摘要： 针对一些经典的关联规则算法进行分析，提出一种基于矩阵的高效关联规则挖掘算法。该算法把交易数据库转化为0-1矩阵形式，只需进行一次数据库搜索，使用逻辑运算方法发现频繁项集并计算它的支持计数，加快了频繁k项目集的验证速度，能大量减少所需的I/O次数，减小了存储空间。实验表明，新算法执行效率明显优于Apriori算法。

关键词: 关联规则, 频繁项集, Apriori算法, 矩阵

Abstract: This paper analyzes some classical algorithm of association rule mining, and gives a high efficient algorithm based on Matrix. The method puts the database changing 0-1 matrix and needs only one pass over database, uses logical operation to discover frequent itemsets calculates it is support, accelerates the verification speed of the frequent k-item sets, reduces I/O overheads greatly and the store space. Test result shows that the performance efficiency of new algorithm is obviously better than Apriori algorithm.

Key words: association rules, frequent items, Apriori algorithm, matrix

中图分类号: 

• TP311.13