As a key element of the sharing economy, ride-sharing plays a vital role in promoting sustainable urban mobility by optimizing vehicle utilization rates, lowering carbon emissions, and alleviating traffic congestion. Despite its cost-efficiency and sustainability benefits, ride-sharing adoption remains limited in the post-pandemic period due to behavioral changes and safety concerns. Accordingly, using survey data from 425 commuters in Hefei, concerns about COVID-19 and satisfaction with ride-sharing services were integrated into the theory of planned behavior framework. Structural equation modeling was applied to examine the relationship between ride-sharing intention and actual usage behaviors. The results indicated that ride-sharing intention was significantly positively affected by subjective norms (β = 0.428 ***), service satisfaction (β = 0.315 ***), and perceived behavioral control (β = 0.162 *), but significantly negatively affected by concerns about COVID-19 (β = −0.183 **). Concerns about COVID-19 significantly negatively affected travelers’ actual ride-sharing behaviors (β = −0.2 **). Furthermore, ride-sharing intention was identified as a significant positive predictor of travelers’ behaviors: specifically, their likelihood of accepting a ride-sharing order (β = 0.395 ***). These findings offer transport authorities evidence-based strategies for designing targeted interventions during health crises, particularly through reinforcing social norms, improving service quality, and implementing transparent health protocols to ensure both user safety and sustainability.

As network scale and demand rise, the Uncapacitated Facility Location Problem (UFLP), a classical NP-hard problem widely studied in operations research, becomes increasingly challenging for traditional methods confined to formulation, construction, and benchmarking. This work generalizes the UFLP to network setting in light of demand intensity and network topology. A new initialization technique called Network- and Demand-Weighted Roulette Wheel Initialization (NDWRWI) has been introduced and proved to be a competitive alternative to random (RI) and greedy initializations (GI). Experiments were carried out based on the TRB dataset and compared eight state-of-the-art methods. For instance, in the ultra-large-scale Gold Coast network, the NDWRWI-based Neighborhood Search (NS) achieved a competitive optimal total cost (9,372,502), closely comparable to the best-performing baseline (RI-based: 9,189,353), while delivering superior clustering quality (Silhouette: 0.3859 vs. 0.3833 and 0.3752 for RI- and GI-based NS, respectively) and reducing computational time by nearly an order of magnitude relative to the GI-based baseline. Similarly, NDWRWI-based Variable Neighborhood Search (VNS) improved upon RI-based baseline by reducing the overall cost by approximately 3.67%, increasing clustering quality and achieving a 27% faster runtime. It is found that NDWRWI prioritizes high-demand and centrally located nodes, fostering high-quality initial solutions and robust performance across large-scale and heterogeneous networks.

Optimizing networks under large-scale, multidimensional, and dynamic demand presents a significant challenge. This complexity is further amplified when solving network-based Uncapacitated Facility Location Problems (UFLP), which are well-known for their NP-hardness. Current large-scale analytical approaches, such as Neighborhood Search (NS) and Variable Neighborhood Search (VNS), face limitations in either solution quality or computational efficiency, predominantly due to their reliance on either unstable Random Initialization (RI) or computationally expensive Greedy Initialization (GI). Considering demand intensity and network topology, this research proposes the Demand-Weighted Roulette Wheel Initialization (DWRWI) strategy. This novel initialization method strategically prioritizes high-demand and centrally located network nodes, thereby creating high-potential initial facility configurations for improvement algorithms. Scenario-based testing demonstrates several distinctive strengths of DWRWI: Compared to RI and GI, DWRWI better identifies high-demand, central network nodes, yielding superior initial solutions early in optimization. For example, in the large-scale network, DWRWI-initialized NS achieved a higher Silhouette score (0.3859) than RI (0.3833) and GI (0.3752). Additionally, DWRWI reduces computation time by about 28% compared to Greedy-initialized NS while maintaining competitive costs. By integrating demand-weighted distance and stochastic optimization theory, DWRWI significantly reduces computation time while ensures more uniform customer-to-facility assignments aligned closely with spatial demand distributions.