Table 2 Details of the three BioNLP 2013 ST datasets
From: Comparing neural models for nested and overlapping biomedical event detection
Dataset | Item | Train | Dev. | Test |
|---|---|---|---|---|
CG 2013 | Documents | 300 | 100 | 200 |
| Â | Sentences | 2640 | 850 | 1610 |
| Â | Pct unknown words | 0% | 10.63% | 10.68% |
| Â | Â Events | 9422 | 3217 | 5530 |
| Â | Â Flat events | 45.31% | 44.07% | NA |
| Â | Â Nested events | 34.95% | 36.46% | NA |
| Â | Â Overlapping events | 41.05% | 43.05% | NA |
| Â | Â Inter-sentence events | 4.08% | 3.11% | NA |
PC 2013 | Documents | 260 | 90 | 175 |
| Â | Sentences | 1900 | 660 | 1254 |
| Â | Â Pct unknown words | 0% | 11.66% | 11.50% |
| Â | Events | 6,657 | 2320 | 4004 |
| Â | Â Flat events | 33.28% | 34.74% | NA |
| Â | Â Nested events | 38.90% | 38.87% | NA |
| Â | Â Overlapping events | 54.88% | 52.80% | NA |
| Â | Â Inter-sentence events | 4.70% | 2.41% | NA |
GE 2013 | Documents | 10 | 10 | 14 |
| Â | Sentences | 1051 | 1104 | 1188 |
| Â | Â Pct unknown words | 0 % | 16.84% | 17.07% |
| Â | Events | 2882 | 3259 | 3301 |
| Â | Â Flat events | 53.57% | 42.25% | NA |
| Â | Â Nested events | 31.15% | 38.96% | NA |
| Â | Â Overlapping events | 26.57% | 35.80% | NA |
| Â | Â Inter-sentence events | 10.55% | 22.55% | NA |