Table 2.

Performance of the species recognizer module on the ISN corpus

Configuration	Recognition			Normalization
	P	R	F	P	R	F
Training set
1. SR	0.968	0.869	0.916	0.968	0.862	0.912
2. NLP+SR	0.962	0.874	0.916	0.962	0.868	0.912
3. NLP+G/P+SR	0.982	0.958	0.970	0.982	0.951	0.966
4. NLP+G/P+ SR-Sentence	0.970	0.949	0.960	0.968	0.940	0.953
LINNAEUS	0.970	0.811	0.884	0.946	0.785	0.858
SPECIES	0.932	0.839	0.883	0.932	0.832	0.880
Test set
1. SR	0.963	0.820	0.886	0.957	0.814	0.880
2. NLP+SR	0.966	0.823	0.889	0.961	0.817	0.883
3. NLP+G/P+SR	0.965	0.929	0.947	0.960	0.920	0.940
4. NLP+G/P+SR-Sentence	0.966	0.917	0.941	0.952	0.900	0.925
LINNAEUS	0.951	0.764	0.847	0.918	0.734	0.816
SPECIES	0.921	0.8	0.856	0.919	0.795	0.852

The best PRF-scores are highlighted in bold. P, precision; R, recall; F, F-measure; NLP, natural language processing; G/P, gene/protein recognized module; SR, species recognizer module.