GripNet: Graph information propagation on supergraph for heterogeneous graphs

• A novel supergraph data structure to segregate a heterogeneous graph into interconnected, semantically-coherent subgraphs for efficient learning. new representation learning framework based on graphs and graph-based integration. Extensive experiments seven large-scale datasets link prediction node classification tasks. Heterogeneous aims learn low-dimensional vector representations of different types entities relations empower downstream Existing popular methods either capture semantic relationships but indirectly leverage node/edge attributes in complex way, or directly without taking account. When involving multiple convolution operations, they also have poor scalability. To overcome these limitations, this paper proposes ible Gr aph i nformation p ropagation Net work (GripNet) framework. Specifically, we introduce consisting supervertices superedges. supervertex is subgraph. superedge defines an information propagation path between two supervertices. GripNet learns the interest by propagating along defined using layers. We construct evaluate against competing show its superiority prediction, classification, The code available at https://github.com/nyxflower/GripNet .