On selection of representative object set for attribute reduction in set-valued information systems

Abstract

In recent years, there have been many researches on tolerance rough set to solve attribute reduction and rule extraction problems in set-valued information systems. For attribute reduction problems, the most important issue is to minimize the time complexity of attribute reduction algorithms. There have been many methods for attribute reduction, but finding reducts in these methods is almost performed on initial object set. In this paper, we propose a method for selection representative object set from initial object set to solve attribute reduction problem in set-valued information systems. Because of the size of representative object set is smaller the size of initial object set, our method reduces significantly the time complexity of attribute reduction algorithms.