Document-Level Relation Extraction in Russian

We explore automatic relation extraction at the document level for Russian texts, considering nested entities. Using state-of-the-art models, we conducted first-of-its-kind experiments on Russian texts and made necessary model adjustments to implement experiments on the NEREL dataset, which is annotated with nested named entities and document-level relations between them. The results validate the efficiency of the proposed techniques in relation extraction from Russian texts, including nested entities.