Public discourse dynamics of shared e-bikes under policy evolution: A social media analysis

Shared e-bikes constitute a critical component of last-mile micromobility infrastructure, yet public perception under shifting regulatory frameworks remains underexplored. This study examines public discourse across four evolving policy stages in China, namely emergence, prohibition, trial, and differentiation, by analyzing 22,337 Weibo comments (2017–2025) using bidirectional encoder representations from transformers (BERT) for sentiment classification, latent Dirichlet allocation (LDA) for topic modeling, and association rule mining. Results reveal that positive sentiment dominates, alongside substantial and stable gender disparities. Specifically, males consistently exhibit higher sentiment scores than females, whereas region exerts minimal overall effect. Temporally, sentiment scores vary significantly across policy stages, and the differentiation stage records peak negativity. Topic modeling extracts eight core themes encompassing technological development, environmental benefits, transit connectivity, and operational governance, revealing that public attention exhibits dynamic shifts across policy stages. Transit connectivity maintains persistently high attention, and sentiment toward market order progressively improves, whereas discussions on environmental values decline substantially over time. Association rule mining demonstrates that the specific pathways driving public sentiment vary significantly across user demographics and regional contexts, revealing distinct operational demands. These findings demonstrate the utility of social media data in capturing policy impacts across diverse contexts, offering empirical insights for urban micromobility governance.