[1L5-OS-4b] OS-4

[1L5-OS-4b-01] Quantifying the Topic, Moral Foundation and Relationship of Toxicity Towards the COVID-19 Vaccine on Twitter

〇Tomoka Nakazato¹, Yuya Shibuya¹, Masaki Onishi², Soichiro Takagi¹ (1. The University of Tokyo, 2. AI research center, AIST)

Keywords:Large-scale social data analysis, Misinformation, Disinformation, SNS, Public Health

This research endeavors to explore the discourse surrounding COVID-19 vaccines on Twitter in 2020, examining the relationship between discourse topics, the moral foundations underpinning vaccine-related keywords, and the extent of toxicity within these communications. Utilizing BERTopic for topic classification and the Japanese Moral Foundations Dictionary (J-MFD) for identifying the underlying moral foundations of vaccine-related keywords, the study uncovers a notable correlation between discourse themes, their moral underpinnings, and the toxicity levels of the tweets. The analysis reveals that toxicity levels fluctuate based on temporal trends, topics, and moral foundations. The research posits that the strategic and accurate dissemination of information could mitigate toxicity for certain topics and keywords prone to inciting toxic responses. Additionally, the findings indicate that toxic tweets may emanate from a broad spectrum of users, rather than a limited group. This implies the need for strategies to monitor toxic speech and the spread of misinformation on social media platforms. This investigation offers critical insights into the dynamics between discourse topics, moral foundations, and toxicity levels in vaccine-related Twitter conversations, suggesting avenues for enhancing public discourse quality on social media.