Table 6.

Evaluation of BioLexicon Coverage on JNLPBA-2004 dataset

	BLTagger	BT-Tagger
Full	55.54	47.96
Left	56.72	55.72
Right	59.24	55.63

The table compares the performance of two dictionary-based POS taggers, the BLTagger and the BT-Tagger, in recognising protein names in the JNLPBA-2004 test dataset. For each tagger, F-scores are reported in terms of full matches (i.e., the sequence of words identified as a biomedical noun by the tagger matches exactly the text span annotated as a protein in the JNLPBA-2004 training dataset) as well as right and left boundary matches (i.e., the sequence of words identified as a biomedical noun by the tagger matches either the left or right boundary of the corresponding annotation in the JNLPBA-2004 training data).