OGB-DDI Link ogbl-ddi Leaderboard
Link prediction on a drug-drug interaction network (4k drugs); predict interactions absent from training. · Metric: Hits@20 (higher is better) · 🟢 Updated 2h ago
| # | Model | Hits@20 | Paper |
|---|---|---|---|
| 1 | HyperFusion | 1.00 | — |
| 2 | ELGNN | 0.98 | — |
| 3 | GCN (node embedding) | 0.95 | — |
| 4 | GIDN@YITU | 0.95 | — |
| 5 | AGDN (AUC loss) | 0.95 | — |
| 6 | Refined-GAE | 0.94 | — |
| 7 | PSG | 0.93 | — |
| 8 | PLNLP | 0.91 | — |
| 9 | GDNN | 0.90 | — |
| 10 | GraphSAGE + Edge Attr | 0.88 | — |
| 11 | CFLP (w/ JKNet) | 0.86 | — |
| 12 | GraphSAGE+anchor distance | 0.82 | — |
| 13 | NeuralCommonNeighbor | 0.82 | — |
| 14 | ELPH | 0.77 | — |
| 15 | DEA + JKNet | 0.77 | — |
| 16 | BUDDY | 0.77 | — |
| 17 | GraphSAGE+Edge Proposal Set | 0.75 | — |
| 18 | LRGA+GCN(Node2Vec+Augment) | 0.74 | — |
| 19 | MAD Learning | 0.68 | — |
| 20 | LRGA + GCN | 0.62 | — |
| 21 | GCN+JKNet | 0.61 | — |
| 22 | NGNN + GraphSAGE | 0.58 | — |
| 23 | NGNN + GCN | 0.55 | — |
| 24 | GraphSAGE | 0.54 | — |
| 25 | GCN | 0.37 | — |
| 26 | SEAL | 0.31 | — |
| 27 | DeepWalk | 0.26 | — |
| 28 | Node2vec | 0.23 | — |