“…Instead of actually running the spreading process, these metrics are mostly based on the local or global topology of a node in the network, for instance, number of immediate neighbors [28,29,59,157], global position [17,158,159,160,161,162], number of shortest paths [163,164,165,166], random walks [167,168,169], eigenvectors [170,171,172,173], path counting [174,175,176,177], etc. Even though the optimal metric that performs best for all spreading dynamics on all underlying networks does not seem to exist [178,179,180], these centrality-based approaches are still persistently used due to their simplicity and relative satisfactory performance in some occasions.…”