A community detection algorithm for multi-view attributed

Multi-view attributed networks community detection is much more challenging than common simple or attributed networks community detection, because each view may contain noisy edges and connections and also some key information are often shared among views. We propose a community detection method called the MvCGCD algorithm for multi-view attributed network in this paper. The proposed algorithm firstly removes the unwanted high frequency noise by using a graph filter, then it employs a graph convolution network (GCN) to aggregate the local neighborhood information of each node to obtain the new representation of the node. Experimental results on multi real work datasets show that MvCGCD algorithm out performed the baseline algorithms, the best performance case was up to 25.39%.