Table 8.

Did authors reuse an existing human-labeled data set (external), create a new human-labeled data set (original), or both?

CountProportion
Only external 58 56.31%
Only original 29 28.16%
Original and external 8.74%
Unsure/no information (but can assume original) 6.80%
Subtotal: assumed to include some original human labeling 45 43.69%
Total of applicable papers (assumed to use human-labeled training data) 103 100.00%
Nonapplicable papers 97 –
CountProportion
Only external 58 56.31%
Only original 29 28.16%
Original and external 8.74%
Unsure/no information (but can assume original) 6.80%
Subtotal: assumed to include some original human labeling 45 43.69%
Total of applicable papers (assumed to use human-labeled training data) 103 100.00%
Nonapplicable papers 97 –
Close Modal