The Tatar News Clustered Dataset is a comprehensive collection of 57,340 Tatar language news articles with topic categories, curated by TatarNLPWorld as part of the Tat2Vec project. The dataset includes full article content, titles, source names, publication dates, and 282 topic categories. It is designed for multi-class text classification, topic modeling, text clustering, and various NLP tasks for the Tatar language, a low-resource Turkic language.

Curated by: TatarNLPWorld Community
Language(s) (NLP): Tatar (tt)
License: other – see Licensing & Legal Notice below.

Licensing & Legal Notice

This dataset follows the practice established by large web-crawled corpora such as HPLT and OSCAR:

Original source texts (news articles, titles, etc.) remain the property of their respective authors and publishers. They are not owned by the TatarNLPWorld team and are not covered by the MIT license or any other open license applied to the annotations.
The structured compilation, metadata, and any original annotations created by the TatarNLPWorld community are released under the MIT License.
Users are solely responsible for ensuring their use of the underlying texts complies with applicable copyright law. For commercial use of verbatim excerpts, permission from the original copyright holders may be required.
A notice-and-takedown procedure is in place: rights holders can request removal of specific articles by contacting the dataset maintainers (see Dataset Card Contact). We commit to responding within 14 business days and removing disputed content in the next release.

Dataset Sources

Repository: https://huggingface.co/datasets/TatarNLPWorld/tatar-news-cluster
Paper: Arabov, M. K., Gilmullin, R. A., & Burnashev, R. A. (2026). Combining Classical and Transformer-based Approaches for Text Classification and Topic Modeling of the Tatar Language. In 2026 IEEE Ural-Siberian Conference on Biomedical Engineering, Radioelectronics and Information Technology (USBEREIT) (pp. 1–4). IEEE. https://doi.org/10.1109/USBEREIT70063.2026.11580632
Demo: [Coming Soon]

Uses

Direct Use

This dataset is intended for:

Multi-class text classification with 282 topic categories
Topic modeling and text clustering
Feature extraction for creating embeddings
Sentence similarity tasks
Text generation and language modeling
Summarization of Tatar news articles
Zero-shot classification and cross-lingual experiments
Source attribution and analysis of Tatar media

Out-of-Scope Use

This dataset should not be used for:

Redistribution of verbatim news articles without permission from original copyright holders
Any use that could misrepresent the original authors' intent (e.g., altering labels or categories without validation)
Applications requiring real-time or up-to-date news content – the dataset is a static snapshot
Automated decision-making in sensitive domains without human oversight

Dataset Structure

Data Fields

Each record contains:

Field	Type	Description
`content`	string	Full article content in Tatar language
`title`	string	Article title
`category`	string	Topic category (282 unique categories)
`source`	string	Original source name
`content_length`	int64	Length of content in characters
`resource`	string	Source URL or identifier
`date`	string	Publication date (ISO format)

Data Splits

The dataset currently provides a single split:

Split	Size	Description
`train`	51,606	Training set (90% of the full data)

Note: The original dataset was split into train (90%) and validation (10%), but the validation split is not included in this release. For evaluation purposes, you can split the train set yourself.

Dataset Creation

Curation Rationale

The Tatar language lacks large-scale labeled datasets for topic classification. This dataset was created to fill that gap by providing a diverse collection of Tatar news articles with rich metadata, enabling research in text classification, topic modeling, and other NLP tasks for this low-resource language.

Source Data

Data Collection and Processing

Texts were collected from multiple Tatar language news sources and portals, including merged Matbugat news, Beznen articles, Azatliq.org, Syuyumbike news, Tatar-inform, and others. The processing pipeline involved:

Web crawling of publicly available Tatar news websites
Extraction of article content, titles, and metadata (source, date)
Assignment of category labels from source metadata (RSS feed categories, site structure)
Cleaning and normalization (removal of HTML tags, extra whitespace)
Calculation of content_length and deduplication

Who are the source data producers?

The original texts were produced by journalists, editors, and media organizations from various Tatar-language outlets. The TatarNLPWorld community performed the collection, structuring, and categorization, but the intellectual content belongs to the original authors and publishers.

Annotations

The dataset does not contain manual annotations in the traditional sense; the category labels were extracted from the structure of the source websites (e.g., RSS categories, URL paths). No additional human labeling was applied.

Personal and Sensitive Information

The dataset consists solely of publicly available news articles. It does not intentionally include personal or sensitive data beyond what is present in the news texts themselves. No anonymization was performed because the data is already public.

Bias, Risks, and Limitations

Domain bias: The dataset reflects the content distribution of the crawled sources; certain categories (e.g., Society, Culture) may be overrepresented.
Source bias: The majority of data comes from a limited number of sources (e.g., merged_matbugat_news accounts for 48%).
Temporal coverage: Articles span 2024–2026, with possible uneven distribution across time.
Label noise: Category labels are derived from source metadata and may contain inconsistencies or misclassifications.
Copyright constraints: The underlying texts are protected; users must respect original rights (see Licensing & Legal Notice).

Recommendations

When training models, consider balancing categories or using weighted loss functions to mitigate class imbalance.
Use the source field to analyze and potentially correct for source bias.
For any commercial application, verify the copyright status of specific articles.
Cite the original sources when using excerpts.

Citation

BibTeX (dataset):

@dataset{tatar_news_clustered_2026,
    title = {Tatar News Clustered Dataset},
    author = {TatarNLPWorld Community},
    year = {2026},
    publisher = {Hugging Face},
    url = {https://huggingface.co/datasets/TatarNLPWorld/tatar-news-cluster}
}

APA (dataset): TatarNLPWorld Community. (2026). Tatar News Clustered Dataset [Data set]. Hugging Face. https://huggingface.co/datasets/TatarNLPWorld/tatar-news-cluster

Related Publication

The following paper used this dataset or a similar version for text classification and topic modeling:

BibTeX:

@inproceedings{arabov2026combining,
  author = {Arabov, M. K. and Gilmullin, R. A. and Burnashev, R. A.},
  title = {Combining Classical and Transformer-based Approaches for Text Classification and Topic Modeling of the Tatar Language},
  booktitle = {2026 IEEE Ural-Siberian Conference on Biomedical Engineering, Radioelectronics and Information Technology (USBEREIT)},
  year = {2026},
  pages = {1--4},
  doi = {10.1109/USBEREIT70063.2026.11580632}
}

APA: Arabov, M. K., Gilmullin, R. A., & Burnashev, R. A. (2026). Combining Classical and Transformer-based Approaches for Text Classification and Topic Modeling of the Tatar Language. In 2026 IEEE Ural-Siberian Conference on Biomedical Engineering, Radioelectronics and Information Technology (USBEREIT) (pp. 1–4). IEEE. https://doi.org/10.1109/USBEREIT70063.2026.11580632

Glossary

Multi-class classification – a task where each input is assigned one of more than two categories.
Topic modeling – discovering abstract topics in a collection of documents.
Low-resource language – a language with limited digital corpora and NLP tools.

More Information

For questions, contributions, or feedback, please open an issue on the Hugging Face repository or contact the TatarNLPWorld community.

Dataset Card Authors

TatarNLPWorld Community

Dataset Card Contact

Hugging Face: https://huggingface.co/TatarNLPWorld
GitHub: https://github.com/TatarNLPWorld
Email: marabov@kpfu.ru

Downloads last month: 19

Total file size:

111 MB