MaksudSharipov/UzbekPosTagger_BERTbek

en Uzbek POS-tagger (Fine-tuned BERTbek Model)

🇺🇿 Uzbek POS-tagger (BERTbek Model asosida qurilgan)

This repository contains a Part-of-Speech (POS) Tagging model for the Uzbek language, fine-tuned from the BERTbek model by Elmurod Kuriyozov.

en Uzbek POS-tagger (Fine-tuned BERTbek Model)

🇺🇿 Uzbek POS-tagger (BERTbek Model asosida qurilgan)

This repository contains a Part-of-Speech (POS) Tagging model for the Uzbek language, fine-tuned from the BERTbek model by Elmurod Kuriyozov.

---

🧠 Model Overview

• Model name: Uzbek POS-tagger based on BERTbek
• Base model: BERTbek (news-big-cased)
• Architecture: BERT (Transformer-based encoder)
• Fine-tuned for: POS tagging task (token classification)
• Training platform: Google Colab (NVIDIA A100 GPU)
• License: CC BY-NC 4.0

---

📘 Dataset

• Source: Manually annotated Uzbek POS-tagged dataset
• Size: 4,000 sentences (50,000 tokens)
• Tags: 16 POS tags based on the Universal Dependencies (UD) tagset
• Annotation: Conducted manually by linguists for high-quality labeling

---

📊 Model Performance

Metric	Score
Accuracy	~91%
F1-score	~87%

---

🧩 Applications

• Linguistic analysis of Uzbek texts
• Corpus annotation
• Preprocessing pipeline for downstream NLP tasks (NER, parsing, etc.)

---

🧑‍💻 Author & Credits

• Model fine-tuning and dataset: Maksud Sharipov
• Base model author: Elmurod Kuriyozov
• Framework: Hugging Face Transformers

---

📜 License

This model is licensed under the Creative Commons Attribution–NonCommercial 4.0 International (CC BY-NC 4.0) license.
You are free to use and adapt the model for non-commercial research with appropriate credit.

---

🗣 Citation

If you use this model in your research, please cite as:

@misc{sharipov2025uzbekpos,
  title  = {Uzbek POS-tagger (Fine-tuned BERTbek Model)},
  author = {Maksud Sharipov},
  year   = {2025},
  howpublished = {Hugging Face},
  url    = {https://huggingface.co/MaksudSharipov/UzbekPosTagger_BERTbek}
}