不完备多尺度决策信息系统的局部泛化约简

Generalization Reducts in Incomplete Multi-scale Decision Tables

在不完备多尺度决策信息系统中,研究面向单个对象的局部泛化约简方法。所提的局部泛化约简方法融合了不同特征的不同尺度上的信息,且保持了广义决策值的一致性。在协调的不完备多尺度决策信息系统中,证明了两种泛化约简方法的一致性,给出了相应的的局部泛化约简算法。在不协调多尺度决策信息系统中,研究了其详细性质,并通过实例说明两种泛化约简的区别。

In incomplete multi-scale decision tables, a type of local generalization reduct method is investigated. The proposed approach integrates the information of different attributes at diffirent scales, and keeps the generalized decision unchanged. In consistent incomplete multi-scale decision tables, two types of generalization reduct methods are shown to be equivalent, and an algorithm of searching for generalization reducts is designed. In inconsistent multi-scale decision tables, the properties are examined in detail and an example is employed to show the differences between two types of generalization reducts.