Datasets:

textdetox
/

multilingual_toxicity_dataset

	---
	language:
	- en
	- ru
	- uk
	- de
	- es
	- am
	- zh
	- ar
	- hi
	- it
	- fr
	- he
	- ja
	- tt
	license: openrail++
	size_categories:
	- 10K<n<100K
	task_categories:
	- text-classification
	dataset_info:
	features:
	- name: text
	dtype: string
	- name: toxic
	dtype: int64
	splits:
	- name: en
	num_bytes: 411178
	num_examples: 5000
	- name: ru
	num_bytes: 710001
	num_examples: 5000
	- name: uk
	num_bytes: 630595
	num_examples: 5000
	- name: de
	num_bytes: 941017
	num_examples: 5000
	- name: es
	num_bytes: 978750
	num_examples: 5000
	- name: am
	num_bytes: 1102628
	num_examples: 5000
	- name: zh
	num_bytes: 359235
	num_examples: 5000
	- name: ar
	num_bytes: 889661
	num_examples: 5000
	- name: hi
	num_bytes: 1842662
	num_examples: 5000
	- name: it
	num_bytes: 791069
	num_examples: 5000
	- name: fr
	num_bytes: 621103
	num_examples: 5000
	- name: he
	num_bytes: 243823
	num_examples: 2011
	- name: hin
	num_bytes: 836167
	num_examples: 4363
	- name: tt
	num_bytes: 764917
	num_examples: 5000
	- name: ja
	num_bytes: 714729
	num_examples: 5000
	download_size: 6802095
	dataset_size: 11837535
	configs:
	- config_name: default
	data_files:
	- split: en
	path: data/en-*
	- split: ru
	path: data/ru-*
	- split: uk
	path: data/uk-*
	- split: de
	path: data/de-*
	- split: es
	path: data/es-*
	- split: am
	path: data/am-*
	- split: zh
	path: data/zh-*
	- split: ar
	path: data/ar-*
	- split: hi
	path: data/hi-*
	- split: it
	path: data/it-*
	- split: fr
	path: data/fr-*
	- split: he
	path: data/he-*
	- split: hin
	path: data/hin-*
	- split: tt
	path: data/tt-*
	- split: ja
	path: data/ja-*
	---

	# Multilingual Toxicity Detection Dataset

	[2025] We extend our binary toxicity classification dataset to more languages! Now also covered: Italian, French, Hebrew, Hindglish, Japanese, Tatar. The data is prepared for [TextDetox 2025](https://pan.webis.de/clef25/pan25-web/text-detoxification.html) shared task.

	[2024] For the shared task [TextDetox 2024](https://pan.webis.de/clef24/pan24-web/text-detoxification.html), we provide a compilation of binary toxicity classification datasets for each language.
	Namely, for each language, we provide 5k subparts of the datasets -- 2.5k toxic and 2.5k non-toxic samples.

	The list of original sources:
	* English: [Jigsaw](https://www.kaggle.com/c/jigsaw-toxic-comment-classification-challenge), [Unitary AI Toxicity Dataset](https://github.com/unitaryai/detoxify)
	* Russian: [Russian Language Toxic Comments](https://www.kaggle.com/datasets/blackmoon/russian-language-toxic-comments), [Toxic Russian Comments](https://www.kaggle.com/datasets/alexandersemiletov/toxic-russian-comments)
	* Ukrainian: [ours](https://huggingface.co/datasets/ukr-detect/ukr-toxicity-dataset)
	* Spanish: [CLANDESTINO, the Spanish toxic language dataset](https://github.com/microsoft/Clandestino/tree/main)
	* German: [DeTox-Dataset](https://github.com/hdaSprachtechnologie/detox), [GemEval 2018, 2021](https://aclanthology.org/2021.germeval-1.1/)
	* Amhairc: [Amharic Hate Speech](https://github.com/uhh-lt/AmharicHateSpeech)
	* Arabic: [OSACT4](https://edinburghnlp.inf.ed.ac.uk/workshops/OSACT4/)
	* Hindi: [Hostility Detection Dataset in Hindi](https://competitions.codalab.org/competitions/26654#learn_the_details-dataset), [Overview of the HASOC track at FIRE 2019: Hate Speech and Offensive Content Identification in Indo-European Languages](https://dl.acm.org/doi/pdf/10.1145/3368567.3368584?download=true)
	* Italian: [AMI](https://github.com/dnozza/ami2020), [HODI](https://github.com/HODI-EVALITA/HODI_2023), [Jigsaw Multilingual Toxic Comment](https://www.kaggle.com/competitions/jigsaw-multilingual-toxic-comment-classification/overview)
	* French: []
	* Hebrew: [Hebrew Offensive Language Dataset](https://github.com/NataliaVanetik/HebrewOffensiveLanguageDatasetForTheDetoxificationProject/blob/main/OLaH-dataset-filtered.xlsx)
	* Hinglish: []
	* Japanese: [filtered](https://huggingface.co/datasets/sobamchan/ja-toxic-text-classification-open2ch) [2chan posts](https://huggingface.co/datasets/p1atdev/open2ch) by Perspective API;
	* Tatar: ours.

