Attribute reduction of interval-valued decision system

In this paper, we establish a dominance relation in interval-valued information systems. With respect to any subset of object set, attribute reduction methods are presented based on substitution of the indiscernibility relation by the dominance relation. Possible and certain decision rules can be derived by possible and uniform distribution functions, respectively. Any part of the decision rules can be identified by these attributes which obtained by the attribute reduction method. The usefulness of each method is illustrated by one example.