解决嵌套问题的中文命名实体识别  

Chinese Named Entity Recognition for Solving Nesting Problem

作　　者：曲晓东 李佳昊 QU Xiaodong;LI Jiahao(Zhongyuan University of Technology,Zhengzhou 450007,China)

机构地区：[1]中原工学院,郑州450007

出　　处：《移动信息》2023年第6期234-236,共3页MOBILE INFORMATION

摘　　要：作为众多任务的子任务,命名实体识别的发展较为迅速,但在中文命名实体识别领域,还存在不少问题,嵌套实体就是其中一个难点。文中根据结点的传入和传出,使用了图卷积神经网络提取图特征,改善了嵌套实体的准确度,并通过使用图神经网络处理中文命名实体识别的问题,更好地融合了词典信息。另外,文中分别对两类数据集进行了实验验证,结果显示,相比其他模型,该模型有所提高。As a subtask of many tasks,named entity recognition has developed rapidly,but there are still many problems in the field of Chinese named entity recognition,and nested entities are one of the difficulties.This paper uses graph convolutional neural networks to extract graph features according to the incoming and outgoing nodes,improves the accuracy of nested entities,and by using graph neural networks to deal with the problem of Chinese named entity recognition,so as to better integrate dictionary information.In addition,experimental verification of two types of datasets is carried out in this paper,and the results show that the model has improved compared with other models.

关 键 词：嵌套问题 中文命名 实体识别 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]