基于深度学习的法院命名实体识别模型

Research on the Recognition Model of Court Judgment Named Entity Based on Deep Learning

命名实体识别作为信息抽取、问答系统、句法分析、机器翻译等应用领域的重要基础工具,在法院判决书信息抽取系统中也得到了广泛应用。然而,已有的技术模型在文本中存在大量专有名词或术语时,实体识别的提取效果会变得很差。双向循环神经网络-条件随机场判别模型可对现有的法院判决书条件随机场模型进行优化,实现自动化特征的选取过程,准确率比已有的条件随机场模型更高。

Named entity recognition,as an important basic tool in such application fields as information extraction,question and answer system,syntactic analysis,machine translation and others,has been widely used in court judgment information extraction system.However,the extraction effect of entity recognition becomes poor when existing technical models have a large number of proper nouns or terminologies in the text.In this paper,a two-way cyclic neural network-conditional random-airport discriminant model is thus developed to optimize the existing court judgment condition with the airport model.The experiment proves that the model can realize the selection process of automatic feature,and the accuracy rate is higher than that in the airport model.