Uzbek Book Corpus With Latin And Cyrillic Scripts

Nearly 40,000 books in the Uzbek language form this text corpus, divided into 'original' and 'lat' branches for OCR and Latin versions. Author murodbek released the dataset to support research on low-resource languages. It was last updated on Hugging Face in June 2025.

Use Cases

• Train masked language models on the 'original' OCRed text branch to handle script variations.
• Develop text generation models using the 'lat' branch's fully Latin version of the corpus.
• Analyze linguistic patterns across nearly 40,000 books for low-resource language research.
• Benchmark OCR post-processing techniques by comparing the 'original' and 'lat' text branches.

Strengths

• Contains nearly 40,000 books, providing a substantial text corpus for a low-resource language.
• Offers two distinct text branches ('original' and 'lat') for comparative analysis.

Limitations

• Specific row counts, column names, and sample data are unavailable for detailed assessment.
• The dataset's license and precise collection methodology are not specified in the provided description.

Provenance

Source

Hugging Face dataset authored by murodbek.

Collection Method

Appears to be compiled from books, with OCR processing for the 'original' branch and conversion to Latin script for the 'lat' branch.

Freshness

Last updated on the platform in June 2025.

Geography

Primary language is Uzbek, suggesting content origin from Uzbekistan or Uzbek-speaking regions.