notesum.ai
Published at November 12Retrieval, Reasoning, Re-ranking: A Context-Enriched Framework for Knowledge Graph Completion
cs.AI
cs.CL
Released Date: November 12, 2024
Authors: Muzhi Li1, Cehao Yang2, Chengjin Xu1, Xuhui Jiang1, Yiyan Qi1, Jian Guo1, Ho-fung Leung3, Irwin King3
Aff.: 1IDEA Research, International Digital Economy Academy; 2Artificial Intelligence Thrust, Hong Kong University of Science and Technology (Guangzhou); 3Department of Computer Science & Engineering, The Chinese University of Hong Kong
| Methods | FB15k237 | WN18RR | ||||||
| MRR | Hits@1 | Hits@3 | Hits@10 | MRR | Hits@1 | Hits@3 | Hits@10 | |
| KG embedding methods | ||||||||
| TransE | 0.279 | 0.198 | 0.376 | 0.441 | 0.243 | 0.043 | 0.441 | 0.532 |
| ComplEx | 0.247 | 0.158 | 0.275 | 0.428 | 0.440 | 0.410 | 0.460 | 0.510 |
| RotatE | 0.338 | 0.241 | 0.375 | 0.533 | 0.476 | 0.428 | 0.492 | 0.571 |
| TuckER | 0.358 | 0.266 | 0.394 | 0.544 | 0.470 | 0.443 | 0.482 | 0.526 |
| GIE | 0.362 | 0.271 | 0.401 | 0.552 | 0.491 | 0.452 | 0.505 | 0.575 |
| HittER | 0.373 | 0.279 | 0.409 | 0.558 | 0.503 | 0.462 | 0.516 | 0.584 |
| Graph neural network-based methods | ||||||||
| CompGCN | 0.355 | 0.264 | 0.390 | 0.535 | 0.479 | 0.443 | 0.494 | 0.546 |
| NBF-Net | 0.415 | 0.321 | 0.450 | 0.599 | 0.551 | 0.497 | 0.573 | 0.666 |
| Text-based methods | ||||||||
| KG-BERT | - | - | - | 0.420 | 0.216 | 0.041 | 0.302 | 0.524 |
| MEM-KGC | 0.346 | 0.253 | 0.381 | 0.531 | 0.557 | 0.475 | 0.604 | 0.704 |
| SimKGC | 0.338 | 0.252 | 0.364 | 0.511 | 0.671 | 0.595 | 0.719 | 0.802 |
| CoLE | 0.389 | 0.294 | 0.429 | 0.572 | 0.593 | 0.538 | 0.616 | 0.701 |
| Large language model-based methods | ||||||||
| KICGPT | 0.412 | 0.327 | 0.448 | 0.581 | 0.564 | 0.478 | 0.612 | 0.677 |
| DIFT + TransE | 0.232 | 0.08 | 0.321 | 0.502 | 0.491 | 0.462 | 0.496 | 0.560 |
| DIFT + SimKGC | 0.402 | 0.338 | 0.418 | 0.528 | 0.686 | 0.616 | 0.730 | 0.806 |
| DIFT + CoLE | 0.439 | 0.364 | 0.468 | 0.586 | 0.617 | 0.569 | 0.638 | 0.708 |
| + TransE | 0.456 | 0.414 | 0.474 | 0.550 | 0.506 | 0.487 | 0.515 | 0.556 |
| + RotatE | 0.456 | 0.400 | 0.476 | 0.569 | 0.520 | 0.495 | 0.520 | 0.550 |
| + GIE | 0.463 | 0.400 | 0.485 | 0.581 | 0.558 | 0.520 | 0.580 | 0.615 |
| + SimKGC | 0.471 | 0.429 | 0.490 | 0.557 | 0.717 | 0.656 | 0.759 | 0.809 |
| + CoLE | 0.507 | 0.455 | 0.537 | 0.612 | 0.635 | 0.579 | 0.676 | 0.723 |
| + NBF-Net | 0.536 | 0.477 | 0.565 | 0.636 | 0.641 | 0.605 | 0.662 | 0.695 |