Topic Modeling of Health Rumors Based on Anti-Rumor Tweets on the WeChat Platform: Machine Learning Analysis (Preprint)

Abstract

Social network has become one of the main channels for the public to obtain health information. However, it has also become a source for the spread of health-related misinformation. Health-related misinformation seriously threatens the public’s physical and mental health. Topic identification is the premise of health-related misinformation governance.

This study aims to explore the types of health-related misinformation favored by WeChat public platform users and their prevalence trends.

We used a web crawler tool to capture health rumor-dispelling tweets collected on the rumor-dispelling public account. We collected text information from health-debunking tweets posted between January 1, 2016, and August 31, 2022. Following word segmentation of the collected text, a document topic generation model, Latent Dirichlet Assignment, is used to identify and generalize the most common topics. The proportion distribution of themes was calculated, and the negative impact of various health rumors in different periods was subsequently analyzed. Additionally, the prevalence of health rumors was analyzed using the number of health rumors generated at each time point.

From January 1, 2016 to August 31, 2022, we collected 9,366 rumor-refuting tweets from WeChat official accounts. Through topic modeling, we divided the health rumors into eight topics, including the prevention and treatment of infectious diseases (n = 1,284, 13.71%), disease therapy and its effects (n = 1,037, 11.07%), food safety (n = 1,243, 13.27%), cancer and its causes (n = 946, 10.10%), regimen and disease (n = 1,540, 16.44%), rumors of transmission (n = 914, 9.76%), healthy diet (n = 1,068, 11.40%), and nutrition and health (n = 1,334, 14.24%). Furthermore, we summarized the eight topics into four themes, including public health, disease, diet and health, and rumor spreading.

Our study shows that the topic model can provide analysis and insights into health rumor governance. The analysis of rumor development trends shows that public health, disease, and diet and health problems are the most affected areas of rumors. Governments still need to consider national conditions, formulate appropriate policies, and deal with health rumors more comprehensively. While ensuring the health of the Internet, we should also improve the level of national quality education. We recommend that additional sentiment analysis-related studies be conducted to verify the impact of health rumor-related topics.