Table 3.

ALC scores, F-measures at the end of 120-minute annotation, and the statistical test P values of AL and PL. Best performance across models for a user is in boldface

Users	ALC scores		F-measures at 120 minutes		P values based on Wilcoxon signed-rank test
	PL	AL	PL	AL
User1	0.633	0.637	0.696	0.695	9.7x10-2
User2	0.574	0.575	0.659	0.671	7.2x10-3
User3	0.608	0.628	0.683	0.690	3.0x10-5
User4	0.615	0.609	0.692	0.680	5.6x10-3
User5	0.619	0.642	0.707	0.717	1.8x10-5
User6	0.580	0.610	0.674	0.691	3.9x10-4
User7	0.521	0.580	0.624	0.671	1.8x10-5
User8	0.599	0.613	0.673	0.691	2.7x10-5
User9	0.606	0.629	0.683	0.693	1.8x10-5
Mean	0.595	0.632	0.677	0.689

Abbreviations: AL, active learning; ALC, under the learning curve; PL, passive learning.