The online food delivery service supply chain constitutes a crucial element in achieving sustainable development goals. With its prosperity, an increasing number of takeaway businesses are drawn to this sector. As their numbers rise, issues such as profitability resilience, environmental friendliness, and fulfillment of social responsibility emerge, posing potential disruptions to the service supply chain. Against this backdrop, our endeavor is to mine the sustainability of takeaway businesses using the triple bottom line. We propose a two-stage approach involving the Bayesian best-worst method and a data mining technique to derive the weights of sustainability criteria and the clusters of takeaway businesses. Subsequently, a case study is conducted focusing on takeaway businesses on the Ele.me platform in China. The results highlight economic sustainability as the most valued criterion, followed by social and environmental sustainability. Clustering outcomes reveal four distinct levels of sustainability, with a stronger performance in social sustainability compared to environmental and economic dimensions. Further discussions explore the relationship between sustainability levels, cuisine categories, and business size. Consequently, this study suggests an effective approach for advancing sustainability initiatives within the online food delivery service supply chain.
Algorithmic management has markedly improved operational efficiency in the platform economy, yet its consequences for the occupational health of platform workers remain insufficiently understood. This study therefore examines how efficiency-driven algorithmic pressures produce occupational health burdens among online food delivery (OFD) riders in China. The analysis drew on 10,103 valid rider comments from a major Chinese social media platform, and it combined Latent Dirichlet Allocation topic modeling with co-occurrence network analysis and sentiment intensity assessment. Semi-structured interviews with 32 OFD riders were added to corroborate these computational findings and to reduce reliance on a single data source. Expert review identified five core health burden dimensions: physical exhaustion, social devaluation, disciplinary distress, injury vulnerability, and health-protection deficit. Network analysis showed that these dimensions form a tightly interconnected structure with physical exhaustion as the central hub, and that pressure cascades from operational demands into punitive mechanisms and then into injury risk and social protection deficits. A clear gap emerged between structural prominence and emotional intensity. Riders discussed time pressure most widely and largely accepted it as routine, while punitive mechanisms and health-protection deficits drew the strongest negative responses. The study contributes an integrated text-mining framework for occupational health research. It reframes rider health as a cascading system of mutually reinforcing burdens rather than a set of isolated risks, and it shows that the most central burden is not the one felt most acutely. These insights point to concrete measures for governments, platforms, rider organizations, and industry associations such as portable occupational-injury insurance, algorithmic transparency, fairer timing and rating rules, and accessible grievance channels.