Skip to main content

Table 12 Pathogen characterisation results. Pathogen identification (PI) is filtered with either a SVM or a BERT classifier that identifies non-relevant documents (NRDs) or non-relevant pathogens / pathogen focus (PF)

From: Classifying literature mentions of biological pathogens as experimentally studied using natural language processing

Method

Precision

Recall

F1

Pathogen identification

0.5632

0.8305

0.6712

PI + NRD filtering SVM

0.6184

0.7966

0.6962

PI + NRD filtering BERT

0.6104

0.7966

0.6912

PI + PF filtering SVM

0.5581

0.8136

0.6621

PI + PF filtering BERT

0.6104

0.7966

0.6912

PI + NRDs SVM + PF BERT

0.6716

0.7627

0.7143