An Enhanced Self-Information Framework Based on Covering Fuzzy Rough Sets for Manufacturing and Medical Diagnosis

Decision self-information measures are widely used in rough set and fuzzy rough set environments to construct models based on similarity relations, which help avoid inconsistencies between lower and upper approximations. However, these approaches are often limited by data-type restrictions. To address this issue, this study employs fuzzy β-coverings to relax such constraints. Four types of self-information measures are introduced within the fuzzy β-covering framework, and their theoretical relationships are analyzed. An algorithm and a numerical example are provided to demonstrate their effectiveness in attribute reduction. Furthermore, the proposed framework is applied to multi-criteria group decision-making (MCGDM) problems, offering a new strategy for solving medical diagnosis and additive manufacturing design problems. This approach enables experts to analyze large-scale data from a granular perspective. Comparative experiments confirm the effectiveness and reliability of the proposed method.