Exploring Conversational Dynamics in Scientific and Pseudoscientific Health Communities on YouTube: A Process Mining and Network Analysis Study (Preprint)

Le résumé fourni par la source

BACKGROUND Social media platforms, particularly YouTube, are primary sources of health information but also significant vectors for misinformation and pseudoscience. While many studies analyze the content and sentiment of this information, the dynamic, sequential nature of user interactions, which shapes belief and community formation, remains poorly understood. OBJECTIVE This study aimed to identify and compare the structural and emotional patterns of conversational flow within YouTube comments sections of videos discussing scientific versus pseudo-scientific health treatments. METHODS We collected a large corpus of YouTube comment threads posted between 2011 and 2025 from videos categorized as either “scientific” (20,387 comments) or “pseudo-scientific” (32,025 comments) using an automated pipeline that combined API-based data extraction, large language model-based video classification, and natural language processing techniques for multilingual sentiment and thematic classification of comments. We then applied process mining to model the temporal sequences of interactions and network analysis to map the relationships between conversational topics. RESULTS Network analysis revealed divergent conversational cores: scientific communities centered on balanced "Expression of feelings" (Positive/Negative) and "Comparison-Negative," with "Medical Treatment" and "Advice request" prominent and "Insult" marginal; pseudo-scientific communities showed high density among positive-affect nodes ("Expression of feelings-Positive," "Thanking," "Compliment") alongside notable "Insult" and negative comparison influence. Process mining confirmed these patterns sequentially: scientific flows incorporated heterogeneous negative/neutral trajectories that resolved without escalation, while pseudo-scientific flows were more homogeneous, dominated by positive reinforcement ("Expression of feelings-Positive" → "Thanking/Compliment/Desires") with negative expressions marginal in dominant sequences. CONCLUSIONS Scientific discourse appears more compatible with mixed-valence evaluation, medical context exchange and action oriented communication, whereas pseudo-scientific discourse shows stronger socio-emotional bonding alongside episodic incivility. These findings suggest that public health strategies should address the affective and community-bonding drivers of engagement in misinformation communities beyond mere information provision.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.