Hugging Face's logo

Datasets:
manueltonneau
/
french-hate-speech-superset

Tasks:
Text Classification
Languages:
French
Size Categories:
10K<n<100K
Tags:
hate speech
Dataset card Files Community
1

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

French Hate Speech Superset

This dataset is a superset (N=18,071) resulting from the preprocessing and merge of all available French hate speech datasets in April 2024. These datasets were identified through a systematic survey of hate speech datasets conducted in early 2024. We only kept datasets that:

  • are documented
  • are publicly available
  • focus on hate speech, defined broadly as "any kind of communication in speech, writing or behavior, that attacks or uses pejorative or discriminatory language with reference to a person or a group on the basis of who they are, in other words, based on their religion, ethnicity, nationality, race, color, descent, gender or other identity factor" (UN, 2019)

The survey is further detailed in our the survey paper.

Columns

The dataset contains six columns:

  • text: the annotated post
  • labels: annotation of whether the post is hateful (== 1) or not (==0). As datasets have different annotation schemes, we systematically binarized the labels.
  • target: target of hate, in case it is provided in the raw datasets
  • source: origin of the data (e.g., Twitter)
  • dataset: dataset the data is from (see "Datasets" part below)
  • nb_annotators: number of annotators by post
  • tweet_id: tweet ID, in case the data originates from Twitter and the ID was made available.

Datasets

The datasets that compose this superset are:

  • CONAN - COunter NArratives through Nichesourcing: a Multilingual Dataset of Responses to Fight Online Hate Speech (CONAN in the dataset column)
  • Multilingual and Multi-Aspect Hate Speech Analysis (MLMA in the dataset column)
  • An Annotated Corpus for Sexism Detection in French Tweets (sexism)
  • CyberAgressionAdo-v1: a Dataset of Annotated Online Aggressions in French Collected through a Role-playing Game (cyberado)
  • Detection of Racist Language in French Tweets (FTR)

Preprocessing

We drop duplicates. In case of non-binary labels, the labels are binarized (hate speech or not). We replace all usernames and links by fixed tokens to maximize user privacy. Further details on preprocessing can be found in the preprocessing code here.

Citation

Please cite our survey paper if you use this dataset.

TODO
Downloads last month
0
Edit dataset card