Matroidal approaches to generalized rough sets based on relations

Rough set theory is a useful tool for dealing with the vagueness, granularity and uncertainty in information systems. This paper connects generalized rough sets based on relations with matroid theory. We define the upper approximation number to induce a matroid from a relation. Therefore, many matroidal approaches can be used to study generalized rough sets based on relations. Specifically, with the rank function of the matroid induced by a relation, we construct a pair of approximation operators, namely, matroid approximation operators. The matroid approximation operators present some unique properties which do not exist in the existing approximation operators. On the other hand, we present an approach to induce a relation from a matroid. Moreover, the relationship between two inductions is studied.