DSGNN, a dual-view supergrid-aware graph neural network, can effectively model the spatial dependencies of both adjacent and distant grid regions to provide accurate regional air quality estimation.
