The Bashkir Wikipedia Corpus is a collection of 43,926 articles from Bashkir Wikipedia and Wikibooks, totaling approximately 10.6 million tokens and 8.9 million words. The data has been extracted from official Wikimedia dumps and processed to provide clean, well‑structured text suitable for NLP tasks. The corpus includes article titles, full content, categories, source identifiers, and a combined text field.

Curated by: BashkirNLPWorld
Language(s) (NLP): Bashkir (ba), Cyrillic script
License: cc-by-sa-4.0 – see Licensing & Legal Notice below.

Licensing & Legal Notice

This dataset is derived from Bashkir Wikipedia (ba.wikipedia.org) and Bashkir Wikibooks (ba.wikibooks.org), which are licensed under the Creative Commons Attribution‑ShareAlike 4.0 International License (CC BY‑SA 4.0). As a derivative compilation, this dataset is also released under the same license.

Attribution must be given to the original Wikipedia/Wikibooks contributors and to this dataset (see Citation below).
ShareAlike – any derivative works must be distributed under the same license.
Users are free to share and adapt the material for any purpose, including commercially, provided they comply with the license terms.
No additional legal restrictions are imposed by the dataset curator beyond those required by the CC BY‑SA 4.0 license.

Dataset Sources

Repository: https://huggingface.co/datasets/BashkirNLPWorld/bashkir-wiki-corpus
Original sources:
- Bashkir Wikipedia: https://ba.wikipedia.org
- Bashkir Wikibooks: https://ba.wikibooks.org

Uses

Direct Use

This dataset is intended for:

Training language models (BERT, GPT) for Bashkir
Token classification (named entity recognition, part‑of‑speech tagging)
Fill‑mask and other pre‑training tasks
Feature extraction and embedding creation
Text classification (category prediction, topic modeling)
Linguistic analysis of Bashkir

Out-of-Scope Use

Redistribution without complying with CC BY‑SA (e.g., without attribution or under incompatible license).
Any use that violates the terms of the original Wikimedia projects.
Automated decision-making in sensitive domains without human oversight.

Dataset Structure

Data Fields

Field	Type	Description
`title`	string	Article title (available for 99.6% of documents)
`content`	string	Main article content
`category`	string	Wikipedia categories (available for 96.6% of documents)
`source`	string	Source identifier (`bawiki_cyrillic` or `bawikibooks_cyrillic`)
`text`	string	Combined field: `title + "\n\n" + content`

Data Splits

Split	Size	Description
`full`	43,926	Complete corpus
`sample`	1,000	Random subset for testing and development

Source Distribution

Source	Documents	Percentage
bawiki_cyrillic	42,262	96.2%
bawikibooks_cyrillic	1,664	3.8%

Dataset Creation

Curation Rationale

Bashkir is a low‑resource Turkic language with limited digital corpora. Wikipedia and Wikibooks provide a large, open, and reliable source of text. This dataset was created to make that content easily accessible in a cleaned, structured format for NLP research and applications.

Source Data

Data Collection and Processing

Data was obtained from official Bashkir Wikipedia and Bashkir Wikibooks dumps. Processing steps included:

Extraction of articles from the Wikimedia XML dump.
Cleaning of HTML tags, templates, and formatting artifacts.
Normalization of Unicode and whitespace.
Category extraction from Wikipedia's category system.
Deduplication of duplicate articles.
Quality filtering (removal of stubs and very short articles).
Creation of the combined text field.

Who are the source data producers?

The original content was created by the Bashkir Wikipedia and Wikibooks community — thousands of volunteer contributors. The dataset curator only performed cleaning and structuring; the intellectual content remains the work of the original authors, licensed under CC BY‑SA 4.0.

Annotations

No additional manual annotations were added. The category field was extracted directly from Wikipedia's category system.

Personal and Sensitive Information

Wikipedia articles are publicly available and may contain information about living persons. The dataset does not include any private information beyond what is already public on Wikipedia. Users should be aware of Wikipedia's policies regarding biographies of living people.

Bias, Risks, and Limitations

Content bias: Like many small Wikipedias, content may be skewed towards certain topics (e.g., geography, sports).
Geographic bias: Stronger coverage of Bashkortostan and Russia; global coverage may be limited.
Temporal coverage: Articles reflect the state of Wikipedia at the time of extraction; no revision history is included.
Missing dates: No publication or last‑updated dates are available.
Copyright: The content is CC BY‑SA; users must comply with license terms (see Licensing & Legal Notice).

Recommendations

Use the category field for topic filtering and classification.
For balanced language modeling, consider combining with other Bashkir corpora.
For historical analysis, use the full Wikipedia revision history (not included here).
Always attribute the original contributors and this dataset when redistributing.

Citation

BibTeX (dataset):

@dataset{bashkirnlpworld2026wikicorpus,
  title = {Bashkir Wikipedia Corpus},
  author = {BashkirNLPWorld},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/BashkirNLPWorld/bashkir-wiki-corpus}
}

APA (dataset): BashkirNLPWorld. (2026). Bashkir Wikipedia Corpus [Data set]. Hugging Face. https://huggingface.co/datasets/BashkirNLPWorld/bashkir-wiki-corpus

Glossary

Wikipedia dump – a snapshot of all Wikipedia articles at a specific point in time.
Category – Wikipedia's classification system for organizing articles.
Cyrillic script – the writing system used for Bashkir.

More Information

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

Dataset Card Authors

BashkirNLPWorld (community)

Dataset Card Contact

For questions, takedown requests, or collaboration, please open an issue on the Hugging Face repository:
https://huggingface.co/datasets/BashkirNLPWorld/bashkir-wiki-corpus/discussions
📧 Direct email: bashkirnlpworld@gmail.com

Downloads last month: 22

Total file size:

102 MB