Table 2.

Performance of negation scope detection, value normalization, and temporal normalization modules in C2Q 1.0 and C2Q 2.0 with 95% confidence intervals using COVID-19 trials

Evaluation task	Metric	C2Q 1.0	C2Q 2.0
Negation scope detection	Accuracy	0.776 [0.751, 0.798]	0.924 [0.907, 0.937]
	Precision	0.792 [0.758, 0.823]	0.963 [0.945, 0.977]
	Recall	0.759 [0.724, 0.791]	0.884 [0.857, 0.908]
	F1-score	0.775 [0.748, 0.800]	0.922 [0.905, 0.937]
Value normalization	Accuracy	0.601 [0.576, 0.624]	0.966 [0.955, 0.973]
Temporal normalization	Accuracy	0.554 [0.522, 0.584]	0.916 [0.896, 0.931]