Fig. 2
From: LocText: relation extraction of protein localizations to assist database curation
LocText full pipeline (NER + RE); intrinsic evaluation. Using the STRING Tagger-extracted (“predicted”) entities, both LocText and Baseline had low and comparable F-measure (F=57% ± 4 and F=51% ± 3, resp.), however LocText was optimized for precision (P=86%)