政务新媒体互动内容分析及诉求回应研究——以人民网留言板数据分析为例  

Research on Interactive Content Analysis and Appeal Response of Government New Media:Taking the Message Boards of People's Daily Online in Central Megacities as an Example

作　　者：熊思斯 滕宇 胡珀[1,2] Xiong Sisi;Teng Yu;Hu Po(School of Computer,Central China Normal University,Wuhan 430079;National Language Resources Monitoring&Research Center for Network Media,Central China Normal University,Wuhan 430079)

机构地区：[1]华中师范大学计算机学院,武汉430079 [2]国家语言资源监测与研究网络媒体中心,武汉430079

出　　处：《情报杂志》2024年第3期150-156,9,共8页Journal of Intelligence

基　　金：国家语委科研项目“政务新媒体互动内容特征分析与摘要生成研究”(编号:YB135-149)研究成果。

摘　　要：[研究目的]网络问政已成为新时代主要的政民互动方式之一。通过充分挖掘网络政民互动内容的特征,可以了解民众的民意诉求,亦可以及时发现地方政府部门工作中有待改善之处。[研究方法]根据框架理论,采用自然语言处理技术挖掘中部超大城市领导留言板互动内容中的语言特征、内容特征及行为特征,并分析不同特征间的潜在关联。[研究结论]研究表明,语言特征方面,民众留言弱者框架的占比最高,抗争框架所占比例较低。政府偏好使用描述话语和共情话语,规则话语的使用频次较低。内容特征上,各项领域内容特征与城市统计年鉴相应数据存在关联性,能够客观反映各领域存在的问题,且通过聚类能够有效发掘热点议题和实体辅助政府回应。行为特征方面,民众投诉/求助类留言比例与人均地区生产总值具有相关性。政府回复的行为特征与民众留言的内容特征存在潜在关联,其中政府行为的处理类型受民众留言的内容特征影响最大,缺乏关注语言特征和行为特征。[Research purpose]Online governance has become one of the main ways of interaction between government and the people in the new era.By fully exploring the characteristics of online political civilian interaction content,we can understand the public's public opinion demands and timely identify areas for improvement in the work of local government departments.[Research method]According to the framework theory,natural language processing technology is used to mine the language characteristics,content characteristics and behavior characteristics in the interactive content of leader message boards in the central megacities,and analyze the potential relationship between different characteristics.[Research conclusion]Research has shown that in terms of language characteristics,the proportion of weak message frames among the public is the highest,while the proportion of resistance frames is relatively low.The government prefers to use descriptive discourse and empathetic discourse,while the frequency of using rule-based discourse is relatively low.In terms of content features,there is a correlation between the content features of various fields and the corresponding data of the urban statistical yearbook,which can objectively reflect the problems in various fields.Clustering can effectively explore hot topics and assist government responses through entities.In terms of behavioral characteristics,the proportion of public complaints/help seeking messages is related to the per capita Gross regional product.There is a potential correlation between the behavioral characteristics of government responses and the content characteristics of public messages,with the processing type of government actions being most influenced by the content characteristics of public messages,lacking attention to language and behavioral characteristics.

关 键 词：政务新媒体 网络问政 留言板 民众留言 政府回复 政府行为 情感分析 主题聚类 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]