Probing Social Support and Self-Disclosure Within Chinese Online Domestic Violence Support Groups: Leveraging Multiple Machine Learning Approaches

This study examines how domestic violence victims can effectively disclose their experiences in online support groups to receive meaningful support, and how community members can provide that support. Analyzing 3455 posts and 55,787 replies from Chinese domestic violence support groups using BERT model, it classifies support types and explores the relationship between disclosure topics, emotional expression, and support types through structural topic modeling and logistic regression. Findings suggest that verbal-violence disclosures receive more emotional support, and expressions of sadness and positive emotions foster empathy. The research provides practical insights for online domestic violence interventions by demonstrating how disclosure topics and emotional expressions influence social support dynamics in Chinese online communities, thereby extending understanding of domestic violence discourse beyond a Western context.