Table 3.

Accuracy of the suggested responses for different models

Method	Precision@1 (%)	Precision@3 (%)	Precision@5 (%)	MRR
BERT‡	59.32 ± 1.64	85.42 ± 0.82	87.56 ± 0.63	0.79 ± 0.00
BiLSTM†	58.98 ± 0.88	83.28 ± 0.75	85.37 ± 0.62	0.75 ± 0.00
Seq2Seq†	53.21 ± 0.61	61.48 ± 4.46	68.63 ± 5.54	0.60 ± 0.02
XGBoost†*	51.33 ± 0.31	79.59 ± 0.52	82.83 ± 0.42	0.69 ± 0.01
SVM†*	47.62 ± 0.42	78.97 ± 0.49	82.25 ± 0.51	0.68 ± 0.00
Weighted TF-IDF	34.41 ± 2.57	46.46 ± 2.97	53.53 ± 1.2	0.42 ± 0.03
TF-IDF	32.32 ± 2.11	44.35 ± 1.28	51.01 ± 0.68	0.42 ± 0.03
Frequency	16.71 ± 0.25	32.20 ± 0.67	42.80 ± 0.74	0.35 ± 0.00

Bold entries are the best performance values given each metric

^‡PubMedBERT embedding; ^†Wikipedia-PubMed embedding; ^*TF-IDF