Аннотация:This paper focuses on removing information redundancy from a covering information system. First, we review the common covering reduction methods and discussed their relationship to each other. Second, we consider the explicit and implicit values of compound attributes in order to lay the basis of our work. Then we obtain a network topology of covering. Third, we prove that network topology is a set of bases in a covering information system through an example, in which we obtain all the irreducible elements. Finally, we discuss the network topology-based attribute reduction method, as well as compare it with other methods. Our discussions enrich the existing attribute reduction theories and methods.
