基于静态IS-树的频繁模式挖掘

Mining Frequent Patterns Based on Static IS-Tree

提出一种基于静态IS-树的频繁模式挖掘有效算法IS-mine,并与经典的Apriori算法和FP-growth算法进行了实验比较。算法直接构造频繁项集,不进行Apriori算法采用的代价较高的候选集产生与测试操作。算法采用深度优先,模式增长的策略,挖掘任务只在一棵静态的IS-树上进行,避免了FP-growth算法所采用的代价较高的动态树的构建。针对不同特征数据集算法采取不同的过滤技术来缩小搜索空间。实验与理论分析表明,对于稠密和稀疏数据两类数据集,算法都具有较好的时空效率。

In this paper, an algorithm is presented for mining frequent patterns based on a static IS- tree. The algorithm builds frequent patterns directly, instead of high-cost candidate sets generation- and-test method used by Apriori. It gen approach, and works on a static IS-tree, In order to reduce search space , it characteristics of datasets. Our perform both dense datasets and sparse datasets. erates frequent rather than a c uses different ance study and patterns by depth first and pattern growth ostly dynamic tree adopted by FP-growth. filter technologies according to different theory analysis show that it is efficient in