应用深度学习的中文命名实体识别研究综述

Review of Chinese Named Entity Recognition Based on Deep Learning

命名实体识别是自然语言处理领域的基础性工作,旨在从非结构化文本中识别出具有特定意义的实体并分类,在多种自然语言处理任务中发挥重要作用。由于中文命名实体没有明显的边界标记,且存在歧义和嵌套等问题,其识别过程比英语等其他语言要更为复杂。近年来,深度学习技术发展迅速,在中文命名实体识别中得到广泛应用,并已成为主流方法。系统梳理中文命名实体识别中深度学习技术的研究进展,重点从文本表示、特征编码、预测解码3个方面,对比分析代表性工作的关联性和关键技术,讨论研究中存在的问题、现有解决方案和未来的研究方向。

Named Entity Recognition(NER)is a fundamental task in Natural Language Processing(NLP)that aims to identify and clarify entities with specific meanings from unstructured text.It is an indispensable part of various downstream NLP fields.Chinese NER is more difficult than English and other languages because there are no obvious boundary markers,and the entities have problems such as ambiguity and nesting,which pose great challenges for existing methods.In recent years,deep learning technology has developed rapidly and has been widely applied in Chinese NER.We give a comprehensive study of recent advances in deep learning research of Chinese NER from perspectives of text representation,context encoding,and tag decoding,focusing on key techniques and relationship among these works.Furthermore,we summarize the main challenges and the latest advances of Chinese NER,and give a discussion on possible future work.