【作者】 陈子阳；

【导师】 郝忠孝； 刘国华；

【作者基本信息】 燕山大学 ， 计算机应用技术， 2000， 硕士

【摘要】 数据挖掘是指从大量的数据中发现潜在的，有用的知识的过程，是解决“数据丰富、信息贪乏”的有效方法，关联规则是数据挖掘的主要研究内容。 已有对关联规则的研究只注重解决算法的时间效率，而忽视了关联规则的多层次性。同时，关联规则只用原始数据表示，由于支持度较低而难以表示数据之间的普遍联系。 本文针对已有方法的不足，利用归纳的抽象的概念层次提出了基于概念层次树的多层次关联规则算法，根据先验估计以概念层次树的中间层次为起点，在计算结点的支持度和可信度的同时对结点之间进行匹配以更高效的发现多层次关联规则。 算法有以下优点：1）高效。与其它方法比较具有较低的空间要求，而且速度更快，从而可充分利用数据，得出准确的知识；2）挖掘出的关联规则是多层次的，同时对得出的多层次关联规则进行清洗，使得到规则更加准确。更多还原

【Abstract】 Data mining is a kind of process that reveals potential useful knowledge from massive, it is an effective way to tackle “Data Rich and Information poor” Association rules are an important aspect of research of DM. The existing research to the association rules only emphasis on the resolving of time efficiency of the algorithm, while ignoring the multiple-leveled performance of the association rules. At the same time, the association rules only represented by raw material, due to the lower supporting degree and it’s hard to represent the universal association between data. This dissertation, in the light of the defects of other methods. Sagest an algorithm based on conception hierarchy tree for mining multiple-leveled association rules, based on initiates from a certain intermediate level of the conception hierarchy, matching discover knowledge more efficiency at the same time of supporting degree and reliability of the counting node. The algorithm has shown advantages such as: 1) high efficiency. Comparing with other approaches, it reaches higher performance with less spatial requirements, witch makes it possible to exploit data thoroughly and discover knowledge accurately. 2) Mining association rules are the multiple-leveled association rules, and clean out the gained multiple-leveled association rules, to make the gained rules more accurate.更多还原