Knowledge graph-based document-level relation extraction:recent advances and perspectives
Document-level Relation extraction(DRE)technology aims to captures the complex semantic relations expressed by multiple entity references in the documents by integrating of both intra-sentence and inter-sentence information.Accurately ex-tracting all the relations between entity pairs from massive unstructured data is not only the key to the automatic construction of knowledge graphs from the bottom to top,but also the basis for the smooth implementation of downstream tasks.In this paper,we sum-marize the research progress of DRE techniques using graph-based methods and non-graph-based methods.We also briefly introduce its potential applications in knowledge graphs,and explore the key directions of DRE's future research.The existing methods have achieved remarkable results in DRE,highlighting its importance in knowledge extraction and automatic knowledge graph construc-tion.We clarify the current development status and forecast future trends in DRE will provide direction for the follow-up studies.
万方数据
