Mining the Specific Skill Demographics in Social Network

GitHub is a popular social coding platform with millions of users and repositories. The social interactions between these users and repositories form a complex network that can provide valuable insights into the programming community. In this paper, we propose a graph convolutional network (GCN) based approach to analyze the GitHub social network and extract meaningful information from it. We use GCN to learn the graph structure and represent the network as a set of high-dimensional node embeddings. We used GCN to distinguish between ML developers and web developers, and by running it we can get a very clear graph to distinguish them, and of course since the two groups have a lot in common, we can see a lot of partial overlap. Our experiments show that our GCN-based approach outperforms traditional methods in predicting user attributes on the GitHub social network. Furthermore, we demonstrate the interpretability of our model by visualizing the learned embeddings and showing their clustering patterns. Overall, our work highlights the potential of GCN-based methods for data mining on large-scale social networks like GitHub.