Table 2 Comparison with the existing approaches for GENETAG data set
From: Biomedical named entity extraction: some issues of corpus compatibilities
System | Approach used | Domain knowledge/resources | F-measure |
|---|---|---|---|
Our system | GA based ensemble | PoS, phrase | 94.70 |
(CRF and SVM) | |||
Song et al. (2005) (Song et al. 2004a) | SVM | - | 66.7 |
Bickel et al. (2004) (Bickel et al. 2004) | SVM | a dictionary | 72.1 |
Kinoshita et al. (2005) (Kinoshita et al. 2005) | TnT (Brants 2000), the Trigrams 'n' Tags | dictionary based postprocessing | 80.9 |
HMM-based part-of-speech tagger | |||
Mitsumoriet al. (2005) (Mitsumori et al. 2005) | SVM | gene/protein name dictionary | 78.09 |
Finkel et al. (2004) (Finkel et al. 2005) | ME+ post processing | 82.2 | |
McDonald and Pereira (2005) (McDonald and Pereira 2005) | CRF | 82.4 | |
GuoDong et al. (2005) (Zhou and Su 2002) | HMM, SVM, Ensemble technique | Post processing | 82.58 |