Table 3 Comparison with the existing approaches for GENIA data set
From: Biomedical named entity extraction: some issues of corpus compatibilities
System | Used approach | Domain knowledge/resources | FM |
|---|---|---|---|
Our system | Classifier ensemble | POS, phrase | 76.52 |
(CRF and SVM) | |||
Zhou & Su (2004) (GuoDong and Jian 2004) Final | HMM, SVM | Name alias, cascaded NEs dictionary, POS, phrase | 72.55 |
Zhou & Su (2004) (GuoDong and Jian 2004) | HMM, SVM | POS, phrase | 64.1 |
Kim et al. (2005) (Kim et al. 2005) | Two-phase model | POS, phrase, | 71.19 |
with ME and CRF | rule-based component | ||
Finkel et. al (2004) (Finkel et al. 2004) | CRF | Gazetteers, web-querying, surrounding abstracts, | 70.06 |
abbreviation handling, BNC corpus, POS | |||
Settles (2004) (Settles 2004) | ME | POS, semantic knowledge sources of 17 lexicons | 70.00 |
Saha et al. (2009) (Saha et al. 2009) | ME | POS, phrase | 67.41 |
Park et. al (2004) (Park et al. 2004) | ME | POS, phrase, domain-salient words using WSJ, | 66.91 |
morphological patterns, collocations from Medline | |||
Song et al. (2004) (Song et al. 2004b) Final | SVM, CRF | POS, phrase, Virtual sample | 66.28 |
Song et al. (2004) (Song et al. 2004b) Base | SVM | POS, phrase | 63.85 |
Ponomareva et al. (2007) (Ponomareva et al. 2007) | HMM | POS | 65.7 |