Skip to main content

Table 3 Results of the ROUGE scale for the two models applied to the Arabic dataset, AHS

From: Arabic text summarization using deep learning approach

Model Inference penalty ROUGE-1 average precision ROUGE-1 average recall ROUGE-1 average F-measure
Baseline model No penalty 57.02 21.05 30.67
Coverage 55.93 19.60 28.95
Length 55.58 21.05 30.46
Length + coverage 55.16 19.60 28.85
Baseline model + copy mechanism No penalty 62.01 33.53 43.42
Coverage 60.43 32.12 41.84
Length 62.13 34.46 44.23
Length + coverage 61.36 32.59 42.47
  1. All our ROUGE scores have a 95% confidence interval of at most \(\pm \;0.25\)