Disentangled Hyperbolic Representation Learning for Heterogeneous Graphs

Abstract--Heterogeneous graphs have attracted a lot of research interests recently due to the success for representing complex real-world systems. However, existing methods have two pain points in embedding them into low-dimensional spaces: the mixing of structural and semantic information, and the distributional mismatch between data and embedding spaces. These two challenges require representation methods to consider the global and partial data distributions while unmixing the information. On the other hand, with the rapid development of graph neural networks, researchers begin to consider using different aggregating functions or processes I. RAPH data, which can be abstracted from a lot of real-world systems (e.g. Driven by the requirements of these However, all these methods fall into a paradigm that learns realistic scenarios and the characteristics of ease for calculation, a mixed representation for each node, ignoring the distinguishing recently graph representation learning has attracted great influences from different aspects of characteristics, attention as a general operation for graph data analysis, and e.g. Take Figure 1 as an example, the has achieved outstanding performances on diverse downstream homogeneous graph and heterogeneous graph have the same tasks, ranging from node clustering [5], [6], node classification topological structure, and keep the same structural information [7], link prediction [8] to community detection [9].