Order-Based Decision Rules Acquisition in Continuous-Valued Decision Information Systems

In a continuous-valued information system, attribute values of objects represent not only the relative distances, but also the ordinal relation of objects. By considering the relative distances of objects, some discretization approaches and other approaches were proposed in previous literature, to discuss the attribute reduces of the system. This paper considers ordinal relations of objects, and utilizes dominance relation rough set model to discuss decision rules acquisition problems. Firstly, concepts of order-based decision rules and generalized order-based decision rules are introduced. Then the concept of optimal order-based decision rule is defined. In order to compute optimal order-based decision rules, the concept of reduct of the object is proposed. Lastly, discernibility functions of objects are constructed, and used to compute optimal order-based decision rules by Boolean reasoning techniques.