Table 1 Performance results of LDMGNN compared with other methods on two datasets, we report the mean and standard deviation of the test sets
Methods | SHS27k | SHS148k | ||||
|---|---|---|---|---|---|---|
Random | BFS | DFS | Random | BFS | DFS | |
SVM [33] | \(75.35 \pm 1.05\) | \(42.98 \pm 6.15\) | \(53.07 \pm 5.16\) | \(80.55 \pm 0.23\) | \(49.14 \pm 5.30\) | \(58.59 \pm 0.07\) |
RF [34] | \(78.45 \pm 0.08\) | \(37.67 \pm 1.57\) | \(35.55 \pm 2.22\) | \(82.10 \pm 0.20\) | \(38.96 \pm 1.94\) | \(43.26 \pm 3.43\) |
LR [35] | \(71.55 \pm 0.93\) | \(43.06 \pm 5.05\) | \(48.51 \pm 1.87\) | \(67.00 \pm 0.07\) | \(47.45 \pm 1.42\) | \(51.09 \pm 2.09\) |
HIN2Vec [37] | \(74.22 \pm 2.38\) | \(49.61 \pm 4.88\) | \(53.78 \pm 3.05\) | \(78.01 \pm 0.62\) | \(56.94 \pm 3.20\) | \(57.15 \pm 2.49\) |
SDNE [38] | \(84.04 \pm 0.91\) | \(47.29 \pm 4.32\) | \(53.42 \pm 2.82\) | \(86.65 \pm 2.73\) | \(58.43 \pm 4.94\) | \(68.84 \pm 1.52\) |
LPI-DLDN [39] | \(77.36 \pm 0.48\) | \(44.68 \pm 2.31\) | \(54.98 \pm 3.94\) | \(83.83 \pm 0.52\) | \(56.41 \pm 5.38\) | \(60.07 \pm 2.71\) |
LPI-deepGBDT [40] | \(72.70 \pm 0.67\) | \(42.25 \pm 3.81\) | \(50.48 \pm 2.76\) | \(81.69 \pm 0.39\) | \(55.51 \pm 7.40\) | \(59.67 \pm 3.29\) |
DTI-CDF [41] | \(79.29 \pm 0.89\) | \(49.60 \pm 5.28\) | \(55.88 \pm 4.19\) | \(83.12 \pm 0.55\) | \(60.04 \pm 8.27\) | \(65.42 \pm 5.89\) |
PIPR [7] | \(83.31 \pm 0.75\) | \(44.48 \pm 4.44\) | \(57.80 \pm 3.24\) | \(90.05 \pm 2.59\) | \(61.83 \pm 10.23\) | \(63.98 \pm 0.76\) |
GAT [42] | \(86.35 \pm 0.86\) | \(53.08 \pm 5.24\) | \(60.09 \pm 1.69\) | \(88.87 \pm 0.31\) | \(62.10 \pm 7.75\) | \(65.49 \pm 0.50\) |
GNN-PPI [26] | \(87.91 \pm 0.39\) | \(63.81 \pm 1.79\) | \(74.72 \pm 5.26\) | \(92.26 \pm 0.10\) | \(71.37 \pm 5.33\) | \(82.67 \pm 0.85\) |
LDMGNN | 89.34 \(\pm \; 0.44\) | 74.56 \(\pm \;3.03\) | 78.20 \(\pm \; 2.69\) | 92.38 \(\pm \;0.08\) | 73.98 \(\pm \;5.51\) | 83.79 \(\pm \; 0.95\) |