Analysis of COVID-19-Related User Content on the Baseball Bulletin Board in 2020 through Text Mining

Shang Chun Ma, Ching Ya Su, Sheng Fong Chen, Shintaro Sato, Shang Ming Ma

Research output: Contribution to journalArticlepeer-review


The world engaged in online sport watching during COVID-19. Fortunately, in Taiwan, the pandemic was stably controlled in 2020, allowing for the continuation of the Chinese Professional Baseball League (CPBL); this attracted international attention and encouraged relevant discussions on social media in Taiwan. In the present study, through text mining, we analyzed user content (e.g., the concepts of sports service quality and social identity) on the Professional Technology Temple (PTT) baseball board—the largest online bulletin board system in Taiwan. A predictive model was constructed to assess PTT users’ COVID-19-related comments in 2020. A total of 422 articles and 21,167 comments were retrieved. PTT users interacted more frequently during the closed-door period, particularly during the beginning of the CPBL in April. Effective pandemic prevention, which garnered global attention to the league, generated a sense of national identity among the users, which was strengthened with the development of peripheral products, such as English broadcasting and live broadcasting on Twitch. We used machine learning to develop a chatbot for predicting the attributes of users’ comments; this chatbot may improve CPBL teams’ understanding of public opinion trends. Our findings may help stakeholders develop tailored programs for online spectators of sports during pandemic situations.

Original languageEnglish
Article number551
JournalBehavioral Sciences
Issue number7
Publication statusPublished - 2023 Jul

All Science Journal Classification (ASJC) codes

  • Ecology, Evolution, Behavior and Systematics
  • Development
  • Genetics
  • General Psychology
  • Behavioral Neuroscience


Dive into the research topics of 'Analysis of COVID-19-Related User Content on the Baseball Bulletin Board in 2020 through Text Mining'. Together they form a unique fingerprint.

Cite this