| Infomap | Louvain | LGI-MCL ER | LGI-MCL RA | LGI-MCL EBC | MCL |
---|
karate | 0.54 | 0.49 | 0.72 | 0.73 | 0.72 | 0.74 |
opsahl 8 | 0.55 | 0.51 | 0.56 | 0.56 | 0.56 | 0.56 |
opsahl 9 | 0.49 | 0.42 | 0.38 | 0.39 | 0.39 | 0.41 |
opsahl 10 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
opsahl 11 | 0.96 | 0.96 | 0.90 | 0.82 | 0.79 | 0.63 |
polbooks | 0.50 | 0.49 | 0.57 | 0.57 | 0.57 | 0.57 |
football | 0.92 | 0.90 | 0.92 | 0.92 | 0.92 | 0.92 |
polblogs | 0.51 | 0.63 | 0.00 | 0.00 | 0.00 | 0.00 |
Mean NMI | 0.68 | 0.68 | 0.63 | 0.62 | 0.62 | 0.60 |
Mean ranking | 3.25 | 4.00 | 3.56 | 3.31 | 3.63 | 3.25 |
- For each real network, 100 perturbed networks have been generated removing at random the 10% of links. The table reports the Normalized Mutual Information (NMI) computed between the ground-truth communities and the ones detected by every community detection algorithm for the 8 real networks, averaged over the 100 repetitions. NMI = 1 indicates a perfect match between the two partitions of the nodes. The methods are ranked by mean NMI over the dataset. The best result for each network as well as the best mean results are marked in bold