Table 3.

The evaluation results of the proposed model and other models.

Method	Dataset
	LegalQA		TrecQA		WikiQA
	MAP	MRR	MAP	MRR	MAP	MRR
KABLSTM [113]	0.751	0.790	0.792†	0.844†	0.732†	0.749†
EATS [75]	0.778	0.810	0.854†	0.881†	0.700†	0.715†
AM-BLSTM [114]	0.786	0.836	0.818	0.827	0.780	0.788
BERT-Base [115]	0.838	0.850	0.823	0.812	0.813†	0.828†
DRCN [116]	0.828	0.859	0.802	0.832	0.804†	0.862†
P-CNN [117]	0.715	0.729	0.680	0.698	0.734†	0.737†
DARCNN [118]	0.700	0.745	0.743	0.725	0.734†	0.750†
DASL [119]	0.804	0.816	0.824	0.831	0.768	0.795
IKAAS [120]	0.825†	0.883†	0.823†	0.868†	0.835	0.849
AS + random weight	0.758 ± 0.000	0.801 ± 0.001	0.796 ± 0.000	0.806 ± 0.002	0.771 ± 0.002	0.792 ± 0.009
AS-BIABC	0.788 ± 0.012	0.815 ± 0.008	0.802 ± 0.005	0.826 ± 0.002	0.803 ± 0.000	0.845 ± 0.025
RLAS	0.855 ± 0.102	0.872 ± 0.018	0.862 ± 0.014	0.883 ± 0.150	0.852 ± 0.025	0.876 ± 0.026
RLAS-BIABC	0.895 ± 0.020	0.912 ± 0.001	0.898 ± 0.015	0.906 ± 0.092	0.888 ± 0.036	0.891 ± 0.017

^† indicates that the results are taken from the articles.