An explainable risk classification model by integrating fuzzy multiple criteria sorting and fuzzy linguistic summarization

Durmaz E. D., AYDOĞAN S., Gölcük İ.

Applied Soft Computing, vol.181, 2025 (SCI-Expanded, Scopus) identifier identifier

  • Publication Type: Article / Article
  • Volume: 181
  • Publication Date: 2025
  • Doi Number: 10.1016/j.asoc.2025.113515
  • Journal Name: Applied Soft Computing
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Applied Science & Technology Source, Compendex, Computer & Applied Sciences, INSPEC
  • Keywords: Explainable artificial intelligence, Fuzzy linguistic summarization, Multiple criteria decision making, Risk assessment
  • Gazi University Affiliated: Yes

Abstract

This study proposes a novel explainable risk assessment framework that integrates the fuzzy Full Consistency Method (FUCOM), fuzzy VlseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR)-Sort, and fuzzy linguistic summarization to address the challenges of uncertainty, prioritization, and interpretability in risk evaluation. The model enables multi-criteria analysis under linguistic uncertainty, classifies risks into predefined categories, and generates human-readable linguistic summaries to support decision-makers. The proposed methodology is applied to a real-world construction case involving 32 risk items, which are categorized into high, medium, and low risk groups. Comparative analysis with six established multiple criteria sorting algorithms reveals strong alignment in classification outcomes. The novelty of the proposed model lies in enhancing the explainability of multiple criteria sorting algorithms by integrating fuzzy linguistic summarization into the classification process—an aspect largely overlooked in the existing literature. The results demonstrate that the proposed model not only produces consistent risk classifications across alternative methods but also enhances decision-makers’ understanding through interpretable linguistic outputs.