2026
doi: 10.12720/jait.17.11.2160-2171
Automated Detection of Cognitive Distortions in Azerbaijani for Mental Health Applications
Laman Guluzada1,2,3,*
,
Samir Rustamov1,3
- 1School of Information Technology, ADA University, Baku, Azerbaijan
- 2School of Engineering and Applied Science, The George Washington University, Washington, DC, USA
- 3MegaSec AI LLC, Baku, Azerbaijan
- Manuscript receivedMay 28, 2026
- revisedJuly 24, 2026
- acceptedSeptember 18, 2026
- publishedNovember 12, 2026
Abstract
This paper examines cognitive distortion classification in Azerbaijani under low-resource conditions. We adapt the Chinese C2D2 dataset and the English Therapist Question-and-Answer dataset through structured translation, and we construct a consensus-filtered synthetic Azerbaijani dataset. The translation workflow combines fixed label mappings with automated field-level quality checks. Across lexical, multilingual transformer, Qwen-based, and hybrid classifiers, dataset structure was as influential as model choice. The hybrid Term Frequency-Inverse Document Frequency (TF-IDF) + multilingual Bidirectional Encoder Representations from Transformers (mBERT) model achieved 0.6762 accuracy and 0.6276 macro-averaged F1-Score on the controlled C2D2-AZE benchmark, whereas performance was markedly lower on Therapist Question-and-Answer (Q&A) dataset. Qwen2.5-3B-Instruct supervised fine-tuning (SFT) achieved 0.8014 accuracy and 0.7793 macro-averaged F1-Score on Synthetic-AZE, showing that the task is learnable under curated conditions. Distorted-span extraction and retrieval-augmented reframing are included as exploratory proof-of-utility extensions. Overall, the study shows that Azerbaijani cognitive distortion detection is feasible, but that meaningful progress depends on careful data construction, label stability, and dataset-aware evaluation.
Keywords
- cognitive distortion detection
- Azerbaijani Natural Language Processing (NLP)
- low-resource NLP
- mental health NLP
- text classification
- Qwen
How to Cite
Laman Guluzada and Samir Rustamov, "Automated Detection of Cognitive Distortions in Azerbaijani for Mental Health Applications," Journal of Advances in Information Technology, vol. 17, no. 11, pp. 2160-2171, 2026. https://doi.org/10.12720/jait.17.11.2160-2171
Copyright & License
Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).
