A novel rough set method based on adjustable-perspective dominance relations in intuitionistic fuzzy ordered decision tables

Abstract

With the increasing diversification and complexity of information and data, it is crucial to monitor and process data and information from multiple perspectives. There is a general consensus that dominance-based rough set approaches are the most effective methods for ordered information system research. Regardless, limitations and irrationality still exist in dominance relations because they cannot reflect the different emphases of data under different feature sets, nor can they meet the requirements for describing data information in the real world. To enable the dominance relation rough set model to be more effectively applied to practical problems in line with human cognition, our work focuses on developing adjustable-perspective dominance relations by fusing three different dominance relations in an intuitionistic fuzzy ordered decision table. On this basis, we construct a single-perspective rough set (SPRS) model to realize knowledge mining and rule extraction from various perspectives. Additionally, we present two types of different-perspective rough sets (DPRS) that reduce the restriction of single-perspective evaluation data in realistic problems and discuss rule extraction. Additionally, we compare SPRS and DPRS to other dominance-based rough set models from the perspectives of the ordinal classification, roughness, and dependence degree. Finally, we analyze eight UCI datasets and present a series of comparative experiments to demonstrate the effectiveness and rationality of the proposed model.