Fuzzy Similarity-Driven Three-Way Decision Model of Rough Fuzzy Sets

Three-way decision (3WD) theory provides an innovative approach to categorizing uncertain issues into acceptance, rejection, and non-commitment regions. The rough fuzzy sets (RFS) model extends rough sets to handle imprecise or fuzzy concepts. A critical challenge in developing the three-way decision model of rough fuzzy sets (3WDRFS) lies in determining and interpreting threshold pairs. However, the traditional expert-based subjective risk parameters often lead to significant misclassification errors in 3WD. To address this limitation, we propose a novel fuzzy similarity-driven 3WDRFS framework (3WDRFS-FS). Our approach involves two key steps: (1) developing a three-way approximation shadowed set (3WA-SS) model using average-step fuzzy sets (AFS), and (2) formulating an objective function to quantify the fuzzy similarity between AFS and 3WA-SS, which is utilized to derive the optimal threshold pair for the 3WDRFS-FS model. Relevant experimental results validate the effectiveness of our proposed model. Moreover, the 3WDRFS-FS model demonstrates superior performance compared to the 0.5-approximation model (0.5-AM).