Skip to Main Content
Table 6: 
Human evaluation results for summaries produced by TextRank and DetRank; proportion of times AMT workers found models Informative (Inf), Succinct (Succ), and Coherent (Coh); All is the average across ratings; symbol † denotes that differences between models are statistically significant (p < 0.05) using a pairwise t-test.
MethodInfSuccCohAll
TextRank 45.45 51.11 42.50 † 46.35 † 
DetRank 54.55 48.89 57.50 53.65 
MethodInfSuccCohAll
TextRank 45.45 51.11 42.50 † 46.35 † 
DetRank 54.55 48.89 57.50 53.65 
Close Modal

or Create an Account

Close Modal
Close Modal