	All credits go to the authors of the original corpora.

	## Citation
	If you would like to acknowledge our work, please, cite the following manuscripts:

	[2024]

	```
	@inproceedings{dementieva2024overview,
	title={Overview of the Multilingual Text Detoxification Task at PAN 2024},
	author={Dementieva, Daryna and Moskovskiy, Daniil and Babakov, Nikolay and Ayele, Abinew Ali and Rizwan, Naquee and Schneider, Frolian and Wang, Xintog and Yimam, Seid Muhie and Ustalov, Dmitry and Stakovskii, Elisei and Smirnova, Alisa and Elnagar, Ashraf and Mukherjee, Animesh and Panchenko, Alexander},
	booktitle={Working Notes of CLEF 2024 - Conference and Labs of the Evaluation Forum},
	editor={Guglielmo Faggioli and Nicola Ferro and Petra Galu{\v{s}}{\v{c}}{\'a}kov{\'a} and Alba Garc{\'i}a Seco de Herrera},
	year={2024},
	organization={CEUR-WS.org}
	}
	```

	```
	@inproceedings{dementieva-etal-2024-toxicity,
	title = "Toxicity Classification in {U}krainian",
	author = "Dementieva, Daryna and
	Khylenko, Valeriia and
	Babakov, Nikolay and
	Groh, Georg",
	editor = {Chung, Yi-Ling and
	Talat, Zeerak and
	Nozza, Debora and
	Plaza-del-Arco, Flor Miriam and
	R{\"o}ttger, Paul and
	Mostafazadeh Davani, Aida and
	Calabrese, Agostina},
	booktitle = "Proceedings of the 8th Workshop on Online Abuse and Harms (WOAH 2024)",
	month = jun,
	year = "2024",
	address = "Mexico City, Mexico",
	publisher = "Association for Computational Linguistics",
	url = "https://aclanthology.org/2024.woah-1.19/",
	doi = "10.18653/v1/2024.woah-1.19",
	pages = "244--255",
	abstract = "The task of toxicity detection is still a relevant task, especially in the context of safe and fair LMs development. Nevertheless, labeled binary toxicity classification corpora are not available for all languages, which is understandable given the resource-intensive nature of the annotation process. Ukrainian, in particular, is among the languages lacking such resources. To our knowledge, there has been no existing toxicity classification corpus in Ukrainian. In this study, we aim to fill this gap by investigating cross-lingual knowledge transfer techniques and creating labeled corpora by: (i){\textasciitilde}translating from an English corpus, (ii){\textasciitilde}filtering toxic samples using keywords, and (iii){\textasciitilde}annotating with crowdsourcing. We compare LLMs prompting and other cross-lingual transfer approaches with and without fine-tuning offering insights into the most robust and efficient baselines."
	}
	```

	```
	@inproceedings{DBLP:conf/ecir/BevendorffCCDEFFKMMPPRRSSSTUWZ24,
	author = {Janek Bevendorff and
	Xavier Bonet Casals and
	Berta Chulvi and
	Daryna Dementieva and
	Ashaf Elnagar and
	Dayne Freitag and
	Maik Fr{\"{o}}be and
	Damir Korencic and
	Maximilian Mayerl and
	Animesh Mukherjee and
	Alexander Panchenko and
	Martin Potthast and
	Francisco Rangel and
	Paolo Rosso and
	Alisa Smirnova and
	Efstathios Stamatatos and
	Benno Stein and
	Mariona Taul{\'{e}} and
	Dmitry Ustalov and
	Matti Wiegmann and
	Eva Zangerle},
	editor = {Nazli Goharian and
	Nicola Tonellotto and
	Yulan He and
	Aldo Lipani and
	Graham McDonald and
	Craig Macdonald and
	Iadh Ounis},
	title = {Overview of {PAN} 2024: Multi-author Writing Style Analysis, Multilingual
	Text Detoxification, Oppositional Thinking Analysis, and Generative
	{AI} Authorship Verification - Extended Abstract},
	booktitle = {Advances in Information Retrieval - 46th European Conference on Information
	Retrieval, {ECIR} 2024, Glasgow, UK, March 24-28, 2024, Proceedings,
	Part {VI}},
	series = {Lecture Notes in Computer Science},
	volume = {14613},
	pages = {3--10},
	publisher = {Springer},
	year = {2024},
	url = {https://doi.org/10.1007/978-3-031-56072-9\_1},
	doi = {10.1007/978-3-031-56072-9\_1},
	timestamp = {Fri, 29 Mar 2024 23:01:36 +0100},
	biburl = {https://dblp.org/rec/conf/ecir/BevendorffCCDEFFKMMPPRRSSSTUWZ24.bib},
	bibsource = {dblp computer science bibliography, https://dblp.org}
	}
	```