Table 9.
Lemmatization performance of the BioLemmatizer resources on CRAFT set
| Silver Standard | |||
|---|---|---|---|
| Recall | Precision | F-score | |
| Base (MorphAdorner lexicon) | 94.37% (5532/5862) | 94.16% (5532/5875) | 94.26% |
| Base + GENIA | 94.20% (5522/5862) | 93.90% (5522/5881) | 94.05% |
| Base + BioLexicon | 98.41% (5769/5862) | 98.23% (5769/5873) | 98.32% |
| Entire Lexicon | 98.60% (5780/5862) | 98.42% (5780/5873) | 98.51% |
| Rule Only | 97.83% (5735/5862) | 97.83% (5735/5862) | 97.83% |
| Rule + Lexicon Validation | 98.67% (5784/5862) | 98.67% (5784/5862) | 98.67% |
| Gold Standard | |||
| Recall | Precision | F-score | |
| Base (MorphAdorner lexicon) | 53.71% (311/579) | 53.34% (311/583) | 53.52% |
| Base + GENIA | 62.69% (363/579) | 61.95% (363/586) | 62.32% |
| Base + BioLexicon | 64.77% (375/579) | 64.10% (375/585) | 64.43% |
| Entire Lexicon | 76.68% (444/579) | 75.90% (444/585) | 76.29% |
| Rule Only | 85.84% (497/579) | 85.84% (497/579) | 85.84% |
| Rule + Lexicon Validation | 90.85% (526/579) | 90.85% (526/579) | 90.85% |