The Tilqazyna National Scientific and Practical Center under the Language Policy Committee of the Ministry of Science and Higher Education presented the first results of the Tilqazyna Kazakh Language Teaching Model based on artificial intelligence.
At the moment, the model is able to perform tasks in such areas of the Kazakh language as vocabulary, morphology, semantics, etc. In particular, it can generate text in Kazakh, create periphrases, work with context, shorten texts, correct grammatical and punctuation errors, reveal the meanings of phraseological units, and translate terms.
This industry-specific LLM model has already been uploaded to the Hugging Face platform and is available to all users. Using this model will make it possible to develop many IT products in the Kazakh language using artificial intelligence. This corresponds to the Message of President Kassym-Jomart Tokayev, which emphasized the importance of Kazakhstan becoming a country that makes extensive use of artificial intelligence and develops digital technologies.
When developing the model, experts from the Til-Kazyna center applied natural language processing algorithms and purposefully analyzed large amounts of data. In particular:
684,876 lexical units were used to check the words.;
To improve the system of checking phrases, 20 212 correct and erroneous variants were used.;
5,558 texts were analyzed to correct punctuation errors.;
3,000 correct and incorrect versions of the texts were prepared to correct the structure of the text.;
A database of 6,000 complete and abbreviated sentences has been created for the shortening function.;
14,790 synonymous series have been collected for the periphrase function;
The total volume of the processed Kazakh language corpus was 35 GB.
This year, the voice communication feature will be added to the model, and a user-friendly interface will be developed. The project will also be able to teach Kazakh at A1, A2 and B1 levels, and by 2026 at B2 and C1 levels.
The end result of the project will be a voice assistant capable of creating an individual Kazakh language training program depending on the user's language proficiency. It will be presented as a mobile application for iOS and Android systems.

Source : https://www.gov.kz/memleket/entities/sci/press/news/details/920165