Typed Abstention for LLM-Elicited Fuzzy Criteria in Multicriteria Transportation Decisions

Authors

• Maikel Maikel Leyva-Vazquez, Universidad Bernardo O'Higgins, Santiago, Chile; Universidad Bolivariana del Ecuador, Duran, Ecuador; Universidad de Guayaquil, Guayaquil, Ecuador
• Nezir Aydin, College of Science and Engineering, Hamad Bin Khalifa University, Qatar

Abstract

Fuzzy multicriteria decision-making (MCDM) increasingly draws its criterion evaluations from large language models (LLMs), but the aggregation step typically treats these elicited judgments as interchangeable with expert opinion, with no mechanism to flag disagreement, insufficiency, or unreliable sources. We port a source-preserving, typed-abstention decision policy into fuzzy-MCDM pipelines, instantiated on a four-alternative sustainable-transportation case, and report two live pilots. Pilot 1 returns all-coherent; Pilot 2, redesigned with criteria of greater normative dependence, shows the gate firing on real disagreement. We report both pilots candidly and verify the gate's logic with synthetic unit tests.

Recommended Citation

Leyva-Vazquez, Maikel Maikel and Nezir Aydin. "Typed Abstention for LLM-Elicited Fuzzy Criteria in Multicriteria Transportation Decisions." Neutrosophic Sets and Systems 101, 1 (2026). https://digitalrepository.unm.edu/nss_journal/vol101/iss1/21