Fig. 2

Approximation of non-spectral centrality measures with various sampling algorithms for the real datasets Epinions(upper left), Slashdot(upper right), Stanford(center left), Adolescent Health(center right), C. Elegans(bottom left) and US Power grid(bottom right). Scatterplots represent average results over 10 runs of sampling, with standard deviation indicated by vertical bars. SpringRank is missing for undirected networks