网民眼中的政府反腐败——基于网络爬虫和结构主题模型的分析(2012-2017)

Government Anti-corruption in the Eyes of Netizens--Analysis Based on Structural Topic Models of Network(2012-2017)

大数据分析技术与廉政建设相结合成为新时期廉政研究的重要方向。通过网络爬虫获取2012-2017年的海量微博数据,运用结构主题模型方法,聚焦党的十八大以来网民讨论的20个政府反腐败主题。研究发现,大部分微博内容与“中国官场—大地震”主题紧密相关,说明网民对党和政府反腐败的力度和决心之大感到震撼,微博所讨论的话题与纪检监察机关的反腐败工作重点具有高度的契合性。通过统计建模发现,网民性别和网络影响力显著影响主题占比高低和主题内容差异。该研究在考量方法上,有利于实现政府反腐败绩效评价的具体化,弥补以往单一指标评价的偏误性;在政策实践上,党风廉政建设宜重点关注公众需求,切实提高公民反腐败获得感。

The combination of big data analysis technology and clean government construction has become an important direction of social science research in the new era. Through the python to obtain massive microblog data(2012-2017), using the structural theme model method, this paper identifies 20 topics discussed by the government against anti-corruption netizens since the 18 th Congress of Communist Party of China. The study found that most of Weibo content is closely related to the theme of ″China Officialdom-Great Earthquake″, indicating that netizens are shocked by the strength and determination of the government to fight corruption. The topics discussed on Weibo and the anti-corruption work focus of the discipline inspection and supervision organs has a high degree of fit. Through statistical modeling, it is found that the gender and network influence of netizens significantly affect the proportion of topics and the difference in subject contents. The above findings have realized the concreteness of government anti-corruption performance evaluation in terms of methods, and got rid of the bias of past single-index evaluation. In policy practice, it is advocated that the government′s clean government construction should focus on public demand and effectively improve citizens′ sense of anti-corruption.