A new reduction model of inconsistent decision table

Two reduction definitions have been introduced in classical rough set theory. One is an algebra definition, and the other is an information definition. However, the two definitions both regard accuracy of rules as criteria of reduction and the former only cares accuracy of deterministic rules, the latter cares accuracy of entire rules. Therefore, for inconsistent decision table, the reduction results that we obtain by using the two definitions respectively could be different. This paper presents a new reduction model that its criteria of reduction combine accuracy and coverage of rules. Correspondingly, this paper proposes two new reduction definitions under the new model. By analyzing and testing, for inconsistent decision table, the reduction results that we obtain by using the two new definitions respectively are uniform. In addition, the reduction results with respect to the new reduction model may arrive at a better effect.