Table 1 Datasets statistics
From: An approach to automatic classification of hate speech in sports domain on social media
No | Purpose | Source | Number of comments | non-HS labels | HS labels | Labels |
|---|---|---|---|---|---|---|
1 | Training | YouTube—entertainment channels | 109,676 after refining 47,884 | 38,789 | 9,095 | Automatically labelled by HS lexicon |
2 | Testing | YouTube—entertainment channels | 5,317 after refining 5,200 | 1,542 | 3,658 | Manually labelled |
3 | Testing | YouTube—sports channels | 270 | 11 | 259 | Manually labelled |
4 | Training | News portals blic.rs and b92.net—sports news | 65,155 | 56,316 | 8,839 | Automatically labelled by HS lexicon |
5 | Testing | News portals blic.rs and b92.net—sports news | 367 | 229 | 138 | Manually labelled |