Node Clustering

Node clustering aims to partition the nodes of a network into groups based on their connectivity and attributes, seeking to maximize similarity within and minimize similarity between clusters. Recent research emphasizes handling diverse network types (graphs, hypergraphs, multiplex graphs), addressing challenges posed by heterophily (dissimilar nodes connecting frequently), and improving scalability for massive datasets. Prominent approaches leverage graph neural networks (GNNs), variational autoencoders (VGAEs), spectral clustering, and novel algorithms incorporating concepts like contrastive learning, modularity optimization, and curvature-based methods. These advancements enhance the accuracy and efficiency of node clustering, with applications spanning social network analysis, bioinformatics, and recommendation systems.