP-Growth algorithm can mine frequent patterns from database effectively,but how to generate association rules from its resulting patterns efficiently remains a complicated problem.This paper proposes a data structure,called threaded frequent pattern tree(TFPT),to maintain frequent patterns,and an efficient algorithm,called Shortest-Pattern-First(SPF)algorithm,to mine association rules from TFPT.While mining association rules from pattern Y,two optimization strategies are used to avoid examining a lot of subsets of Y that can’t be the left of any association rule of the form XY-X.In this way,SPF algorithm achieves a good performance.Our experiments shows that combined with a FP-Growth-like algorithm,SPF runs far faster than Apriori algorithm and has excellent scalability.更多还原