Context-sensitive graph representation learning

Graph representation learning, which maps high-dimensional graphs or sparse into a low-dimensional vector space, has shown its superiority in numerous learning tasks. Recently, researchers have identified some advantages of context-sensitive graph methods functions such as link predictions and ranking recommendations. However, most existing depend on convolutional neural networks recursive to obtain additional information outside node, require community algorithms extract multiple contexts focus only the local neighboring nodes without their structural information. In this paper, we propose novel method, Context-Sensitive Representation Learning (CSGRL), simultaneously combines attention variant auto-encoder learn weighty about various aspects participating nodes. The core CSGRL is utilize an asymmetric encoder aggregate structures optimize goal. main benefit that it does not need features for node. message spread through encoder. Experiments are conducted three real datasets both tasks prediction node clustering, results demonstrate can significantly improve effectiveness all challenging compared with 14 state-of-the-art baselines.