Typical Hypergraph Computation Tasks

Abstract After hypergraph structure generation for the data, next step is how to conduct data analysis on hypergraph. In this chapter, we introduce four typical computation tasks, including label propagation, clustering, imbalance learning, and link prediction. The first task which predict labels vertices, i.e., assigning a each unlabeled vertex in hypergraph, based labeled information. general cases, propagation propagate information from vertices through structural of hyperedges. part, discuss cut hypergraphs random walk interpretation hypergraphs. second formulated as dividing into several parts Laplacian smoothing filter an embedded model clustering tasks. third cost-sensitive targets learning with different mis-classification costs. fourth prediction, aims discover missing relations or new coming hyperedges observed