Table 5 Evaluation of CID results
From: Improving chemical disease relation extraction with rich features and weakly labeled data
Team/training corpus | Using text-mined entity mentions | Using gold entity mentions | ||||
|---|---|---|---|---|---|---|
Precision | Recall | F-score | Precision | Recall | F-score | |
Co-occurrence baseline | 16.43 | 76.45 | 27.05 | Â | Â | Â |
Avg team results | 47.09 | 42.61 | 43.37 | – | – | – |
Best team results | 55.67 | 58.44 | 57.03 | – | – | – |
1. Train | 51.55 | 59.19 | 55.11 | 62.07 | 64.17 | 63.10 |
2. Train + dev | 64.24 | 52.06 | 57.51 | 68.15 | 66.04 | 67.08 |
3. Train + dev + 1000 | 63.78 | 53.85 | 58.39 | 68.12 | 68.95 | 68.53 |
4. Train + dev + 5000 | 62.50 | 56.75 | 59.49 | 67.63 | 72.33 | 69.90 |
5. Train + dev + 10,000 | 64.49 | 56.57 | 60.27 | 69.64 | 71.86 | 70.73 |
6. Train + dev + 18,410 | 65.59 | 56.94 | 61.01 | 71.07 | 72.61 | 71.83 |