TARJIMA XATOLARINI ANIQLASH VA TAHRIRLASHDA SUN’IY INTELLEKTDAN FOYDALANISH: CHATGPT-4 VA DEEPSEEK BILAN QIYOSIY TAJRIBA
Detecting errors in the translation process and editing them has always been an important and responsible task. In the traditional approach, this work was solely entrusted to human translators and editors, but with the development of artificial intelligence in recent years, the situation has changed dramatically [1]. Artificial intelligence tools, particularly large language models, have brought about significant changes in translation practice over the past two years [2]. Based on practical experiments conducted in this article, the capabilities of two most popular models – OpenAI's ChatGPT-4 and DeepSeek – for detecting and editing translation errors are comparatively analyzed
1. Abdurahmonov, Sh. (2023). Neyron tarmoqlar asosida tarjimani post-tahrirlash: muammo va yechimlar. Toshkent: “Zamonaviy filologiya” nashriyoti, 45-78 b.
2. Ahmedov, B. (2022). “Katta til modellari va tarjima sifatini baholash muammolari”. O‘zbekiston tarjimashunoslik jurnali, 4-son, 23-41 b.
3. Bahodirova, N., & Karimova, D. (2023). Sun’iy intellekt va tarjima xatolari tipologiyasi: qiyosiy tahlil. “Filologiya masalalari” ilmiy to‘plami, 2-jild, 67-89 b.
4. Brown, T., & Smith, J. (2023). Large Language Models for Translation Post-Editing: A Comprehensive Review. Journal of Artificial Intelligence Research, 78(3), 245-278.
5. Chen, L., Wang, Y., & Zhang, W. (2024). Comparative Analysis of GPT-4 and DeepSeek-V3 for Low-Resource Language Translation. Computational Linguistics, 50(2), 112-145.
6. Elov, B., & Sobirova, M. (2025). “Avtomatik matn tahrirlash vositalari: imkoniyat va cheklovlar”. Amaliy tilshunoslik masalalari, 3-jild, 112-130 b.
7. G‘ulomova, M. (2021). “Tarjima tahririda neyron tarmoqlarni qo‘llash tajribasi”. “Innovatsion texnologiyalar va ta’lim” jurnali, 8-son, 90-108 b.
8. Inoyatova, F. (2025). “Post-redaktirlash kompetensiyasi: zamonaviy tarjimon kasbining yangi talabi”. Tarjima nazariyasi va amaliyoti, 5(1), 34-52.
9. Johnson, P., Lee, H., & Martinez, R. (2023). Prompt Engineering for Translation Error Detection and Correction. arXiv preprint arXiv:2305.09876v3.
10. Karimov, D. (2024). ChatGPT-4 va DeepSeek: qiyosiy imkoniyatlar tahlili. “Raqamli transformatsiya” ilmiy-amaliy konferensiya materiallari, 15-17 mart, Toshkent, 55-74 b.
11. Kim, E., & Vasiliev, A. (2024). Deep Learning Approaches to Translation Quality Assessment. Machine Translation Review, 44(1), 15-33.
12. Wu, Z., Chen, L., & Smith, J. (2024). DeepSeek vs GPT-4: A Practical Benchmark on Post-Editing Across Multiple Language Pairs. Computational Linguistics Review, 44(2), 112-141.
13. Liu, X., & Zhou, Q. (2024). Evaluating ChatGPT and DeepSeek for Professional Translation Tasks. Natural Language Engineering, 30(4), 567-592.
14. Mirzayeva, S. (2023). “Sun’iy intellekt va tarjimon kasbining kelajagi”. Xalqaro ilmiy-amaliy konferensiya materiallari, 12-14 sentyabr, Samarqand, 134-142 b.
15. Rahmonova, S. (2024). “Korpus lingvistikasi va neyron tarjima modellari integratsiyasi”. O‘zbek tilshunosligining dolzarb muammolari, 2(3), 88-106.
16. Tursunov, O. (2023). Neyron tarmoqlar asosida tarjima sifatini baholash: metodologiya va amaliyot. Samarqand: “Ilm ziyo” nashriyoti, 200-235 b.
17. Xolmatov, B. (2022). “Tarjimon va sun’iy intellekt: hamkorlik modeli”. “Zamonaviy tarjimashunoslik: muammo va yechimlar” respublika ilmiy konferensiyasi, 15-17 noyabr, Buxoro, 134-139 b.
18. Yusupova, D. (2025). “Tabiiy tilni qayta ishlash modellarining o‘zbek tiliga moslashuvi muammolari”. “Kompyuter lingvistikasi va sun’iy intellekt” jurnali, 6(2), 45-63.
19. Zhang, W., & Liu, Y. (2024). Prompt Optimization for Translation Error Correction with LLMs. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics, 1123-1140.
20. Zokirov, A. (2024). “Sun’iy intellekt vositalari bilan ishlash: tarjimonlar uchun amaliy qo‘llanma”. Toshkent: “Fan va texnologiya” nashriyoti, 150-178 b.
Copyright (c) 2026 «ACTA NUUz»
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
.jpg)
1.png)
