Skip to Main Content
Table 4: 

The percentage increase in error of selector-predictor highlight methods compared to an equivalent architecture model which was trained to classify complete text. We report the numbers reported in previous work whenever possible (italics means our results). Architectures are not necessarily consistent across the table, thus they do not imply performance superiority of any method. The highlight lengths chosen for each experiment were chosen with precedence whenever possible, and otherwise chosen as 20% following Jain et al. (2020) precedence.

SST-2SST-3SST-5AG NewsIMDBEv. Inf.MultiRCMoviesBeer
Lei et al. (2016) 22.65 7.09 9.85 33.33 22.23 36.59 31.43 160.0 37.93 
Bastings et al. (2019) 3.31 0 2.97 199.02 12.63 85.19 75.0  13.64 
FRESH 90.0 17.82 13.45 50.0 14.66 9.76 0.0 20.0  
SST-2SST-3SST-5AG NewsIMDBEv. Inf.MultiRCMoviesBeer
Lei et al. (2016) 22.65 7.09 9.85 33.33 22.23 36.59 31.43 160.0 37.93 
Bastings et al. (2019) 3.31 0 2.97 199.02 12.63 85.19 75.0  13.64 
FRESH 90.0 17.82 13.45 50.0 14.66 9.76 0.0 20.0  
Close Modal

or Create an Account

Close Modal
Close Modal