Table 3.
Development test results (F-scores)
| Method | Single-class | Multiclass |
|---|---|---|
| Baseline | 86.93 | 81.23 |
| Truecasing | 87.12 | 81.45 |
| Non-local features | 87.81 | 81.82 |
| UMLS, tokens | 89.07 | 82.79 |
| UMLS, longest phrases | 88.57 | 82.64 |
| UMLS, all phrases | 89.65 | 83.50 |
| OBO, tokens | 87.58 | 81.76 |
| OBO, longest phrases | 88.81 | 82.56 |
| OBO, all phrases | 88.40 | 82.44 |
| Brown, news, c = 100 | 87.20 | 80.92 |
| Brown, news, c = 320 | 87.71 | 81.23 |
| Brown, news, c = 1000 | 86.58 | 80.68 |
| Brown, news, c = 3200 | 87.11 | 80.80 |
| Brown, bio, c = 100 | 87.44 | 81.67 |
| Brown, bio, c = 320 | 89.56 | 82.03 |
| Brown, bio, c = 1000 | 88.94 | 81.78 |
| Brown, bio, c = 3200 | 88.55 | 81.33 |