Topic Modeling of Social Media Discourse of Autism Support Groups

Social media platforms serve as critical channels for autism support groups to communicate and seek assistance. This study employed Latent Dirichlet Allocation (LDA) topic modeling to analyze discourse patterns within the Autism Bar on Baidu Tieba, a major Chinese social media. A dataset of 14,151 posts was collected through web crawling, with 12,667 posts retained after preprocessing. The analysis revealed two key findings: (1) The discourse among autism support communities on Baidu Tieba focuses on four central themes: intervention and therapy, early educational journey, early symptom detection and family interaction, and access to educational resources and community support. (2) Sociocultural factors exert a significant influence on autism-related discourse, particularly in shaping societal attitudes toward individuals with autism and the formation of support networks. Traditional Chinese cultural values, such as collectivism and familial centrality, impact the behavioral patterns and decision-making processes of families with autistic children. This study has demonstrated the unique needs and challenges faced by the autism support community, while also informing strategies to promote social media platforms as spaces for support and information exchange. The findings have practical implications for designing targeted interventions and support mechanisms for individuals with autism and their families.