Using artificial intelligence to develop a machine translation system and teaching resources in the Tuvan language

Abstract

The advancement of computer technologies applied in the humanities and the progress in the development of large language models based on machine learning and neural network technologies have reached an exceptionally high level of sophistication. The linguistic potential of large language models elicits a natural interest among researchers, which constitutes a justified reflection of the relevance and importance of using artificial intelligence to create machine translation systems and educational resources. The article explores the experience of creating a large language model for the Tuvan language using machine learning and artificial intelligence. The authors undertook an attempt to develop a large language model capable of recognizing the Tuvan language, translating phrases into Russian and back. In addition, the possibilities of generating text in Tuvan were examined and tested, which can be used both in the field of language teaching and when conducting various kinds of linguistic research. This experience is unique since, as of now, the Tuvan language is not represented in any well-established machine translation systems. A secondary aim of the research is to analyze the level of the language’s digital presence on the Internet, as well as to provide recommendations for devising an optimal algorithm for building similar systems and web services based on machine learning. The research outcomes are of practical value not only with respect to the Tuvan language but can also be extrapolated to other official languages in the Russian Federation.