This study analyses the agenda setting on social media in the COVID-19 pandemic by exploiting one of the disruptive technologies, big data analytics. Our purpose is to examine whether the agenda of news organisations matches the public agenda on social media in crisis situations, and to explore the feasibility and efficacy of applying big data analytics on social media data. To this end, we used an unsupervised machine learning approach, structural topic modelling and analysed 129,965 tweets posted by UK news media and citizens during April 2, and 8, 2020. Our study reveals a wide diversity of topics in the tweets generated by both groups and finds only a small number of topics are similar, indicating different agendas set in the pandemic. Moreover, we show that citizen tweets focused more on expressing feelings and sharing personal activities while news media tweets talked more about facts and analysis on COVID-19. In addition, our results find that citizens responded more significantly to breaking news. The findings of the study contribute to the agenda setting literature and offer valuable practical implications.

本研究利用其中一种颠覆性技术大数据分析，分析了 COVID-19 大流行期间社交媒体上的议程设置。我们的目的是检查危机情况下新闻机构的议程是否与社交媒体上的公共议程相匹配，并探讨将大数据分析应用于社交媒体数据的可行性和有效性。为此，我们使用了一种无监督机器学习方法、结构化主题建模，并分析了英国新闻媒体和公民在 2020 年 4 月 2 日至 8 日期间发布的 129,965 条推文。我们的研究揭示了这两个群体生成的推文中主题的广泛多样性并发现只有少数主题相似，表明在大流行中设定了不同的议程。而且，我们表明，公民推文更侧重于表达感受和分享个人活动，而新闻媒体推文更多地谈论 COVID-19 的事实和分析。此外，我们的结果发现，公民对突发新闻的反应更为显着。该研究的结果有助于议程设置文献，并提供有价值的实际意义。