Similarity-guided and divergence-driven ELECTRE methodology for decision-making within circular intuitionistic fuzzy environments

Building upon the Elimination and Choice Translating Reality (ELECTRE) framework, this study develops a novel similarity-guided and divergence-driven decision-making methodology under Circular Intuitionistic Fuzzy (CIF) environments. The proposed approach introduces an advanced evaluation scheme that refines CIF similarity measures by incorporating favorable and unfavorable comparisons relative to ideal and worst-case benchmarks. To ensure numerical stability while maintaining theoretical soundness, a minuscule positive constant is introduced in the refined similarity formulations. The study further formulates similarity-guided evaluation metrics for both CIF importance weights and evaluative ratings, thereby enabling more accurate and context-sensitive assessment of alternatives. In addition, multiple similarity-guided consistency indicators and divergence-driven inconsistency indicators are developed to assess preference coherence and detect structural disadvantages, respectively. These elements are systematically integrated into enhanced CIF ELECTRE I and II procedures, comprising five phases: deriving criteria weights via a similarity-based approach, examining concordance relations, investigating discordance relations, conducting partial ordering through CIF ELECTRE I, and producing comprehensive rankings using CIF ELECTRE II. Key theoretical properties—such as boundedness, optimality conditions, and asymptotic consistency—are also established to reinforce methodological rigor. The proposed methodology is applied to evaluate Artificial Intelligence (AI)-enhanced Clinical Decision Support System (AI-CDSS) service providers for a leading hospital in Taiwan. Evaluations are conducted across multiple decision criteria, including AI functionality, automation, and clinical workflow optimization. A series of juxtaposed evaluations is undertaken to scrutinize the implications of differing metric configurations on ranking coherence and decisional efficacy, thereby substantiating the resilience and real-world relevance of the CIF ELECTRE framework in intricate, ambiguity-rich contexts. Overall, this work contributes a comprehensive and scalable ELECTRE-based paradigm that advances fuzzy decision analytics under circular uncertainty.