Multi-Label Graph Convolutional Network Representation Learning

Knowledge representation of networked systems is fundamental in many disciplines. To date, existing methods for learning primarily focus on networks with simplex labels, yet real-world objects (nodes) are inherently complex nature and often contain rich semantics or labels. For example, a user may belong to diverse interest groups social network, resulting multi-label applications. A network not only has multiple labels each node, the highly correlated making ineffective even fail handle such correlation node learning. In this article, we propose novel graph convolutional (MuLGCN) representation. fully explore label-label topology structures, model as two Siamese GCNs: node-node-label label-label-node graph. The GCNs one aspect nodes respectively, seamlessly integrated objective function. learned label representations can effectively preserve intra-label interaction properties, aggregated enhance under unified training framework. Experiments comparisons classification validate effectiveness our proposed approach.