中文摘要：

根据分层递阶约简算法,提出了一种直接在不完备信息系统上进行数据挖掘的方法。该方法首先将信息系统中由所有属性构成的单层知识表示转变成由部分属性所构成的多层知识表示,即由完备属性和不完备属性表示;然后建立了两个不同层次的子系统,并推导出各个子系统的规则集;最后,将该方法应用于心脏病诊断系统的研究。仿真结果证明,该方法具有较强的实用性和有效性,并能提高知识约简的速度。

英文摘要：

Based on the hierarchical reduction algorithm, this paper proposed a data mining method based on rough sets theory for incomplete information system. Knowledge was presented hierarchically with multiple layers. The attributes were firstly partitioned into complete parts , incomplete parts and two sub-systems with various levels were created accordingly, Then the reduction was hierarchically applied to each sub-system. Finally, the method was applied to the diagnosis system of heart disease, Simulations show that the method has a strong applicability and a more rapid reduction speed. Therefore the effectiveness of the method is verified.