Interpretable Topic Modelling of Mental Health Conversations in Australian Online Forums Using Network-Based Insights
摘要
Mental health conditions such as anxiety, depression, post-traumatic stress disorder (PTSD), and suicidal thoughts significantly impact the wellbeing of many Australians. Beyond Blue, a leading national mental health platform, offers a unique forum where individuals anonymously share experiences and seek support. Compared to global platforms like Reddit and Twitter, this community remains relatively underexplored as a data source for understanding lived experiences of mental illness. In this study, we apply topic modelling and quantitative analysis to Beyond Blue forum posts, identifying key themes and subtopics within categories such as medication, therapy, relationships, and work stress. Using MentalBERT-enhanced BERTopic, we also construct a visual network of subtopics, capturing both diversity and overlap across conditions. The findings reveal the complexity of how Australians articulate distress, cope with challenges, and engage in peer support. This work provides insights to guide tailored interventions, inform digital mental health services, and support policies grounded in lived experience.