Graph neural networks (GNNs) have achieved superior performance and gained significant interest in various domains. However, most of the existing GNNs are considered for homogeneous graphs, whereas real-world systems are usually modeled as heterogeneous graphs or heterogeneous information networks (HINs). |