Attribute significance for F — Parallel reducts

Attribute significance in a family of decision subsystems is defined in this paper, and its properties are discussed. It is the extension of attribute significance for a single decision system. We apply it to obtain parallel reducts, and an algorithm with the attribute significance in a family of decision subsystems is proposed. Experimental results show that the method overmatches the matrix of attribute significance in both time complexity and space complexity as well as the length of reducts. Moreover, a new rough set model called F-rough sets is proposed, it is consistent with parallel reducts.