基于粗集（ＲＳ）理论，针对知识表达系统提出一种新的归纳学习方法。对该方法中条件属性的简化，核值表的求取，决策规则的约简进行了详细讨论，并给出相应的求解算法。本方法为机器学习以及从数据库中进行机器发现提供了新的思路。 Based on rough sets (RS) theory, a new inductive learning method is proposed for knowledge representation system. The simplification of the condition property, the calculation of the nuclear value table and the reduction of decision rules are discussed in detail, and the corresponding algorithm is given. This method provides a new idea for machine learning and machine discovery from the database.