metadata

dataset_info:
  features:
    - name: id
      dtype: string
    - name: tokens
      sequence: string
    - name: pos_tags
      sequence:
        class_label:
          names:
            '0': ADJ
            '1': ADP
            '2': ADV
            '3': AUX
            '4': CCONJ
            '5': DET
            '6': INTJ
            '7': NOUN
            '8': NUM
            '9': PART
            '10': PRON
            '11': PROPN
            '12': PUNCT
            '13': SCONJ
            '14': SYM
            '15': VERB
            '16': X
  splits:
    - name: latin
      num_bytes: 114634
      num_examples: 500
    - name: cyrillic
      num_bytes: 143553
      num_examples: 500
  download_size: 99179
  dataset_size: 258187
configs:
  - config_name: default
    data_files:
      - split: latin
        path: data/latin-*
      - split: cyrillic
        path: data/cyrillic-*
license: apache-2.0
task_categories:
  - token-classification
language:
  - uz
tags:
  - pos
  - uz
  - upos
pretty_name: uzbekpos
size_categories:
  - n<1K

Dataset Card for "uzbekpos"

Dataset Description
Dataset Structure
Dataset Creation
Considerations for Using the Data
Additional Information

Dataset Description

Homepage: Uzbek UD
Repository: UD_Uzbek-UT (conllu format)
Paper: BBPOS: BERT-based Part-of-Speech Tagging for Uzbek
Point of Contact: latofatbobojonova@gmail.com or arofat.akhundjanova@gmail.com
Size of downloaded dataset files: 99.2 kB

Dataset Summary

Uzbek POS: First UPOS tagged dataset for Part-of-Speech tagging task

This dataset is an annotated dataset for POS tagging. It contains 250 sample sentences collected from news outlets and fictional books respectively. The dataset is presented in both Uzbek scripts i.e., Latin and Cyrillic. The annotation was done manually according to UPOS tagset.

Languages

Northern Uzbek (a.k.a Uzbek)

Dataset Structure

Data Instances

Size of downloaded dataset files: 99.2 kB
Size of the generated dataset: 99.2 kB
Total amount of disk used: 99.2 kB

An example of 'latin' looks as follows.

{
  'id': 0,
  'tokens': "['Doimiy', 'g‘ala-g‘ovur', ',', 'to‘lib-toshgan', 'peshtaxtalar', ',', 'mahsulotlarning', 'o‘ziga', 'xos', 'qorishiq', 'isi', '…']",
  'pos_tags': '[0, 7, 12, 15, 7, 12, 7, 10, 0, 0, 7, 12]'
}

Data Fields

The data fields are the same among all splits:

id (string): ID of the example.
tokens (list of string): Tokens of the example text.
pos_tags (list of class labels): POS tags of the tokens, with possible values:
- 0: ADJ
- 1: ADP
- 2: ADV
- 3: AUX
- 4: CCONJ
- 5: DET
- 6: INTJ
- 7: NOUN
- 8: NUM
- 9: PART
- 10: PRON
- 11: PROPN
- 12: PUNCT
- 13: SCONJ
- 14: SYM
- 15: VERB
- 16: X

Data Splits

Dataset consists of two splits according to its written script.

name
latin	500
cyrillic	500

Dataset Creation

Source Data

news articles:
- Kun.uz
- Daryo.uz
fictional books:
- “Og‘riq Tishlar” and “Dahshat” by Abdulla Qahhor
- “Shum Bola” and “Yodgor” by G‘afur G‘ulom
- “Sofiya”, “Hazrati Hizr Izidan”, “Bibi Salima va Boqiy Darbadar”, “Olisdagi Urushning Aks-Sadosi” and “Genetik” by Isajon Sulton
- “Buxoro, Buxoro, Buxoro. . . ”, “Ozodlik” and “Lobarim Mening. . . ” by Javlon Jovliyev
- “Ko‘k Tog‘”, “Insonga Qulluq Qiladurmen”, “Fano va Baqo” and “Chodirxayol” by Asqar Muxtor
- “Ajinasi Bor Yo‘llar” by Anvar Obidjon
- “Kecha va Kunduz” and “Qor Qo‘ynida Lola” by Cho‘lpon.

Initial Data Collection and Normalization

All sentences were handpicked to ensure the quality of the data.

Annotations

Annotation process

Manual

Who are the annotators?

Arofat Akhundjanova (M.Sc. Language Science and Technology, Saarland University)

Citation Information

@inproceedings{bobojonova-etal-2025-bbpos,
    title = "{BBPOS}: {BERT}-based Part-of-Speech Tagging for {U}zbek",
    author = "Bobojonova, Latofat  and
      Akhundjanova, Arofat  and
      Ostheimer, Phil Sidney  and
      Fellenz, Sophie",
    editor = "Hettiarachchi, Hansi  and
      Ranasinghe, Tharindu  and
      Rayson, Paul  and
      Mitkov, Ruslan  and
      Gaber, Mohamed  and
      Premasiri, Damith  and
      Tan, Fiona Anting  and
      Uyangodage, Lasitha",
    booktitle = "Proceedings of the First Workshop on Language Models for Low-Resource Languages",
    month = jan,
    year = "2025",
    address = "Abu Dhabi, United Arab Emirates",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.loreslm-1.23/",
    pages = "287--293",
    abstract = "This paper advances NLP research for the low-resource Uzbek language by evaluating two previously untested monolingual Uzbek BERT models on the part-of-speech (POS) tagging task and introducing the first publicly available UPOS-tagged benchmark dataset for Uzbek. Our fine-tuned models achieve 91{\%} average accuracy, outperforming the baseline multi-lingual BERT as well as the rule-based tagger. Notably, these models capture intermediate POS changes through affixes and demonstrate context sensitivity, unlike existing rule-based taggers."
}

Datasets:

latofat
/

uzbekpos