数智时代政务新媒体中的政府表达与公众互动机制研究——基于自然语言处理与全国政务微博面板数据的分析
Government Communication and Public Engagement Mechanisms on Government Social Media in the Digital-Intelligent Era: Evidence from Natural Language Processing and a National Panel of Government Weibo Posts
摘要
数智技术正在重塑政府与公众的连接方式，政务新媒体不只是信息发布工具，更是观察平台化治理中政府表达、公众回应与互动秩序形成的重要场域。研究以337个认证政务微博账号的328544条原创博文为样本，运用NLP和机器学习方法测度可读性、情绪强度、媒体丰富度与议程波动，构造残差绩效指标，并采用账号和月份双向固定效应模型进行分析。结果显示：可读性具有稳定正向影响，媒体丰富度和议程波动呈现当期或较弱正向作用，情绪强度缺乏稳定直接促进效应；政策关注度直接作用不稳定，但具有一定调节效应，尤其会改变情绪表达的边际作用。研究认为，政务微博互动提升的关键在于降低公众理解成本，并依据政策关注情境以及地区、账号差异优化表达方式。
Abstract
Digital-intelligent technologies are reshaping the ways in which governments connect with the public. Government social media is not merely a tool for information dissemination, but also an important arena for observing government communication, public response, and the formation of interactive governance orders in platform-based governance. Drawing on 328,544 original posts from 337 verified government Weibo accounts in China, this study employs natural language processing and machine learning methods to measure four content features: readability, emotional intensity, media richness, and agenda volatility. It further constructs a residual performance indicator to capture excess public engagement after accounting for baseline account popularity and routine posting capacity, and estimates account-month two-way fixed effects models.The results show that readability has a stable and positive effect on excess interaction performance, suggesting that reducing citizens cognitive costs is central to effective government communication. Media richness and agenda volatility exert contemporaneous or relatively weaker positive effects, while emotional intensity does not generate a stable direct positive effect. Policy attention does not show a robust direct effect, but it serves as an important contextual moderator, especially by reshaping the marginal effect of emotional expression. The study argues that improving public engagement on government Weibo depends less on creating superficial visibility than on providing clear, credible, understandable, and context-sensitive communication. These findings contribute to the study of government social media by shifting the focus from communication performance evaluation to the mechanisms linking government expression, public attention, and interactive response.关键词
政务新媒体/互动绩效/数智治理/自然语言处理/注意力分配Key words
Government social media/interaction performance/digital-intelligent governance/natural language processing/attention allocation引用本文复制引用
周璜达,赵乐言.数智时代政务新媒体中的政府表达与公众互动机制研究——基于自然语言处理与全国政务微博面板数据的分析[EB/OL].(2026-07-09)[2026-07-11].https://chinaxiv.org/abs/202607.00053.学科分类
信息传播、知识传播/计算技术、计算机技术
