Hyper-relational Inductive Link Prediction on JFFI V2
0.2777MRR (H/T)THOR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| THORInductive setting=Fully-inductive2026.02 | 0.2777 | 0.2309 | |
| TRIXInductive setting=Fully-inductive2026.02 | 0.2445 | — | |
| MetaNIRInductive setting=Fully-inductive2026.02 | 0.2376 | 0.1768 | |
| ULTRAInductive setting=Fully-inductive2026.02 | 0.2105 | — | |
| MAYPLInductive setting=Fully-inductive2026.02 | 0.1859 | 0.1633 | |
| MOTIFInductive setting=Fully-inductive2026.02 | 0.1792 | — | |
| KG-ICLInductive setting=Fully-inductive2026.02 | 0.1098 | — | |
| RMPIInductive setting=Fully-inductive2026.02 | 0.0881 | — | |
| INGRAMInductive setting=Fully-inductive2026.02 | 0.04 | — | |
| Neural LPInductive setting=Rule-based2026.02 | 0.0281 | — | |
| DRUMInductive setting=Rule-based2026.02 | 0.0249 | — | |
| NS-HARTInductive setting=Semi-inductive2026.02 | 0.0129 | 0.0165 | |
| NBFNetInductive setting=Semi-inductive2026.02 | 0.0081 | — | |
| QBLP (w/o pretrain)Inductive setting=Semi-inductive, Pre-training status=Without pre-training2026.02 | 0.0016 | — |