Knowledge Reduction of Incomplete Information Systems with Negative Decision Rules

In this paper, within rough set theory we study attribute reduction of incomplete decision tables (IDT). The concept of negative support of descriptors in an IDT is proposed firstly, from the lower and upper approximations of decision descriptors, certain and possible decision rules can be induced. By use of the lower and upper approximations, for simplifying the decision rules, two types of attribute reduction of IDT are considered, and methods of attribute reduction are given by employing discernibility attribute sets.