Community discovery algorithm of complex network attention model

In terms of understanding the structure of complex networks or the functional characteristics of complex networks, community discovery is of great significance. This paper uses the attention model to combine the second-order neighbor similarity matrix with the modularity matrix, extracts relatively more comprehensive complex network feature information from multiple angles for network division. Firstly, perform complex network preprocessing, and perform division preprocessing according to the value of the attention similarity matrix. Secondly, complete the merger of the community game according to the connection strength between the two different communities. By comparing with other algorithms on computer-generated networks and real-world networks, it is proved that this algorithm has obtained good results in terms of the number of communities, running time, normalized mutual information and modularity.