Share Paper: Tweeting about Teachers and COVID-19: An Emotion and Sentiment Analysis Approach

  1. Jason Harron, Kennesaw State University, United States
  2. Sa Liu, Harrisburg University of Science and Technology, United States
Wednesday, April 13 4:45 PM-5:15 PM Shorebreak

Abstract: This paper applies Natural Language Processing (NLP) techniques to analyze 20-months of Twitter data related to the general publics’ discourse about the COVID-19 pandemic and teachers. A total of 68,340 English-language tweets were analyzed using semi-supervised topic modeling, which generated 17 topics of interest that focused on schools opening/closing, online learning, vaccines, teacher appreciation, and masks. Using a Robustly optimized BERT approach (RoBERTa), emotion and sentiment analysis were performed longitudinally across all generated topics. Analysis indicated an overwhelmingly negative sentiment, with a focus on emotions of anger and sadness over the 20-month period. Positive sentiment was associated with teacher appreciation ...