Social network is an important part of human society, containing a large number of user data, but also contain a large number of social relations. In social networks, graph convolution neural networks are often used to analyze and mine social networks. Aiming at this problem, this project studies the data of social network based on graph convolution neural network. This project takes social networks as the research object, based on graph, establishes graph modeling method, and uses graph convolution neural network to identify and recognize nodes in social network. The results show that the application of graph convolution neural network in social network is effective. The result shows that the algorithm based on image convolution network has 88.9% precision. Through the above research, we find that the big data of social network based on graph convolution neural network has a broad application prospect in social network. This project will explore the role of graph convolution neural network in social network, and develop more accurate and efficient algorithms and tools for social network research.