Table 3.
Quantitative Recall-Oriented Understudy for Gisting Evaluation (ROUGE)-1, ROUGE-2, and ROUGE-L results on the summarization task of clinical diagnostic interviews. All the scores have a 95% CI of at most ±0.18 (SD).
| Methods | ROUGEa-1 | ROUGE-2 | ROUGE-L | |||||
|
|
Recall | F1 score | Recall | F1 score | Recall | F1 score | ||
| KiASb | 22.53 | 30.43 | 10.65 | 14.62 | 24.46 | 32.57 | ||
| Abstractive summarization | 9.89 | 15.22 | 4.09 | 6.42 | 12.80 | 20.53 | ||
| AoESc | 8.69 | 12.37 | 2.25 | 3.21 | 11.51 | 17.24 | ||
| Extractive summarization | 9.79 | 14.37 | 4.18 | 6.52 | 12.72 | 20.47 | ||
aROUGE: Recall-Oriented Understudy for Gisting Evaluation.
bKiAS: knowledge-infused abstractive summarization.
cAoES: abstraction over extractive summarization.