The graph embedded topic model

Dingge Liang, Marco Corneli, Charles Bouveyron, Pierre Latouche

Abstract: Highlights•A document similarity-based graph convolutional network (DS-GCN).•An end-to-end node clustering approach with estimates of the cluster posterior probabilities.•A set of experiments on simulated data to illustrate the proposed methodologies.•In-depth analysis of a real-world network.