LDA-Based Topic Modelling on COVID-19 FLiRT Variant in Social Media
摘要
This study examines the sentiments of Filipinos toward the FLiRT variant of COVID-19, as expressed on social media, particularly Facebook. Using Apify.com’s scraping tool, 1,452 comments were collected from posts discussing isolation and emergence, vaccines and immunity, and quarantines. To guarantee data consistency and accuracy, the comments underwent thorough preprocessing, including translation from Tagalog to English and tokenization and lemmatization. Then, Latent Dirichlet Allocation (LDA) for topic modeling revealed underlying themes in public discourse. The model’s performance was assessed using coherence scores and perplexity, with the results highlighting prevalent concerns about vaccine hesitancy, pandemic fatigue, and increasing distrust in government and healthcare institutions. The study highlights the significant role of social media in shaping public perceptions and behaviors during the pandemic. It shows the importance of clear, accurate communication from authorities to address misinformation and public fears effectively. Future research should expand the dataset across multiple platforms, enhance translation processes, and validate topic modeling results through expert collaboration to improve the accuracy and applicability of findings in addressing public health challenges.