Abstract <p>The use of neural network language models to increase the efficiency of processing scientific and technical information (STI) on the Internet from the point of view of cognitive information retrieval is considered. A conceptual model is proposed to identify subject areas that corresponding to the object of information need based on existing STI classification systems. The model is based on the principles of coordinate indexing incorporating methods of detecting and tracking topics and the use of large language models to analyze the information space of the user’s information needs, metasearch of Internet resources, automatic abstracting and ranking of relevant documents.</p>