Knowledge Graph Extraction from Russian News

In the modern world there are multiple instruments for data analysis and representation, and they keep evolving. Graphs are considered to be a part of these instruments, because they are used for knowledge representation. In their structure they have nodes, which represent some entities, and edges, which store information about those entities. Since knowledge can be collected not only from structured forms of data, but also from text, there is a strong need for special NLP tools and methods. The current study covers this issue, as it is aimed to build knowledge graphs on the basis of Russian news texts. There is a lack of studies that cover extracting knowledge graphs from texts that are written in Russian language. In the current study a model of syntactic relationship was constructed, and it is used for extracting triplets of two entities and the relationship between them from texts. These triplets are then used for knowledge graph extraction. The knowledge graphs were extracted from Russian news of the online news agency Lenta.ru. The constructed graphs can be used to visualize any kinds of information, relationships between specific entities, and for representing dynamically evolving knowledge.