A k-MCST Based Algorithm for Discovering Core-Periphery Structures in Graphs

Core-periphery structures are examples of meso-scale characteristics of graphs. Most existing algorithms for core-periphery (CP) structures work by first finding the dense cores of a network and then discovering the peripheral nodes around them. Our algorithm presented here seeks to query a graph to return the CP structures centered around any selected query node. Our algorithm significantly reduces the computational complexity of repeatedly querying the CP structures from a network. Our algorithm repeatedly extracts minimum cost spanning trees (MCSTs), first from the original network, and then successively from the residual networks. From the union of these MCSTs, our algorithm efficiently answers the queries for CP structures around nodes. We validate our algorithm on example networks taken from two domains.