Skip to main content
Fig. 4 | BMC Bioinformatics

Fig. 4

From: pyMeSHSim: an integrative python package for biomedical named entity recognition, normalization, and comparison of MeSH terms

Fig. 4

Recall, Precision and F1 of pyMeSHSim, DNorm and TaggerOne. a-d. Performance of pyMeSHSim without SCRs (a), pyMeSHSim with SCRs (b), DNorm (c) and TaggerOne (d). The similarity between MeSH terms identified by the tools and Nelson’s manual work were called as a true positive or false positive when their similarity was higher or lower than the determined threshold. When the similarity threshold is set to 1, only perfect matched terms would be considered as true positives. The recall (\( \frac{TP}{TP+ FN} \)), precision (\( \frac{TP}{TP+ FP} \)) and F1 (\( \frac{2\times precision\times recall}{precision+ recall} \)) of the tools were calculated at each similarity threshold

Back to article page