Table 3. Comparing performance with benchmark models.

Model	Weighted-AUCROC	Weighted-AUCPR	Weighted-F1	MCC	ACC
HTCInfoMax	0.9861	0.9060	0.9030	0.8989	0.9087
ConvTextTM	0.9806	0.8658	0.8210	0.8393	0.8478
BERT+HiMatch	0.9932	0.9668	0.9172	0.9078	0.9174
HDLTex	0.9865	0.9063	0.9030	0.8989	0.9087
DocBERT	0.9880	0.9433	0.9173	0.9077	0.9174
Ours	0.9922	0.9639	0.9350	0.9273	0.9348

Note:

Bold indicates the highest performance.