Abstract: Firstly，this paper indicates the localization of single-support Apriori algorithm，and analyzes the imperfection of multiple supports Apriori algorithm that is proposed for overcoming the localization of single-support Apriori algorithm.Aiming at some latent rules of affair，the paper proposes a sub-support Apriori algorithm.It does not simply expand or improve the classical Apriori algorithm.The proposed algorithm destroys Apriori algorithm’s qualities—global and high frequency in theory，and uses the method of sub-support to data mining for the database.It discovers that classical and multiple supports Apriori algorithm are difficult to find the strong relevancy regulation.The proposed algorithm can be realized for a quick speed.

Key words: subsection, support, Apriori algorithm, colony character, data mining

摘要： 首先指出单支持度的Apriori算法的局限性，分析了目前为克服单支持度Apriori算法的局限性而提出的多支持度的Apriori算法的不完备性，针对事务中的一些潜在规则，提出了一种分段支持度Apriori算法。算法不是简单地对经典Apriori算法进行扩展或改进，而是从理论上破坏了Apriori算法全局、高频两个性质，采用分段支持度的方法对数据库进行数据挖掘，可以发现经典和多支持度Apriori算法不能发现或很难发现的强关联规则，并以较快的速度得以实现。

关键词: 分段, 支持度, Apriori算法, 群体特征, 数据挖掘