Skip to main content
Figure 6 | Journal of Biomedical Semantics

Figure 6

From: Evaluating gold standard corpora against gene/protein tagging solutions and lexical resources

Figure 6

The precision and recall performances in these diagrams are based on cos98 evaluation. The left and the right diagram give an overview on the precision and recall measurements of the PGN taggers against the selected GSCs. For ML-Tag methods both parameters are best, if the taggers have been tested against those corpora that they have been trained on. The LexTag methods have been sorted from left to right according to their precision performance with the best performing solution to the left. With minor deviations, all LexTag methods show best performances against the FsuPrge corpus for both, precision and recall. The “trade-off” between precision and recall for methods with similar F1-measures is well known, but it is still remarkable that the LexTag solutions can differ considerably on their precision and recall values, whereas the F1-measure seems to be almost invariable across them. Again, precision and recall performances of the LexTag solutions against the corpora reach higher values for FsuPrge and PennBio than for BioCreative-II and Jnlpba.

Back to article page