Link Property Prediction On Ogbl Ppa
평가 지표
Ext. data
Number of params
Test Hits@100
Validation Hits@100
평가 결과
이 벤치마크에서 각 모델의 성능 결과
모델 이름 | Ext. data | Number of params | Test Hits@100 | Validation Hits@100 | Paper Title | Repository |
|---|---|---|---|---|---|---|
| Node2vec | No | 73878913 | 0.2226 ± 0.0083 | 0.2253 ± 0.0088 | node2vec: Scalable Feature Learning for Networks | |
| GraphGPT(SMTP) | No | 145263360 | 0.6876 ± 0.0067 | 0.7017 ± 0.0044 | GraphGPT: Graph Learning with Generative Pre-trained Transformers | |
| NGNN + GCN | No | 410113 | 0.3683 ± 0.0099 | 0.3834 ± 0.0082 | Network In Graph Neural Network | - |
| **Neural Common Neighbor ** | No | 33538 | 0.6119 ± 0.0085 | 0.6021 ± 0.0037 | Neural Common Neighbor with Completion for Link Prediction | |
| GraphSAGE | No | 424449 | 0.1655 ± 0.0240 | 0.1724 ± 0.0264 | Inductive Representation Learning on Large Graphs | |
| SEAL | No | 709122 | 0.4880 ± 0.0316 | 0.5125 ± 0.0252 | Labeling Trick: A Theory of Using Graph Neural Networks for Multi-Node Representation Learning | |
| Common Neighbor | No | 0 | 0.2765 ± 0.0000 | 0.2823 ± 0.0000 | - | - |
| AGDN | No | 36904259 | 0.4123 ± 0.0159 | 0.4332 ± 0.0092 | Adaptive Graph Diffusion Networks | |
| GCN (node embedding) | No | 148144898 | 0.6354 ± 0.0121 | 0.6524 ± 0.0096 | Can GNNs Learn Link Heuristics? A Concise Review and Evaluation of Link Prediction Methods | - |
| Resource Allocation | No | 0 | 0.4933 ± 0.0000 | 0.4722 ± 0.0000 | Predicting Missing Links via Local Information | |
| NGNN + SEAL | No | 735426 | 0.5971 ± 0.0245 | 0.5995 ± 0.0205 | Network In Graph Neural Network | - |
| S3GRL (PoS Plus) | No | 32270001 | 0.4242 ± 0.0180 | 0.6512 ± 0.0109 | Simplifying Subgraph Representation Learning for Scalable Link Prediction | |
| SIEG | No | 1993965 | 0.6322 ± 0.0174 | 0.6533 ± 0.0234 | - | - |
| MPLP | No | 147794531 | 0.6524 ± 0.0150 | 0.6685 ± 0.0073 | Pure Message Passing Can Estimate Common Neighbor for Link Prediction | |
| GCN | No | 278529 | 0.1867 ± 0.0132 | 0.1845 ± 0.0140 | Semi-Supervised Classification with Graph Convolutional Networks | |
| MLP+CN&RA&AA | No | 163330 | 0.5062 ± 0.0035 | 0.4906 ± 0.0029 | - | - |
| SUREL+ | No | 52802 | 0.5432 ± 0.0044 | 0.5492 ± 0.0112 | SUREL+: Moving from Walks to Sets for Scalable Subgraph-based Graph Representation Learning | |
| RA+Edge Proposal Set | No | 0 | 0.5324 ± 0.0000 | 0.5142 ± 0.0000 | Edge Proposal Sets for Link Prediction | |
| NGNN + GraphSAGE | No | 556033 | 0.4005 ± 0.0138 | 0.4058 ± 0.0123 | Network In Graph Neural Network | - |
| Matrix Factorization | No | 147662849 | 0.3229 ± 0.0094 | 0.3228 ± 0.0428 | Open Graph Benchmark: Datasets for Machine Learning on Graphs |
0 of 26 row(s) selected.