The paper describes the subject area data that make up the content of the semantic library. The study is aimed at solving the problem of creating a dataset for training large language models to work with Russian-language scientific subject areas. The procedures for filling the library with publications after processing weakly structured data, checking the quality of the knowledge graph and navigating through the data are shown. The application of the work is to create recommender systems for working with scientific Russian-language journals.

applied ontology, knowledge graph, data sources, ontology development, scientific results in the digital environment