Figure 6

Largest component experiments for unweighted-undirected networks. In (a), after removing approximately 40% of the nodes ranked by our algorithm, the largest component size is reduced to the lowest, while the other metrics need to remove approximately 80% or 90% of the nodes. Panel (b)clearly shows that the performance of URank compared with other metrics. In (c), the figure shows the performance of our metric is more uniform, and there is little difference in the removal ratio of nodes when each metric makes the largest component size reach the lowest. In panel (d), when removing the first 40% of the nodes, our metric makes the largest component size decline the fastest until it reaches the minimum.