Uncovering High-Order Cohesive Structures: Efficient (k,g)-Core Computation and Decomposition in Hypergraphs

Abstract

Hypergraphs offer a versatile framework for analysing complex networks with higher-order interactions. This paper introduces the (k,g)-core model for cohesive subgraph discovery in hypergraphs, extend ing the traditional k-core model by incorporating co-occurrence constraints. The (k,g)-core identifies subgraphs in which each node has at least k neighbours co-occurring in at least g hyperedges, effec tively capturing both connectivity and interaction strength. To compute these structures efficiently, we propose a top-down, memory-efficient algorithm. Extensive experiments on real-world hypergraphs demonstrate the effectiveness of the (k,g)-core model and the computational efficiency of the proposed algorithm.