Abstract
<jats:p>The last decade has seen significant developments in natural language processing methods: automatic translation has improved significantly, the quality of text classification, sentiment and argumentation analysis, and other tasks has increased. Radical improvements have been achieved in the task of text generation, where generated texts are almost indistinguishable from those written by humans, and voice assistants (chatbots) are able to conduct a full-fledged dialogue in natural language. Due to this, automatic natural language processing is becoming increasingly important for everyday life: most people constantly use the listed technologies, and commercial companies are increasing their staff of natural language processing specialists and conducting their own scientific research. The textbook considers several stages in the development of automatic text processing methods: both methods based on dictionaries and rules, and machine learning methods that occupy a central place in modern science and are capable of self-learning based on human-labeled data. Neural network approaches based on the transformer architecture are described in detail, including large language models and methods for training them. The textbook is intended for senior undergraduate, graduate and postgraduate students majoring in «Applied Mathematics and Computer Science» and “Fundamental and Applied Linguistics”, as well as researchers, developers and scientists interested in methods of natural language processing and machine learning. It can be used for retraining and advanced training of specialists in applied mathematics, machine learning and artificial intelligence. The textbook will also be useful for linguists and philologists who want to become more deeply familiar with methods of automatic natural language processing.</jats:p>