Table 4. Chinese datasets. Ten-fold cross-validation was used to split datasets.

Datasets	Positive(Pos)	Negitive(Neg)
Hotela	5,322	2,444
WaiMaib	4,000	6,000
SinaWeiboc	5,993	5,993
DouBand	6,000	6,000

Notes.

^a https://raw.githubusercontent.com/SophonPlus/ChineseNlpCorpus/master/datasets/ChnSentiCorp_htl_all/ChnSentiCorp_htl_all.csv.

^b https://raw.githubusercontent.com/SophonPlus/ChineseNlpCorpus/master/datasets/waimai_10k/waimai_10k.csv.

^c https://github.com/SophonPlus/ChineseNlpCorpus/blob/master/datasets/weibo_senti_100k.

^d https://www.kaggle.com/utmhikari/doubanmovieshortcomments.