From: Recognizing speculative language in biomedical research articles: a linguistically motivated perspective
Threshold
Precision
Recall
Accuracy
F1 score
1
0.66
0.89
0.86
0.76
1.5
0.81
0.79
0.90
0.80
2
0.83
0.69
0.75
3
0.88
0.53
0.87
5
0.98
0.25
0.40
6
0.13
0.24