一种基于分治策略与位运算的频繁项集挖掘算法
A Frequent Itemsets Mining Based on Divide-conquer and Bittable
易宗剑 1彭月英 1覃晓 1唐涛1
作者信息
- 1. 广西师范学院 计算机与信息工程学院,广西 南宁 530023
- 折叠
摘要
为提高挖掘频繁项集的效率,在垂直数据格式下,结合分治思想提出一种基于分治策略与位运算频繁项集挖掘算法DC-FIMBII。利用分治将数据库中的事务划分为多个非重叠部分,对每一部分采用位运算求交计算支持度,从而减少操作时项集的规模和项集的比较次数。在mushroom、pumsb_star和 T40I10D100K等数据集上,对DC‐FIMBII、Apriori、Eclat、DF‐FIMBII等算法进行比较。实验表明,DC‐FIMBII具有更高的效率。
Abstract
For the purpose of efficiency improvement ,a partition principle frequent itemset mining based on bittable and inverted index (PP‐FIMBII) is proposed in this paper I.t divides database transac‐tions into multiple nonoverlapping sections , and calculates the support counts of each two items through the bit operations .Thus both of the Tidset's quantity and the comparing times can be reduced when operating intersects .We compared the execution time of Apriori ,Eclat in three datasets such as mushroom ,pumsb_star and T40I10D100K .The experiment results show that PP‐FIMBII has a more efficiency .
关键词
垂直数据表示/频繁项集/DC-FIMBII算法/分治策略Key words
vertical data layout/frequent itemset/PP-FIMBII/partition principle引用本文复制引用
出版年
2015