Table 3 Performance comparison of the different pruning approaches and the baseline methods (TEES/EVEX) on the official BioNLP Shared Task GE data sets
| Â | Predictions | Precision | Recall | F1-score |
|---|---|---|---|---|
TEES-2011 (Shared Task 2011) | Original TEES | 61.76 | 48.78 | 54.51 |
| Â | Pruned-TEES (Unsupervised Method) | 62.39 | 48.75 | 54.74 |
| Â | Pruned-TEES (Manual Annotation Method) | 62.04 | 48.78 | 54.62 |
| Â | Pruned-TEES (Aggregation Method) | 62.26 | 48.78 | 54.70 |
| Â | Pruned-TEES (Aggregation Method + SVM) | 62.27 | 48.78 | 54.71 |
TEES-2013 (Shared Task 2013) | Original TEES | 56.32 | 46.17 | 50.74 |
| Â | Pruned-TEES (Unsupervised Method) | 57.13 | 46.02 | 50.97 |
| Â | Pruned-TEES (Manual Annotation Method) | 56.63 | 46.17 | 50.87 |
| Â | Pruned-TEES (Aggregation Method) | 56.97 | 46.17 | 51.00 |
| Â | Pruned-TEES (Aggregation Method + SVM) | 57.01 | 46.17 | 51.02 |
EVEX-2013 (Shared Task 2013) | Original EVEX | 58.03 | 45.44 | 50.97 |
| Â | Pruned-EVEX (Unsupervised Method) | 58.77 | 45.29 | 51.15 |
| Â | Pruned-EVEX (Manual Annotation Method) | 58.32 | 45.44 | 51.08 |
| Â | Pruned-EVEX (Aggregation Method) | 58.66 | 45.44 | 51.21 |
| Â | Pruned-EVEX (Aggregation Method + SVM) | 58.71 | 45.44 | 51.23 |