Application of artificial intelligence methods in multilingual technologies

Read full text Summarise with VIRAT
Authors: M.V. Karaseva, T.S. Karaseva
Cited by: 0
References: 6

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This article systematizes and comparatively analyzes modern artificial intelligence methods used in multilingual technologies. It examines the evolution of approaches to machine translation, from rule-based and statistical models to neural architectures and transformers. Particular attention is paid to multilingual models such as Multilingual BERT, XLM-R, and NLLB, as well as large language models. Self-attention mechanisms, cross-lingual embeddings, and transfer learning methods that ensure knowledge transfer between languages and efficient translation of resource-poor languages are analyzed. A comparative analysis of the architectures based on quality and scalability metrics is conducted, identifying their advantages and limitations, including computational costs and the problem of language imbalance. It is shown that neural transformer models significantly outperform previous generations of systems in terms of translation quality and versatility. The article concludes that artificial intelligence is of strategic importance for the development of global digital communications and the preservation of linguistic diversity, and identifies promising areas for further research.

Fields of Study & Subjects

Natural Language Processing Techniques Topic Modeling Speech Recognition and Synthesis

Publication details

Journal / source
All-Russian (National) Scientific Conference Achievements of Science and Technology
Pages
330–336
Publication date
1 January 2026
Language
EN