HateBR is the first large-scale, expert-annotated dataset of Brazilian Instagram comments for offensive language and hate speech detection on the web and social media. The dataset was collected from Brazilian Instagram comments directed at politicians and manually annotated by specialists. It comprises 7,000 documents annotated across three different layers: (i) binary classification (offensive vs. non-offensive comments), (ii) offensiveness level (highly, moderately, and slightly offensive messages), and (iii) hate speech targets. Each comment was annotated by three expert annotators, achieving high inter-annotator agreement. Furthermore, baseline experiments were conducted, outperforming existing baselines in the literature for the Portuguese language. We hope that this expert-annotated dataset fosters further research on hate speech detection in the field of Natural Language Processing.
This repository contains the HateBR 2.0 corpus and its explainable version, the HateBRXplain corpus. The following image and tables provide detailed definitions of offensive language and hate speech used to label the corpus, along with the number of samples for each class. The offensiveness levels and hate speech targets are available only in HateBR 1.0. If you need access to this version, please contact me at [email protected].
| class | label | total |
|---|---|---|
| offensive | 1 | 3,500 |
| non-offensive | 0 | 3,500 |
| Total | 7,000 |
| class | label | rationales | total |
|---|---|---|---|
| offensive | 1 | human-annotated rationales | 3,500 |
| non-offensive | 0 | null | 3,500 |
| Total | 7,000 |
In addition, we also provide baseline machine learning results for both tasks: offensive language and hate speech detection. The best-obtained models are available here in .pkl files. File names are organized as [classification (offensive or hate)_representation (ngram or tfidf)_algorithms (nb, svm, mlp or lr)]. For example, the file offensive_tfidf_svm.pkl presents the model of offensive detection with tf-idf representation using the support vector machine algorithm.
Please cite our paper if you use our dataset:
@inproceedings{vargas-etal-2022-hatebr,
title = "{H}ate{BR}: A Large Expert Annotated Corpus of {B}razilian {I}nstagram Comments for Offensive Language and Hate Speech Detection",
author = "Vargas, Francielle and
Carvalho, Isabelle and
Rodrigues de G{\'o}es, Fabiana and
Pardo, Thiago and
Benevenuto, Fabr{\'\i}cio",
booktitle = "Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022)",
year = "2022",
address = "Marseille, France",
publisher = "European Language Resources Association",
url = "https://aclanthology.org/2022.lrec-1.777",
pages = "7174--7183",
} @article{Vargas_Carvalho_Pardo_Benevenuto_2024,
author={Vargas, Francielle and Carvalho, Isabelle and Pardo, Thiago A. S. and Benevenuto, Fabrício},
title={Context-aware and expert data resources for Brazilian Portuguese hate speech detection},
DOI={10.1017/nlp.2024.18},
journal={Natural Language Processing},
year={2024},
pages={435-456},
volume{31},
number={2},
url={https://www.cambridge.org/core/journals/natural-language-processing/article/contextaware-and-expert-data-resources-for-brazilian-portuguese-hate-speech-detection/7D9019ED5471CD16E320EBED06A6E923#},
}@inproceedings{salles-etal-2025-hatebrxplain,
title = "{H}ate{BRX}plain: A Benchmark Dataset with Human-Annotated Rationales for Explainable Hate Speech Detection in {B}razilian {P}ortuguese",
author = "Salles, Isadora and
Vargas, Francielle and
Benevenuto, Fabr{\'i}cio",
editor = "Rambow, Owen and
Wanner, Leo and
Apidianaki, Marianna and
Al-Khalifa, Hend and
Eugenio, Barbara Di and
Schockaert, Steven",
booktitle = "Proceedings of the 31st International Conference on Computational Linguistics (COLING 2025)",
year = "2025",
address = "Abu Dhabi, UAE",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.coling-main.446/",
pages = "6659--6669",
}