Abstract:

From the view of attribute sets, the attribute reduction in the variable precision rough set model is studied. A new β lower approximation binary relation is defined. Using relations the equivalence relation is constructed on attribute sets and its power sets, and a dependence space is also produced. By applying the binary relation and dependence space,  judgement theorems for judging β lower approximation consistent sets are obtained, from which we provide a new attribute reduction algorithm. It can preserve β lower approximation in every decision class. Finally, a practical example illustrates the validity of the method.
 

Key words: variable precision rough sets; attribute reduction; β lower approximation binary relation; dependence space