Deep Learning Based Abstractive Text Summarization: Approaches, Datasets, Evaluation Measures, and Challenges

Table 6

Evaluation measures of several abstractive text summarisation methods over the CNN/Daily Mail datasets.

Reference	Year	Authors	Model	ROUGE1	ROUGE2	ROUGE-L
[55]	2016	Nallapati et al.	Words-lvt2k-temp-att	35.46	13.30	32.65
[56]	2017	See et al.	Pointer-generator + coverage	39.53	17.28	36.38
[57]	2017	Paulus et al.	Reinforcement learning, with intra-attention	41.16	15.75	39.08
[57]	2017	Paulus et al.	Maximum-likelihood + RL, with intra-attention	39.87	15.82	36.90
[58]	2018	Liu et al.	Adversarial network	39.92	17.65	36.71
[30]	2018	Song et al.	ATSDL	34.9	17.8	—
[35]	2018	Al-Sabahi et al.	Bidirectional attentional encoder-decoder	42.6	18.8	38.5
[59]	2018	Li et al.	Key information guide network	38.95	17.12	35.68
[60]	2018	Kryściński et al.	ML + RL ROUGE + Novel, with LM	40.19	17.38	37.52
[61]	2018	Yao et al.	DEATS	40.85	18.08	37.13
[62]	2018	Wan et al.	BiSum	37.01	15.95	33.66
[63]	2019	Wang et al.	BEAR (large + WordPiece)	41.95	20.26	39.49
[64]	2019	Egonmwan et al.	TRANS-ext + filter + abs	41.89	18.9	38.92
[65]	2020	Liu et al.	BERTSUMEXT (large)	43.85	20.34	39.90
[49]	2020	Peng et al.	DAPT + imp-coverage (RL + MLE (ss))	40.72	18.28	37.35