ETHOS (multi-labEl haTe speecH detectiOn dataSet)
Introduced by Mollas et al. in ETHOS: an Online Hate Speech Detection Dataset
ETHOS is a hate speech detection dataset. It is built from YouTube and Reddit comments validated through a crowdsourcing platform. It has two subsets, one for binary classification and the other for multi-label classification. The former contains 998 comments, while the latter contains fine-grained hate-speech annotations for 433 comments.
Source: https://arxiv.org/abs/2006.08328Benchmarks
| Trend | Task | Dataset Variant | Best Model | Paper | Code |
|---|---|---|---|---|---|
|
Ethos Binary
|
BiLSTM + static BE
|
|||
|
Ethos MultiLabel
|
MLARAM
|
Papers
| Paper | Code | Results | Date | Stars |
|---|
