Figure 5From: Learning Statistical Models for Annotating Proteins with Function Information using Biomedical TextROC and precision-recall graphs for predictions made without weakly-labeled data. ROC and precision-recall graphs for GO code predictions made by the system aggregated over the three ontologies, when Informative Term models were learned without using any weakly labeled data. The solid line represents predictions made by the Informative Term/Naïve Bayes path in Figure 1. The dashed line represents predictions made by the Regular Expression/Naïve Bayes path in Figure 1.Back to article page