A Data-Driven Framework For Analyzing And Predicting Social Media Engagement During Rumor Propagation

The spread of rumors on social media has become a significant concern, as platforms like X (formerly known as Twitter) enable rapid dissemination of unverified information. These rumors can shape public perception and behavior, making it crucial to understand the dynamics of their spread. The main objective of this study is to explore how users interact with rumor-related tweets and identify key factors that predict tweet engagement. By analyzing tweets from seven different rumor events, this research aims to uncover patterns in user engagement and provide insights into the spread of misinformation. The study utilized a dataset of tweets from events concerning COVID-19, Antifa, the 2020 Presidential Election, and the war in Ukraine. Machine learning models were employed to analyze engagement metrics such as retweets, quotes, likes, and replies. Key factors like user information and tweet sentiment were identified as significant predictors of engagement.The findings reveal that tweets from users with a large number of followers and those perceived to have positive or negative sentiment receive higher engagement. User information emerged as the most influential factor, followed by tweet sentiment. These insights can guide efforts to control and manage the spread of rumors on social media, helping to develop strategies to mitigate the negative effects of misinformation and enhance the reliability of information shared online.