Mapping the Pandemics Echo: Dynamic Narrative Detection and Spatio-Temporal Sentiment Modeling of COVID-19 Discourse on Twitter

Traditional sentiment analysis treats tweets as independent, static samples, missing the evolving and geographically heterogeneous nature of public opinion during the pandemic. To address this, the authors integrated COVID-Twitter-BERT for fine-grained sentiment classification with BERTopic for dynamic topic modeling, enabling automatic discovery and tracking of narratives over time and space. Using a corpus of 2.4 million geolocated tweets collected from January 2020 to June 2022, the analysis reveals three distinct pandemic phases: early fear-driven narratives about mask shortages (Q1 2020), a period of vaccine optimism followed by polarization (2021), and pandemic fatigue (2022). Regional comparisons show significant differences, with US discourse dominated by freedom-versus-mandate debates while European discussions emphasized collective solidarity. The framework achieved a 76% F1-score in sentiment classification and successfully identified 50 distinct narratives with high coherence scores. This work provides a powerful methodology for real-time epidemiological narrative surveillance and crisis communication monitoring, allowing public health officials to track public concerns and adjust messaging accordingly.