摘要
[目的/意义]针对当前对突发公共卫生重大传染病事件的决策支持研究较少,相关实体知识抽取模型建设不足的情况,本研究基于语料库技术、机器学习、深度学习模型,构建重大传染病实体知识抽取模型,为突发公共卫生重大传染病事件提供决策支持。[方法/过程]以人民网传染病新闻、百度百科传染病词条、中科院病毒案例库为研究对象,采用条件随机场模型、循环神经网络模型、预训练文本表征模型,基于BERT对突发公共卫生重大传染病事件的实体知识进行识别与分析,并通过可视化的方式,对传染病的时序演化情况进行了分析。[结果/结论]构建了基于BERT的突发公共卫生重大传染病事件实体知识自动抽取模型,其精准率、召回率、调和平均值分别达到84.09%、87.71%、85.86%,可为相关部门决策提供及时、可靠、有效的信息。
[Purpose/significance]In view of the lack of research on decision-making support for major public health emergencies in the current research,and the insufficient construction of relevant knowledge bases,this study established a corpus of physical knowledge of major infectious diseases for public health emergencies Provide decision support for major infectious disease events.[Method/process]Taking People's Daily Online infectious disease news,Baidu Encyclopedia infectious disease entry,Chinese Academy of Sciences virus case database as the research objects,using conditional random field model,recurrent neural network model,pretraining text representation model,BERT for public health emergencies of the entity knowledge of major infectious disease events is identified and analyzed,and the temporal evolution of infectious diseases is analyzed through visualization.[Result/conclusion]A BERT-based automatic extraction model of physical knowledge of major public health infectious disease events was constructed,and its accuracy,recall,and reconciliation average reached 84.09%,87.71%,85.86%.It can provide timely,reliable and effective information for decision-making of relevant departments.
作者
江川
王东波
JIANG Chuan;WANG Dongbo(School of Information Management,Nanjing Agricultural University,Nanjing 210095)
出处
《科技情报研究》
2021年第2期23-35,共13页
Scientific Information Research
基金
国家社会科学基金重点项目“总体国家安全观下的国家情报工作制度创新研究”(编号:20ATQ004)。
关键词
BERT
重大传染病
决策支持
实体识别
深度学习
公共卫生
BERT
major infectious diseases
decision support
entity recognition
deep learning
public health
