<p>This study systematically investigates the spatial evolution and driving mechanisms of urban tourism flow networks in Shanghai by integrating complex network analysis, GIS-based spatial modeling, and a negative binomial regression framework, using user-generated travelogue data from online platforms. The findings show that, despite significant disruption caused by the COVID-19 pandemic, Shanghai’s tourism flow network demonstrated strong resilience, exhibiting prominent small-world properties and high clustering coefficients. Its topological structure features a stable core, multiple emerging nodes, and outward spatial diffusion. Traditional landmarks such as the Bund, Lujiazui, and Nanjing Road remain critical hubs. At the same time, emerging and suburban attractions are gaining prominence, signaling an ongoing expansion of tourism activity into peripheral areas. Hierarchical analysis indicates a structural shift from fragmented, loosely connected multi-cluster configurations toward a core-dominated, multi-cluster collaborative system. The network’s integration level has improved, with the Bund maintaining absolute centrality throughout the observation period, as the number of subgroups declined. Centrality metrics highlight persistent spatial concentration alongside dynamic adjustments, with inner-city attractions maintaining high connectivity and consistently strong betweenness centrality. Regression results show that a site’s geographic location, popularity, historical significance, and spatial capacity all positively influence its network centrality. Among these, spatial location and scale effects are especially strong, underscoring the key role of intrinsic attributes in shaping network position. In contrast, transportation accessibility is not statistically significant, likely due to Shanghai’s overall high level of urban connectivity.</p>