Social network community discovery method based on graph neural network

The invention discloses a social network community discovery method based on a graph neural network, and the method comprises the steps: constructing a graph model of a social network according to data provided by the social network; randomly sampling neighbor nodes of each node to serve as aggregation feature nodes; when k-th aggregation is carried out on the node v, firstly, aggregation operation is carried out on neighbors u in a k-order subgraph of the node v; aggregating the embedding information obtained by the (k-1) th aggregation of the node v and the aggregation information of the neighbor node of the node v in a CONCAT mode; updating and embedding of the node v are obtained; forming a probability matrix of the category to which each node belongs; after training is completed, label information is output, the nodes with the same labels are divided into a community, and a community division result is obtained. According to the community division method, the neighbor feature information of the nodes and