Table 3 Per class metrics for BioLinkBERT fine-tuned model
From: Leveraging pre-trained language models for mining microbiome-disease relationships
Class | Precision | Recall | F1-score |
|---|---|---|---|
NA | \(0.556 \pm 0.131\) | \(0.447 \pm 0.186\) | \(0.457 \pm 0.096\) |
Negative | \(0.827\pm 0.022\) | \(0.894\pm 0.043\) | \(0.859\pm 0.025\) |
Positive | \(0.831\pm 0.015\) | \(0.854\pm 0.048\) | \(0.841\pm 0.027\) |
Relate | \(0.829\pm 0.040\) | \(0.798\pm 0.053\) | \(0.811\pm 0.025\) |