一种更新频繁项目集的快速算法

数据库的更新会引起数据库中的关联规则的更新,找出更新后的所有的频繁项目集,也就能生成更新后的关联规则,因此关联规则的更新就转化为频繁项目集的更新。UWEP算法 利用以前的挖掘结果来减少挖掘新的频繁项目集的开销,采用了一些优化技术来减少数据库的扫描次数和候选项目集的数量,但UWEP算法只能处理增加新事务的情况。本文提出 的UWEP2算法是UWEP算法的扩展,能处理数据库中事务的增加、删除、修改等情况。我们将它与另一种更新频繁项目集的算法FUP2比较,实验显示,UWEP2算法比FUP2算法生成的候选项目集要少,性能要高。

An Efficient Algorithm to Update Frequent Itemsets

While new transactions are being appended to a database and obsolete ones are being removed,association rules already discovered also have to be updated.After finding out all frequent itemsets,finding out all the association rules will be straightforward.Thus,major research efforts have been spent on updating frequent itemsets.A previously proposed algorithm UWEP makes use of the previous mining results to cut down the cost of finding the new rules in an updated database.But UWEP can only handle the problem in the case of insertion.The proposed algorithm UWEP2 is a generalization of UWEP that handles insertions to and deletions from an existing set of transactions.The experiments on synthetic data show that UWEP2 outperforms another algorithm FUP2 in terms of the generated candidates and efficiency.