基于深度学习的医疗命名实体识别  

Medical Text Entities Recognition Method Base on Deep Learning

作　　者：张帆[1] 王敏[1] ZHANG Fan WANG Min(Hunan University, College of Electric and Information Engineering, Changsha, Hunan 410082)

机构地区：[1]湖南大学电气与信息工程学院,湖南长沙410082

出　　处：《计算技术与自动化》2017年第1期123-127,共5页Computing Technology and Automation

基　　金：湖南省科技厅重点研发计划(2015WK3049)

摘　　要：在较为深入地研究医疗文本实体识别的现有方法的基础上,设计一种基于深度学习的医疗文本实体识别方法。本文在医疗文本数据集上进行实体识别对比实验,所识别目标实体包含疾病,症状,药品,治疗方法和检查五大类。实验结果表明,设计的深度神经网络模型能够很好的应用到医疗文本实体识别,本文所设计的方法比传统算法(如CRF)具有较少人工特征干预及更高的准确率和召回率等优点。This article do deep research on many medical text entity recognition methods, and design a medical text entity recognition method base on deep learning. This article do entity recognition controlled experiments on medical text data set, the target entity contain diseases, symptoms, medicine, treatment and inspection. The experimental result show that our neural network model can be applied to medical text entity recognition. Our neural network model have higher precision, higher recall and less artificial features than traditional methods such as CRF.

关 键 词：实体识别 数据挖掘 深度学习 医疗信息 

分 类 号：U491.14[交通运输工程—交通运输规划与管理]