Approaches for dynamically updating set approximations in dominance based rough sets while condition attributes value coarsening and refining

This paper introduces the concepts of attributes coarsening and refining in dominance based rough sets (DBRS), and analyzes the basic knowledge granules' variation while condition attributes' coarsening and refining happen. Incremental approaches are proposed for dynamically maintaining the knowledge in dynamic decision systems with respect to condition attributes' coarsening and refining. Examples are employed to validate the feasibility of proposed approaches.