摘要
[研究目的]大数据时代,提升政府治理能力和人民群众生活幸福感是各级政府核心任务,尤其是从群众留言文本中获取民众关注的热点问题,从而快速响应并及时解决群众反馈的问题,从群众留言文本中提取有价值的热点对政府部门显得尤其紧迫和必要。[研究方法]以从上海市政府信箱、上海市委信箱、上海信访的“互联网+政务”网络平台爬取的群众留言29074条数据为研究对象,对比分析BERT、LSTM以及FastText三个模型的分类效果,选择BERT模型构建群众留言热点追踪框架,对反馈最多的住房规划、医疗卫生和交通出行三类问题使用Affinity Propagation聚类算法与BERT实体命名识别相结合的方式对每个分类下的问题进行聚类,最后通过聚类形成的问题进行热度计算得出群众反馈的热点问题。[研究结论]从热点追踪的结果可知住房安全、疫情防控和两港快线设立分别为住房规划、医疗卫生和交通出行三类留言中讨论热度最高的问题,据此提出针对性建议。
[Research purpose]In the era of big data,improving government's governance ability and the people's happiness in life is the core task of governments at all levels.It is especially urgent and necessary to extract valuable hotspots from the mass message text,which could help identify the problems concerned by the people,enabling government to solve the problems in time.[Research method]This article crawls data from the"Internet+Government Affairs"network platform of Shanghai Municipal Government Mail,Shanghai Municipal Committee,Shanghai Letters,the Top 29074 mass messages are used as the research object.The classification effects of three models of BERT,LSTM,and FastText are compared.The BERT model is selected to build the mass message hotspot tracking framework,the housing planning,health and transportation problems are clustered via combing the Affinity Propagation clustering algorithm with BERT entity naming identification.Finally the hotspot problems of the mass are calculated out of the problems identified by clustering.[Research conclusion]From the results of hotspot tracking,it can be seen that housing safety,epidemic prevention and control,and the establishment of the two-port express line are the most discussed issues in the three categories of housing planning,medical care,and transportation.Based on this,targeted suggestions are put forward.
作者
徐绪堪
印家伟
王晓娇
Xu Xukan;Yin Jiawei;Wang Xiaojiao(School of Business,Hohai University,Nanjing 211100;Changzhou Key Laboratory of Industrial Big Data and Knowledge Management,Changzhou 213022)
出处
《情报杂志》
CSSCI
北大核心
2022年第9期136-142,78,共8页
Journal of Intelligence
基金
国家社会科学基金重大项目“总体国家安全观下重大突发事件的智能决策情报体系研究”(编号:20&ZD125)子课题“总体国家安全观下重大突发事件智能决策情报体系的基本理论与方法研究”
河海大学中央高校业务费资助项目(编号:B220207038)研究成果之一。
