Knowledge Reduction Based on Binary Discernibility Matrix in Variable Precision Rough Set

Attributes reduction is one of the most important subjects of knowledge discovery in information systems. As an approach of attributes reduction, binary discernibility matrices have many interesting properties. Due to the disadvantage of binary discernibility matrices in dealing with variable precision rough set (VPRS), this paper presents an approach to knowledge reduction based on the binary discernibility matrix which is redefined formally based on traditional binary discernibility matrices, and is called the extensive binary discernibility matrix. From the practice point of view, two-valued decision attributes are general, our discussions are concentrated on this kind of decision information system (DIS), some conclusions about attributes reduction and computative examples are given in this paper