Table 8. Performance of the multi-label misinformation classification models trained and evaluated using ArCOV19-MLM.

Boldface represents the best performance scores for each metric.

Classifier	Binary Relevance (BR)			Label Powerset(LP)
	Jaccard	Macro-F1	Micro-F1	Jaccard	Macro-F1	Micro-F1
RF	39.35%	48.61%	55.25%	62.65%	55.4%	62.54%
Linear SVC	60.29%	67.84%	75.23%	76.3%	68.89%	76.69%