一种基于记录分区的多值关联规则挖掘算法

A Record Partition Based Algorithm for Mining Quantitative Association Rules

根据多值关联规则的属性互不相交的特点 ,提出基于记录分区的多值关联规则的定理和算法 ,利用反倾斜技术 ,在合理的内存支持下 ,该算法能以至多 3次扫描数据库 (I O)的时间复杂度 ,有效地在分布不均的数据记录集中产生高频项目集(frequentitemsets)

We propose an efficient record partition based algorithm for mining quantitative association rules on the basis of the analysis that the precedent attributes and consequent attributes are disjoined. The algorithm can discover frequent itemsets in skew dataset, requiring only three scans over dataset in the worst case.