Moara: a Java library for extracting and normalizing gene and protein mentions

BMC Bioinformatics

Table 1 Results for the CBR-Tagger evaluated with the BioCreative 2 GM test set

Training set	Recall	Precision	F-Measure
CbrBC2	64.11	76.01	69.56
CbrBC2y	42.90	80.98	56.08
CbrBC2m	29.14	76.08	42.14
CbrBC2f	51.05	73.66	60.30
CbrBC2ymf	24.53	77.00	37.21
Best BioCreative	85.97	88.48	87.21
BANNER	82.78	87.18	84.92
ABNER	51.49	86.93	64.68

The BioCreative 2 Gene Mention test set consists of 5,000 sentences. The first five numerical lines represent the results (recall, precision and F-measure) according to the corpus used for training the CBR-Tagger: BioCreative 2 Gene Mention task only (CbrBC2) or combined with the BioCreative 1 task 1B for yeast (CbrBC2y), mouse (CbrBC2m), fly (CbrBC2f) or all three (CbrBC2ymf). The last two lines present the best results of the BioCreative 2 Gene Mention task and BANNER and ABNER results when trained with the latter training corpus.

ISSN: 1471-2105