Planted hitting set recovery in hypergraphs

Abstract In various application areas, networked data is collected by measuring interactions involving some specific set of core nodes. This results in a network dataset containing the nodes along with potentially much larger fringe that all have at least one interaction node. many settings, this type arises for structures are richer than graphs, because they involve sets; example, might be individuals under surveillance, where we observe attendees meetings individuals. We model such scenarios using hypergraphs , and study problem recovery: if hypergraph but not labels nodes, can recover ‘planted’ hypergraph? provide theoretical framework analyzing recovery use our theory to develop practical scalable algorithm recovery. The crux analysis hitting hypergraph, meaning every hyperedge has node demonstrate efficacy on number real-world datasets, outperforming competitive baselines derived from centrality core-periphery measures.