实体关系抽取作为信息抽取、自然语言理解、信息检索等领域的核心任务和重要环节，能够从文本中抽取实体对间的语义关系.近年来，深度学习在联合学习、远程监督等方面上的应用，使关系抽取任务取得了较为丰富的研究成果.目前，基于深度学习的实体关系抽取技术，在特征提取的深度和模型的精确度上已经逐渐超过了传统基于特征和核函数的方法.围绕有监督和远程监督两个领域，系统总结了近几年来中外学者基于深度学习的实体关系抽取研究进展，并对未来可能的研究方向进行了探讨和展望.

Entity relation extraction is a core task and an important part in the fields of information extraction, natural language understanding, and information retrieval. It can extract the semantic relationships between entity pairs from the texts. In recent years, the application of deep learning in the fields of joint learning, remote supervision has resulted in relatively abundant research results in relation extraction tasks. At present, entity relationship extraction technology based on deep learning has gradually exceeded the traditional methods which are based on features and kernel functions in terms of the depth of feature extraction and the accuracy. This paper focuses on the two fields of supervision and remote supervision. It systematically summarizes the research progress of Chinese and overseas scholars' deep relationship-based entity relationship extraction in recent years, and discusses and prospects future possible research directions as well.

国家重点研发计划（2018YFB1403501）