An Improved Variable Precision Model of Dominance-Based Rough Set Approach

摘要

The classification, ranking and sorting performance of Dominance-based rough set approach (DRSA) will be affected by the inconsistencies of the decision tables. 'Two relaxation models (VC-DRSA and VP-DRSA) have been proposed by Greco and Inuiguchi respectively to relax the strict dominance principle. But these relaxation methods are not always suitable for treating inconsistencies. Especially, some objects which should be included in lower-approximations are excluded. After analyzing the inadequacies of the two models, an improved variable precision model, which is called ISVP-DRSA, based on inclusion degree and supported degree is proposed in this paper. The basic concepts are defined and the properties are discussed. Furthermore, the lower approximations of ISVP-DRSA are the union of those of VC-DRSA and VP-DRSA, and the upper approximations are the intersection of those of the two models. 'Then more objects will be included in lower approximations and the quality of approximation classification is not poor than the above two models. Finally, the efficiency of ISVP-DRSA is illustrated by an example.