TABLE 4. Additional Evaluation Measures of the Methods on the Litcovid and Hoc datsets.

	Label-based measures								Instance-based measures
	Macro-Precision		Macro-Recall		Micro-Precision		Micro-Recall		Precision		Recall
	Mean	Max	Mean	Max	Mean	Max	Mean	Max	Mean	Max	Mean	Max
LitCovid BioCreative
ML-Net	0.8364	0.8559	0.7309	0.7632	0.8756	0.8827	0.8142	0.8227	0.8849	0.8901	0.8514	0.8591
Binary BERT	0.9103	0.9498	0.8350	0.8690	0.9304	0.9448	0.8969	0.9173	0.9349	0.9408	0.9210	0.9381
Linear BERT	0.9071	0.9388	0.8354	0.8701	0.9276	0.9396	0.8870	0.9093	0.9368	0.9443	0.9143	0.9308
LitMC-BERT (ours)	0.9131	0.9226	0.8574	0.8814	0.9313	0.9366	0.8952	0.9145	0.9418	0.9480	0.9212	0.9355
Hoc
ML-Net	0.7949	0.8356	0.7389	0.7622	0.7667	0.8053	0.7253	0.7448	0.8045	0.8296	0.7826	0.8013
Binary BERT	0.8471	0.8644	0.8661	0.8834	0.8363	0.8562	0.8548	0.8724	0.8565	0.8735	0.8909	0.9095
Linear BERT	0.8772	0.8930	0.8475	0.8630	0.8614	0.8812	0.8496	0.8619	0.8929	0.9116	0.8955	0.9024
LitMC-BERT (ours)	0.8868	0.8983	0.8641	0.8910	0.8718	0.8938	0.8582	0.8787	0.9035	0.9148	0.9038	0.9193