Framework Neutrosófico AHP–TOPSIS Asistido por una Cadena de Expertos basada en LLMs para la Evaluación del Riesgo Crediticio en PyMES

Abstract

Credit risk assessment in small and medium-sized enterprises (SMEs) is a complex decision-making process due to the coexistence of multiple financial criteria, incomplete information, and high levels of uncertainty. Classical multicriteria decision-making approaches, such as AHP and TOPSIS, provide structure and transparency but do not explicitly model indeterminacy. This study proposes a neutrosophic AHP–TOPSIS framework assisted by a Chain of Experts implemented through Large Language Models (LLMs). The approach is developed under a Design Science Research paradigm, constructing a methodological artifact for SME credit risk evaluation. The framework integrates Neutrosophic Analytic Hierarchy Process (N-AHP) and Neutrosophic TOPSIS (N-TOPSIS) to represent truth, indeterminacy, and falsity in expert judgments. Validation is conducted through a case study using simulated but realistic financial data, and results are compared with a classical AHP–TOPSIS model. Findings show that the neutrosophic approach improves decision robustness and explicitly captures uncertainty without necessarily altering the final ranking. The decision process was supported by a Neutrosophic Chain of Experts implemented through Large Language Models. This work is presented as an exploratory methodological study aimed at demonstrating the feasibility and usefulness of neutrosophic decision-making frameworks assisted by LLMs in credit risk assessment contexts.

Keywords