
- 9
- 20.11.2026 12:33:00
- 1 min read
According to the Ministry of Digital Technologies, a large Uzbek-language text corpus containing 5.6 million documents and nearly 6 billion words has been created and made publicly available on the Hugging Face platform.
The project was implemented following instructions given by Shavkat Mirziyoyev during a meeting on October 21, 2025, aimed at ensuring that artificial intelligence systems can fully understand the Uzbek language.
The corpus includes texts in both Latin and Cyrillic Uzbek scripts. It contains books, newspapers, archival materials, legislative documents, Wikipedia articles, web pages, and public Telegram posts.
The collected data has undergone special processing, quality filtering, and classification by topic. Personal information has been replaced with special markers.
The main objective of the project is to improve AI models' ability to understand and process Uzbek, as well as generate high-quality text in the language.
The ministry is currently working to provide this database to leading artificial intelligence developers. In particular, cooperation with the Qwen model has already begun.
The database is expected to contribute to the further development of artificial intelligence systems capable of operating in Uzbek.