ADAPTING AI CHATBOTS TO THE UZBEK LANGUAGE FOR EDUCATIONAL PURPOSES
Keywords:
Uzbek language, educational chatbot, large language models, low-resource languages, tokenization, transliteration, retrieval-augmented generation, natural language processing, school education, artificial intelligence.Abstract
Large language model chatbots are entering school education, yet most of them were built around English and work noticeably worse in Uzbek. This article examines what has to be changed in such a chatbot before it can serve Uzbek-speaking pupils reliably. The linguistic and technical obstacles are analyzed first: agglutinative morphology, the parallel use of Latin and Cyrillic scripts, inconsistent apostrophe characters in the letters oʻ and gʻ, code-switching with Russian and the shortage of educational data. An original experiment then measures how four widely used tokenizers process 30 parallel school-domain sentences in English and Uzbek. Uzbek in Latin script required 1.63–2.04 times more tokens than English, Cyrillic Uzbek up to 5.24 times more, and the choice of apostrophe character alone changed token counts by up to 5.7%. On this basis a five-layer adaptation framework is proposed, covering text normalization, safety, curriculum-based retrieval, an Uzbek-adapted language model and a pedagogical response policy, together with criteria for its evaluation.References
1.Musaev M., Mussakhojayeva S., Khujayorov I., Khassanov Y., Ochilov M., Varol H.A. USC: An Open-Source Uzbek Speech Corpus and Initial Speech Recognition Experiments // Speech and Computer: SPECOM 2021. – Springer, 2021. – P. 437–447.
2.Salaev U., Kuriyozov E., Gómez-Rodríguez C. A machine transliteration tool between Uzbek alphabets // CEUR Workshop Proceedings. – 2022. – Vol. 3315. – arXiv:2205.09578.
3.Fozilov E., Urinov A., Juraev K. Llama-3.1-8B-Instruct-Uz: model card. – Hugging Face, 2024. – URL: https://huggingface.co/behbudiy/Llama-3.1-8B-Instuct-Uz
4.alloma-8B-Instruct: Uzbek-optimized Llama model, model card. – Hugging Face. – URL: https://huggingface.co/uzlm/alloma-8B-Instruct
5.Bobojonova L., Akhundjanova A., Ostheimer P., Fellenz S. BBPOS: BERT-based Part-of-Speech Tagging for Uzbek. – arXiv:2501.10107, 2025.
Downloads
Published
Issue
Section
License

This work is licensed under a Creative Commons Attribution 4.0 International License.
You are free to:
- Share — copy and redistribute the material in any medium or format for any purpose, even commercially.
- Adapt — remix, transform, and build upon the material for any purpose, even commercially.
- The licensor cannot revoke these freedoms as long as you follow the license terms.
Under the following terms:
- Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.
- No additional restrictions — You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.
Notices:
You do not have to comply with the license for elements of the material in the public domain or where your use is permitted by an applicable exception or limitation.
No warranties are given. The license may not give you all of the permissions necessary for your intended use. For example, other rights such as publicity, privacy, or moral rights may limit how you use the material.






