Extracting Official Agencies’ Communication Patterns During the COVID-19 Pandemic: A Text Mining Approach

This study offers an extensive analysis of government authorities’ online COVID-19 communication throughout the pandemic, with a specific case study in Hyogo Prefecture, Japan. Examining 85 documents, our research seeks to discern the primary themes within government messaging and assess the conveyed tone and sentiment. Utilizing diverse text mining techniques, including word analysis, collocation analysis, topic modeling, sentiment analysis, and correlation analysis, we identify six themes in government COVID-19 communication: measures for business and healthcare support, preventive measures and vaccination efforts, communication and awareness initiatives, the healthcare system, testing and patient support, and sales and business operations within restricted environments, as well as event management and safety protocols. Significantly, our analysis reveals a transition in the emotional sentiment from predominantly positive to negative in September 2020, closely mirroring a peak in COVID-19 cases in mid-August of the same year. Correlation analysis demonstrates moderate positive associations between the top five emotions and COVID-19 cases in Hyogo Prefecture. These findings provide valuable insights into government communication and its interaction with the evolving COVID-19 landscape.