Skip to main content

View full-text article in PMC

. 2021 Sep 17;23(9):e30161. doi: 10.2196/30161

Table 3.

Performance of our models in round 5 of the TREC-COVID challenge.

Model	NDCG@20^a	P@20^b	Bpref^c	MAP^d	# rel^e
bm25	0.6320	0.6440	0.5021	0.2707	6533
bow + rrf	0.6475	0.6650	0.5174	0.2778	6695
mlm + rrf	0.7716	0.7880	0.5680	0.3468	6963
bow + mlm + rrf	0.7826	0.8050	0.5616	0.3719	7006
bow + mlm + lm	0.7297	0.7460	0.5759	0.3068	6834
bow + mlm + lr	0.7375	0.7450	0.5719	0.3439	6976
bow + mlm + lr + rrf	0.7961	0.8260	0.5659	0.3789	6939

^aNDCG@20: normalized discounted cumulative gain at 20 documents.

^bP@20: precision at 20 documents.

^cBpref: binary preference.

^dMAP: mean average precision.

^e# ref: total number of relevant documents retrieved by the model for the 50 queries.