Dominance-based fuzzy rough sets in multi-scale decision tables

Aiming at the situation that fuzzy condition attribute values in multi-scale decision table have dominance relations and the decision attribute values are fuzzy, we establish dominance-based fuzzy rough set model (DFRS) in multi-scale decision table (MSDT). In order to investigate the knowledge acquisition efficiency of DFRS in MSDT, we give the optimal scale selection and reduction method to obtain all the optimal scales and all the optimal scale reducts. Besides, we also propose a simple algorithm to obtain an optimal scale reduct. Finally, we verify the effectiveness and practicability of our method through an example of information system security audit risk judgment and a comparative experiment. Experimental results show that our method has obviously improved the knowledge acquisition efficiency compared with the traditional dominance-based fuzzy rough set and effectively integrates the optimal scale selection of the multi-scale decision table with attribute reduction.