Abstractive Arabic Text Summarization Based on Deep Learning

Table 5

Performance comparisons of the proposed system by using AraBERT preprocess and without using it.

Dataset	Preprocess	ROUGE-1			ROUGE-2			ROUGE-L			BLEU
		Precision	Recall	F1	Precision	Recall	F1	Precision	Recall	F1
AHS	With AraBERT	54.95	50.48	51.49	13.1	12.01	12.27	37.84	35.19	34.37	0.41
	Without AraBERT	52.44	49.5	49.79	12.23	11.5	11.6	36.12	34.66	33.22	0.39
AMN	With AraBERT	48.15	42.65	44.28	19.46	17.93	18.35	35.48	32.86	32.46	0.41
	Without AraBERT	45.9	38.3	40.7	17.63	15.53	16.18	33.19	29.21	29.15	0.36

The highest result is given in bold.