Link Property Prediction On Ogbl Collab

评估指标

Ext. data
Number of params
Test Hits@50
Validation Hits@50

评测结果

各个模型在此基准测试上的表现结果

Paper TitleRepository
E2NNo5268510.9515 ± 0.14100.9546 ± 0.1270Edge2Node: Reducing Edge Prediction to Node Classification-
HyperFusionNo10644462120.7129 ± 0.00180.7385 ± 0.0099--
GIDN@YITUNo604490250.7096 ± 0.00550.9620 ± 0.0040GIDN: A Lightweight Graph Inception Diffusion Network for High-efficient Link Prediction-
PLNLP + SIGNNo349808640.7087 ± 0.00331.0000 ± 0.0000--
PLNLP (random walk aug.)No349808640.7059 ± 0.00291.0000 ± 0.0000Pairwise Learning for Neural Link Prediction
HOP-RECNo301911040.7012 ± 0.00161.0000 ± 0.0000--
PLNLP+ LRGANo352006560.6909 ± 0.00551.0000 ± 0.0000Global Attention Improves Graph Networks Generalization
PLNLP (val as input)No351121920.6872 ± 0.00521.0000 ± 0.0000Pairwise Learning for Neural Link Prediction
Refined-GAENo1266698250.6816 ± 0.00411.0000 ± 0.0000Reconsidering the Performance of GAE in Link Prediction
TopoLinkNo4833638450.6792 ± 0.00740.6771 ± 0.0083--
S3GRL (PoS Plus)No59130250.6683 ± 0.00300.9861 ± 0.0006Simplifying Subgraph Representation Learning for Scalable Link Prediction
ELPHNo32840650.6636 ± 0.58760.6631 ± 0.0021--
BUDDYNo11848670.6572 ± 0.00530.6621 ± 0.0016--
Adamic Adar+Edge Proposal SetNo00.6548 ± 0.00000.9735 ± 0.0000Edge Proposal Sets for Link Prediction
SEAL-nofeat (val as input)No5015700.6474 ± 0.00430.6495 ± 0.0043Labeling Trick: A Theory of Using Graph Neural Networks for Multi-Node Representation Learning
Adamic AdarNo00.6417 ± 0.00000.6349 ± 0.0000--
Common NeighborNo00.6137 ± 0.00000.6036 ± 0.0000--
SEAL-nofeatNo5015700.5471 ± 0.00490.6495 ± 0.0043Labeling Trick: A Theory of Using Graph Neural Networks for Multi-Node Representation Learning
GraphSAGE (val as input)No4602890.5463 ± 0.01120.5688 ± 0.0077Inductive Representation Learning on Large Graphs
NGNN + GraphSAGENo5918730.5359 ± 0.00560.6281 ± 0.0046Network In Graph Neural Network-
0 of 34 row(s) selected.
Link Property Prediction On Ogbl Collab | SOTA | HyperAI超神经