中文摘要：

使用粗糙集理论对训练集进行属性约简，再使用决策树算法得到决策树规则；然后，根据规则信息量及规则可信度的概念提出两条规则筛选准则，并将两条准则应用于极小极大规则学习方法，形成有判定的极小极大规则学习。将该算法应用于决策树规则的简化，缩小了简化的范围，并能保证规则覆盖的一致性，且可减少规则的总数量。

英文摘要：

This paper conducts attribute reduction for training set using Rough Set theory, and then obtains the decision tree rules by use of decision tree algorithm. Afterwards, two crite- ria on rule screening are proposed in accordance with the concept of rule information quantity and rule credibility, and the two criteria are applied to minimal and maximal rules learning method, which forms the Judgemental Minimal and Maximal Rules Learning. The simplification of decision tree rules by this algorithm can narrow the scope of simplification, ensure consistency of coverage of the rules, and reduce the total number of rules.