一种基于事务修剪的约束关联规则的挖掘算法

Constraint association-rule mining algorithm based on transaction clip

针对一类常见而简单的规则中有项或缺项的约束,提出了一种基于事务数据修剪的约束关联规则的快速挖掘算法。该算法先扫描一遍数据库对事务进行水平和纵向的修剪,接着在修剪后的数据集上挖掘频繁项集,形成规则的候选头集、体集和规则项集,最后一次扫描后由最小可信度约束得到所要求的关联规则。实验表明,与按简洁约束采取的一般策略相比,该算法的性能有较明显的提高。

Aiming at a familiar and simple constraint that some items must or must not present in rules, a fast clippedtransaction-based constraint association-rule mining algorithm was put forward, This algorithm firstly scanned data base to clip transactions horizontally and vertically, then mined frequent item sets from clipped data set to form rules＇ candidate head sets, body sets and rule item sets. Finally, it scanned original data base again to gain association rules according to minimum confidence constraint. Experiments show that, compared with common strategy of succinct constraint, this algorithm has better performance.