Skip to Main Content
Table 6: 

Accuracy of supervised models. p ¡ {0.05, 0.001} with paired bootstrap compared to BERT.

Normative ArgumentsNon-normative Arguments
ACCAUCF1F1supF1attF1neuACCAUCF1F1supF1attF1neu
TGA Net 71.5 88.3 62.2 43.5 54.3 88.7 76.6 90.8 69.8 62.9 53.9 92.5 
Hybrid Net 66.8 78.2 56.2 42.9 42.4 83.4 71.8 82.2 65.7 55.6 51.4 90.2 
 
BERT 79.5 92.4 73.3 60.5 65.2 94.2 83.8 94.6 79.2 72.3 68.8 96.6 
BERT+LX 79.2 92.1 72.7 58.7 65.6 93.8 83.7 94.6 79.2 70.8 69.9 96.9 
BERT+MT 79.3 92.6 73.4 63.8 63.6 92.7 83.6 94.7 79.2 71.8 69.7 96.1 
 
LogBERT 80.0 92.8 74.3 63.6 66.2 93.2 84.3 95.0 80.2 73.1 71.4 96.1 
(a) Kialo 
Normative ArgumentsNon-normative Arguments
ACCAUCF1F1supF1attF1neuACCAUCF1F1supF1attF1neu
TGA Net 71.5 88.3 62.2 43.5 54.3 88.7 76.6 90.8 69.8 62.9 53.9 92.5 
Hybrid Net 66.8 78.2 56.2 42.9 42.4 83.4 71.8 82.2 65.7 55.6 51.4 90.2 
 
BERT 79.5 92.4 73.3 60.5 65.2 94.2 83.8 94.6 79.2 72.3 68.8 96.6 
BERT+LX 79.2 92.1 72.7 58.7 65.6 93.8 83.7 94.6 79.2 70.8 69.9 96.9 
BERT+MT 79.3 92.6 73.4 63.8 63.6 92.7 83.6 94.7 79.2 71.8 69.7 96.1 
 
LogBERT 80.0 92.8 74.3 63.6 66.2 93.2 84.3 95.0 80.2 73.1 71.4 96.1 
(a) Kialo 
Close Modal

or Create an Account

Close Modal
Close Modal