Method for named entities linking with concepts of the Wikidata knowledge base

The paper presents a method for automatically linking named entities in Russian-language texts with concepts from the Wikidata knowledge base. It is based on the use of named entity search tools with subsequent semantic analysis of the correspondence of the found entity to the concept in the knowledge base. The resulting links can later be used to form a linked corpus of texts in any subject area. The difference between the presented method and existing ones is the analysis of both the named entity itself, its attributes, as well as associated words without the use of machine learning methods. This approach allows to increase the accuracy of searching for a corresponding concept in the knowledge base and eliminates the need to constantly retrain the neural network model to recognize new entities added to the knowledge base.