a novel Romanian language dataset for offensive message detection with manually annotated comment from a local Romanian news website (stiri de cluj) into five classes:
- non-offensive
- targeted insults
- racist
- homophobic
- sexist
Resulting in 4052 annotated messages
Class | Label | # examples | Percentage |
---|---|---|---|
Non-offensive | 0 | 2682 | 66.19% |
Targeted insult | 1 | 777 | 19.18% |
Racist | 2 | 252 | 6.22% |
Homophobic | 3 | 186 | 4.59% |
Sexist | 4 | 155 | 3.82% |
TOTAL | 4052 |
Csv containing all annotated comments (Fields comment_text and LABEL)
additionally, all comments have the article_id for the originating article and, if they were a reply to another comment, the reply-to comment_id
contains all 160 articles that were scraped for comments. The available fields are - Publishing Date, Article Title, Article Tags list and Article content
@inproceedings{cojocaru2022news-ro-offense,
title={News-RO-Offense - A Romanian Offensive Language Dataset and Baseline Models Centered on News Article Comments},
author={Cojocaru, Andreea and Paraschiv, Andrei and Dascalu, Mihai},
booktitle={Proceedings of the international conference on human-computer interaction-RoCHI 2022},
pages={--},
year={2022}
}