ALYT: Abusive Language on YouTube

Paper link

This repository contains the data from the paper Abusive Language on Social Media Through the Legal Looking Glass, including:

• the Abusive Language on YouTube dataset;
• the data statement related to the dataset;

Data Format

ALYT_data.csv has 19,915 rows, one per YouTube comment, with these columns:

Column	Description
comment_id	YouTube comment identifier
comment	comment text (English)
label	annotators' score on the 7-point scale described in the paper (1 to 7)
label_normalised	collapsed label used in the paper: scores 1-3 become -1 (non-abusive), 4 becomes 0, 5-7 become 1 (abusive)
video_id	YouTube video identifier
video_topic	topic of the source video: transgender, veganism or workplace
video_type	type of the source video: official, personal or reaction

The annotation framework, guidelines and training process are described in the paper and in DATA STATEMENT.md.

Citing

If you use the dataset in your research, you are kindly asked to acknowledge it in any publications.

@inproceedings{bertaglia-etal-2021-abusive,
    title = "Abusive Language on Social Media Through the Legal Looking Glass",
    author = "Bertaglia, Thales  and
      Grigoriu, Andreea  and
      Dumontier, Michel  and
      van Dijck, Gijs",
    booktitle = "Proceedings of the 5th Workshop on Online Abuse and Harms (WOAH 2021)",
    month = aug,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.woah-1.20",
    doi = "10.18653/v1/2021.woah-1.20",
    pages = "191--200"}