Figure 11From: Evaluating gold standard corpora against gene/protein tagging solutions and lexical resourcesThe diagram displays the FP results of the tagging solutions in form of error frequencies normalized against the annotation numbers in the GSC. As expected, GP7 produces lot of FP errors in all corpora which have been reduced in Wh-Ukpmc (GP7) through FP filtering using Chang2. All solutions produce larger numbers of FP errors in Jnlpba and PennBio leading to the conclusion that both corpora are less compliant with the current terminological standards in the tagging of PGNs.Back to article page