JBT Journal
Advanced

Raqamli Transformatsiya va Sun’iy Intellekt

AKADEMIK MAQOLALARDAN METAMA’LUMOTLARNI EKSTRAKSIYA QILISHDA MAVJUD YECHIMLAR

Rashid Turg‘unboyev1

1 Qo‘qon davlat universiteti

Keywords

BERT katta til modellari multimodal yondashuv akademik maqolalar GPT-4 metama’lumotlarni ekstraksiya qilish

Abstract

Ushbu maqolada katta til modellarining akademik maqolalardan metama’lumotlarni ekstraksiya qilishdagi imkoniyatlari tahlil qilinadi. GPT-4, BERT va LLaMA kabi modellar an’anaviy usullarga nisbatan yuqori moslashuvchanlik, murakkab ma’lumotlarni ajratish va ko‘p tilli ishlash afzalliklariga ega. ChatExtract usuli 90% F1-natijaga erishgan. Biroq resurs talabchanligi, faktik xatolar va maxfiylik muammolari asosiy cheklovlardir. Tor sohaga moslashtirilgan modellar tejamkorligi bilan ajralib turadi.

Views: 98 Published: 2026-08-26 Pages: 100-105 Volume: 4 Issue: 4

References

  1. Alyafeai, Z., Al-Shaibani, M. S., & Ghanem, B. (2025). MOLE: Metadata extraction and validation in scientific papers using LLMs. arXiv preprint arXiv:2505.19800.
  2. Usta, G. (2026). Automating MARC Records from Turkish Title Pages: A Performance and Efficiency Comparison of Fine-Tuned BERT vs. VLMs. Journal of Library Metadata, 1–19. https://doi.org/10.1080/19386389.2026.2627732
  3. Kohút, J., Dočekal, M., Hradiš, M., Vaško, M. (2026). BiblioPage: A Dataset of Scanned Title Pages for Bibliographic Metadata Extraction. In: Yin, XC., Karatzas, D., Lopresti, D. (eds) Document Analysis and Recognition – ICDAR 2025. ICDAR 2025. Lecture Notes in Computer Science, vol 16025. Springer, Cham. https://doi.org/10.1007/978-3-032-04624-6_17
  4. J. Dagdelen et al., “Structured information extraction from scientific text with large language models,” Nature Communications, vol. 15, no. 1, 2024, doi: 10.1038/s41467–024-45563-x.
  5. Johan Krause, Igor Shapiro, Tarek Saier, and Michael Färber. 2021. Bootstrapping Multilingual Metadata Extraction: A Showcase in Cyrillic. In Proceedings of the Second Workshop on Scholarly Document Processing, pages 66–72, Online. Association for Computational Linguistics.
  6. T. Saier, M. Ohta, T. Asakura, and M. Färber, “HyperPIE: Hyperparameter Information Extraction from Scientific Publications,” Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14609 LNCS, pp. 254–269, 2024, doi: 10.1007/978-3-031-560606_17.
  7. Polak, M.P., Morgan, D.: Extracting accurate materials data from research papers with conversational language models and prompt engineering - Example of ChatGPT (Mar 2023). https://doi.org/10.48550/arXiv.2303.05352
  8. Boukhers, Z., & Yang, C. (2025). Comparison of Feature Learning Methods for Metadata Extraction from PDF Scholarly Documents. arXiv preprint arXiv:2501.05082.
  9. Choi, W., Yoon, H. M., Hyun, M. H., Lee, H. J., Seol, J. W., Lee, K. D., ... & Kong, H. (2023). Building an annotated corpus for automatic metadata extraction from multilingual journal article references. PloS one, 18(1), e0280637.