Vietnamese Sentiment Analysis under Limited Training Data Based on Deep Neural Networks

Table 4

The size of new samples for each polarity when applying the data augmentation techniques.

No.	Data augmentation techniques	Size
1	Vietnamese EDA	4,000
2	Sentence shuffling	1,000
3	Back translation	100
4	Syntax-tree transformation	100
5	Contextual substitution ()	1,000
6	Contextual substitution ( + )	1,000
7	Masked language model	200