Detecting communities with multiple topics in attributed networks via self-supervised adaptive graph convolutional network

Chaobo He, Junwei Cheng, Guohua Chen, Quanlong Guan, Xiang Fei, Yong Tang

Abstract: Highlights•An end-to-end method SSAGCN for multiple topics community detection in attributed networks is proposed.•SSAGCN is implemented by autoencoder-style self-supervised learning.•SSAGCN can adaptively fuse topology information and attribute information.•SSAGCN has a good capability to describe the communities using multiple topics.•SSAGCN outperforms state-of-the-art baselines.