. 2019 Oct 29;2019:baz085. doi: 10.1093/database/baz085

Table 7.

Performance results of the three largest single-annotator sets for the PMRA, BM25 and TF-IDF methods. Metrics include MCC, AUC (ROC), MRR and P@N, given as (mean ± stdev).

		#	MCC	AUC	MRR	P@5	P@10	P@20
A1	BM25	`92`	`0.610 ± 0.25`	`0.771 ± 0.19`	`0.740 ± 0.34`	`0.335 ± 0.19`	`0.226 ± 0.13`	`0.150 ± 0.09`
	PMRA		`0.580 ± 0.26`	`0.728 ± 0.23`	`0.726 ± 0.35`	`0.313 ± 0.19`	`0.212 ± 0.11`	`0.149 ± 0.08`
	TFIDF		`0.601 ± 0.27`	`0.752 ± 0.21`	`0.735 ± 0.35`	`0.337 ± 0.23`	`0.220 ± 0.13`	`0.155 ± 0.09`
A2	BM25	`55`	`0.538 ± 0.23`	`0.710 ± 0.19`	`0.882 ± 0.24`	`0.611 ± 0.29`	`0.498 ± 0.29`	`0.408 ± 0.27`
	PMRA		`0.547 ± 0.17`	`0.699 ± 0.15`	`0.892 ± 0.24`	`0.553 ± 0.28`	`0.420 ± 0.24`	`0.342 ± 0.22`
	TFIDF		`0.564 ± 0.20`	`0.729 ± 0.15`	`0.900 ± 0.25`	`0.615 ± 0.30`	`0.480 ± 0.28`	`0.377 ± 0.26`
A3	BM25	`49`	`0.550 ± 0.25`	`0.733 ± 0.18`	`0.966 ± 0.14`	`0.780 ± 0.22`	`0.698 ± 0.26`	`0.627 ± 0.28`
	PMRA		`0.561 ± 0.25`	`0.732 ± 0.19`	`0.944 ± 0.17`	`0.788 ± 0.24`	`0.718 ± 0.24`	`0.614 ± 0.28`
	TFIDF		`0.504 ± 0.25`	`0.697 ± 0.20`	`0.922 ± 0.20`	`0.710 ± 0.26`	`0.659 ± 0.29`	`0.605 ± 0.31`

The bold values indicate maximal metric value between the three baseline methods.