Directed Graph Convolutional Network
This work addresses the problem of processing directed graph-structured data for applications like citation networks and co-purchase analysis, representing an incremental improvement over existing GCNs.
The paper tackled the limitation of Graph Convolutional Networks (GCNs) to undirected graphs by extending spectral-based convolution to directed graphs using first- and second-order proximity, resulting in a new model called DGCN that achieved superior performance against state-of-the-art methods on citation networks and co-purchase datasets.
Graph Convolutional Networks (GCNs) have been widely used due to their outstanding performance in processing graph-structured data. However, the undirected graphs limit their application scope. In this paper, we extend spectral-based graph convolution to directed graphs by using first- and second-order proximity, which can not only retain the connection properties of the directed graph, but also expand the receptive field of the convolution operation. A new GCN model, called DGCN, is then designed to learn representations on the directed graph, leveraging both the first- and second-order proximity information. We empirically show the fact that GCNs working only with DGCNs can encode more useful information from graph and help achieve better performance when generalized to other models. Moreover, extensive experiments on citation networks and co-purchase datasets demonstrate the superiority of our model against the state-of-the-art methods.