中文摘要：

Apriori算法是关联规则挖掘中的经典算法，算法的核心思想是一种基于频繁理论的自底向上的递推方法。文中对Apriori算法进行分析，发现其中存在的问题。对Apriori算法做了改进。改进后的算法基于自顶向下的思想。利用有向图给出计算候选项集和项集支持度计数的更快的方法，同时简化了Apriori算法的连接和剪枝操作，从而在时间和空间上提高了Apriori算法的效率。

英文摘要：

Apriori is classical in association rules mining. The core of the algorithm is a recursive Bottom-up approach based on the frequent theory. An analysis of the Apriori algorithm indicated its some problems in this article, and provides an improvement on Apriori algorithm. The improved algorithm is based on the Top-down thinking. It can compute the candidated itemsets and the count of support of itemsets more quickly by directed graph. It can also simplify the join step and the prune step in Apriori algorithm. Thus the efficiency of Apriori algorithm gets improvement both in time and in space.