Table 1.
Token-level | Field-level | |||||
precis. | recall | F1 | precis. | recall | F1 | |
development set | ||||||
likely-protein | 0.816 | 0.313 | 0.452 | 0.667 | 0.268 | 0.382 |
possible-protein | 0.439 | 0.885 | 0.587 | 0.304 | 0.566 | 0.396 |
dictionary | 0.501 | 0.469 | 0.484 | 0.245 | 0.439 | 0.314 |
YAPEX test | ||||||
likely-protein | 0.949 | 0.648 | 0.770 | 0.872 | 0.621 | 0.725 |
possible-protein | 0.490 | 0.974 | 0.652 | 0.472 | 0.825 | 0.600 |
YAPEX system | 0.678 | 0.664 | 0.671 |
Performance of the NER systems on the mouse development corpus and the YAPEX test corpus. The row labeled "dictionary" is for simple dictionary lookup using the mouse synonym list. The row labeled "YAPEX system" is from [15].