. 2021 Jul 27;185:115632. doi: 10.1016/j.eswa.2021.115632

Table 2.

Description of datasets.

Dataset	#Tweets	#Tweets extracted	Labels	Class distribution
Waseem and Hovy (2016)	16,000	14,949	Sexist, Racist, Non-Hateful	Hateful: 4839
Waseem and Hovy (2016)	16,000	14,949	Sexist, Racist, Non-Hateful	Non-Hateful: 10,110

SemEval 2019 (Basile et al., 2019)	9000	9000	Hateful, Non-Hateful	Hateful: 3783
SemEval 2019 (Basile et al., 2019)	9000	9000	Hateful, Non-Hateful	Non-Hateful: 5217

$C o v i d - H a t e_{H L}$ (Ziems et al., 2020)	2400	1637	Hate, Neutral	Hate: 677
$C o v i d - H a t e_{H L}$ (Ziems et al., 2020)	2400	1637	Hate, Neutral	Neutral: 960

$C o v i d - H a t e_{M L}$ (Ziems et al., 2020)	30M	10,674	Hate, Neutral	Hate: 4968
$C o v i d - H a t e_{M L}$ (Ziems et al., 2020)	30M	10,674	Hate, Neutral	Neutral: 5706

US elections	1105	1105	Hate, Neutral	Hate: 665
US elections	1105	1105	Hate, Neutral	Neutral: 440