An effective heuristic clustering algorithm for mining multiple critical nodes in complex networks

Abstract Influence maximization is of great significance in complex networks, and many methods have been proposed to solve it. However, they are usually time-consuming or cannot deal with the overlap of spreading. To get over the flaws, an effective heuristic clustering algorithm is proposed in this paper: (1) nodes that have been assigned to clusters are excluded from the network structure to guarantee they do not participate in subsequent clustering. (2) the K-shell ( k s ) and Neighborhood Coreness ( N C ) value of nodes in the remaining network are recalculated, which ensures the node influence can be adjusted during the clustering process. (3) a hub node and a routing node are selected for each cluster to jointly determine the initial spreader, which balances the local and global influence. Due to the above contributions, the proposed method preferably guarantees the influence of initial spreaders and the dispersity between them. A series of experiments based on Susceptible–Infected–Recovered (SIR) stochastic model confirm that the proposed method has favorable performance under different initial constraints against known methods, including VoteRank, HC, GCC, HGD, and DLS-AHC.