Joint entity-relation extraction: A key technique for knowledge graph construction—A survey

<p indent="0mm">Knowledge graph (KG), which aims to represent unstructured semantic knowledge in the form of structured triplet, is crucial to establishing the knowledge base for artificial intelligence (AI) research. There are many techniques proposed for knowledge graph construction, among which joint entity-relation extraction is a key technique. Since the proposal of joint entity-relation extraction, numerous methods have emerged, profoundly impacting the automatic construction of KG. However, there remains a notable lack of systematic survey in this field. In this paper, we provide a thorough review of joint entity-relation extraction. We point out the ambiguous task definition and clarify the task scope of joint entity-relation extraction thereby. We further detail the different methods from the perspective of formulations, rather than the commonly used encoder evolution perspective in related survey work. We also introduce the datasets, analyze the performance of classical work and offer our insight for the future development.