Link Prediction Methods and Their Accuracy for Different Social Networks and Network Metrics
Table 6
Prediction methods accuracy result (AUC).
Datasets
AUC
CN
JI
PA
AA
Cosin
Sor
HPI
HDI
LHN
Facebook
0.6688
0.6758
0.6803
0.6753
0.8369
0.6738
0.6715
0.6708
0.6694
0.6694
Flickr
0.89
0.8702
0.841
0.8922
0.8839
0.8812
0.865
0.844
0.8511
0.6944
UC Irvine
0.6625
0.6421
0.8412
0.6738
0.8048
0.6414
0.6359
0.6303
0.6427
0.6322
PWr
0.6815
0.6466
0.7924
0.6913
0.7979
0.651
0.6514
0.6422
0.6491
0.6382
Enron
0.8157
0.7937
0.9015
0.8196
0.9312
0.7921
0.7995
0.794
0.7977
0.7881
YouTube
0.8525
0.7957
0.9109
0.8571
0.9157
0.7938
0.7503
0.8017
0.7984
0.7587
Average
0.7618
0.7374
0.8279
0.7682
0.8617
0.7389
0.7289
0.7305
0.7374
0.6968
Variance
0.0105
0.0091
0.0071
0.0099
0.0032
0.0095
0.0084
0.0087
0.0083
0.0041
The accuracy of selected prediction methods measured by AUC. The average performance and the variance for each method are also listed. aIn our experiment, we choose = 0.0005.