Table 14.

Evaluation on the challenge corpus. Highest F-measures are in bold. The F-measure differences from Stat De-id in PHI and in non-PHI are significant at α = 0.05.

Method	Class	Precision	Recall	F-measure
Stat De-id	PHI	98.69%	97.37%	98.03%
IFinder	PHI	25.10%	49.10%	33.20%*
H+D	PHI	36.24%	55.84%	43.95%*
CRFD	PHI	86.37%	84.79%	85.57%*
Stat De-id	Non-PHI	99.83%	99.92%	99.86%
IFinder	Non-PHI	97.25%	92.47%	94.80%*
H+D	Non-PHI	97.67%	94.95%	96.29%*
CRFD	Non-PHI	99.55%	99.65%	99.60%*