Figure 2From: Finding genomic ontology terms in text using evidence contentGO evaluation of FiGO predictions in tasks 2.1 and 2.2. A prediction was considered perfect when it provided high evidence of both GO term and protein. Thus, a prediction could provide high evidence of the GO term without being perfect. These charts disregard the protein evaluation and show the number of predictions submitted by FiGO to BioCreative tasks 2.1 and 2.2, which provided a high, generally and low evidence of the GO term for each value of α used.Back to article page