The HaSpeeDe2 dataset collects 8,012 tweets and 500 news headlines annotated for the presence of hate speech, stereotypes and nominal utterance.
The dataset has been used in the context of the HaSpeeDe task (
http://www.di.unito.it/~tutreeb/haspeede-evalita20/index.html), organized as part of the EVALITA 2020 evaluation campaign (
http://www.evalita.it/2020).
In order to meet the GDPR requirements, texts have been pseudonymized replacing all original IDs in both datasets… See the full description on the dataset page:
https://huggingface.co/datasets/evalitahf/hatespeech_detection.
