Document-level entity relation extraction involves identifying and categorizing relationships between entities mentioned within whole documents, rather than individual sentences.