中文摘要：

关联分类通常产生大量的分类规则，导致在分类新实例时经常产生规则冲突问题。针对这种规则冲突问题，提出了一种基于改进关联分类的两次学习框架。利用频繁且互关联的项集产生分类规则改进关联分类算法，有效减少了规则数。应用改进的关联分类算法产生的一级规则一次性分离出训练集中规则冲突的所有实例。然后，在冲突实例上应用改进的关联分类算法进行第二次学习得到二级规则。分类新实例时，首先利用第一级规则进行分类。如果出现规则冲突，则利用第二级规则分类该实例。实验结果表明，基于改进关联分类的两次学习方法降低了规则冲突比率，并且显著提高了分类准确率。

英文摘要：

Associative classification usually generates numerous rules, resulting in rule conflicts in stage of classification. To address this problem, a double learning method based on the improved associative classification is proposed. The improved associative classification reduces the number of rules significantly by discovering the frequent and mutual associated itemsets. All training conflict instances in training set are separated by applying the first level rules generated by the improved associative classifi- cation. Then, the second level rule set is induced by applying the improved associative classification on the conflict instances. When classifying a new instance, the first level rule set is applied. If the rules are not consistent with the instance, the second rules set is used to classify this instance. The experimental results show that the double learning method based on the improved associative classification can im prove the classification accuracy effectively.