A spatiotemporal knowledge graph for driving factor analysis and growth prediction of regional economy in China

Knowledge Graphs (KGs), as a tool for organizing and integrating multidimensional data, have become increasingly important in addressing multifaceted challenges in geographic research. However, most existing KGs rely on static data to organize information, overlooking the dynamic and spatiotemporal properties inherent in real-world scenarios, thereby resulting in incomplete representations of spatiotemporal knowledge. Similarly, traditional economic analysis methods struggle to unravel the complex relationships among numerous factors. Thus, to address these limitations, we integrated multidimensional spatiotemporal data and combined the complex relationships between multiple factors of economic activities to construct a regional economic spatiotemporal knowledge graph (REST-KG), enabling the modeling of systemic economic interactions. Utilizing relational graph convolutional networks (R-GCNs), we trained knowledge representations on REST-KG to further identify the main driving factors of economic growth and predict future trends in key economic indicators. Our findings reveal that the primary drivers of China's economic growth exhibit temporal variations across different periods. Notably, night lights and employment emerge as the most significant factors influencing GDP growth from 2001 to 2021. Concurrently, the influence of traditional statistical indicators on GDP is progressively diminishing. Our prediction results for China's average annual GDP growth rate from 2022 to 2025 (with absolute error under 0.3%) show that China's economy will maintain stable and sustained growth. Furthermore, economic inequality, marked by inter-city disparities, narrowed significantly from 2000–2012 but has since widened slowly, with projections of further increases without targeted policies. An external shock module enhances predictions under disruptions. These findings indicate that China's economy is progressing toward balanced and sustainable development, yet targeted policy interventions remain essential to achieve further common prosperity.