Relation Extraction on SciER
62.36Hierarchical Relation F1HGNet
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| HGNetLearning Paradigm=Supervised2026.03 | 62.36 | 67.02 | 65.38 | |
| HGNetLearning Paradigm=Supervised, Loss Configuration=w/o LDHL2026.03 | 59.1 | 65.95 | 62.79 | |
| HGNetLearning Paradigm=Supervised, Loss Configuration=w/o LCAF Loss2026.03 | 54.75 | 61.21 | 58.67 | |
| GCNLearning Paradigm=Supervised2026.03 | 47.37 | 63.89 | 57.35 | |
| GATLearning Paradigm=Supervised2026.03 | 47.35 | 64.29 | 57.64 | |
| GATLearning Paradigm=Supervised, Loss Configuration=w/o LDHL2026.03 | 47.03 | 64.23 | 57.3 | |
| GCNLearning Paradigm=Supervised, Loss Configuration=w/o LDHL2026.03 | 46.85 | 64.22 | 56.89 | |
| HGERELearning Paradigm=Supervised2026.03 | 43.79 | 64.35 | 58.47 | |
| PL-MarkerLearning Paradigm=Supervised2026.03 | 40.25 | 61.84 | 56.78 | |
| PURELearning Paradigm=Supervised2026.03 | 38.53 | 56.21 | 49.35 | |
| openai/gpt-oss-120bLearning Paradigm=Zero-Shot2026.03 | 27.93 | 27.52 | 27.64 | |
| qwen/qwen3-32bLearning Paradigm=Zero-Shot2026.03 | 24.02 | 24.45 | 24.28 | |
| llama-3.3-70b-versatileLearning Paradigm=Zero-Shot2026.03 | 23.97 | 25.06 | 24.59 | |
| llama-3.1-8b-instantLearning Paradigm=Zero-Shot2026.03 | 17.15 | 17.69 | 17.43 | |
| GPT-3.5 TurboLearning Paradigm=Zero-Shot2026.03 | 8.35 | 8.91 | 8.58 |