Dataset Card for Jigsaw Toxic Comments

  Dataset





  Dataset Description

The Jigsaw Toxic Comments dataset is a benchmark dataset created for the Toxic Comment Classification Challenge on Kaggle. It is designed to help develop machine learning models that can identify and classify toxic online comments across multiple categories of toxicity.

Curated by: Jigsaw (a technology incubator within Alphabet Inc.) Shared by: Kaggle Language(s) (NLP): English License: CC0 1.0… See the full description on the dataset page: https://huggingface.co/datasets/anitamaxvim/jigsaw-toxic-comments.

The comments in this dataset come from an archive of Wikipedia talk page comments. These have been annotated by Jigsaw for toxicity, as well as (for the main config) a variety of toxicity subtypes, including severe toxicity, obscenity, threatening language, insulting language, and identity attacks. This dataset is a replica of the data released for the Jigsaw Toxic Comment Classification Challenge and Jigsaw Multilingual Toxic Comment Classification competition on Kaggle, with the test dataset merged with the test_labels released after the end of the competitions. Test data not used for scoring has been dropped. This dataset is released under CC0, as is the underlying comment text.

To use this dataset:

import tensorflow_datasets as tfds

ds = tfds.load('wikipedia_toxicity_subtypes', split='train')
for ex in ds.take(4):
 print(ex)

See the guide for more informations on tensorflow_datasets.

This dataset is a toxicity dataset consisting of several columns capturing various attributes of nanoparticles (NPs) and their toxicological effects. The dataset contains various features related to nanoparticles (NPs) and their properties, which are likely related to toxicity classification. Here is an overview of the columns in the dataset:

NPs: Type of nanoparticles (e.g., Al2O3). coresize: Core size of the nanoparticles. hydrosize: Hydrodynamic size of the nanoparticles. surfcharge: Surface charge of the nanoparticles. surfarea: Surface area of the nanoparticles. Ec: Electric Charge Expotime: Exposure time. Dosage: amount of material used. e: Energy-related feature. NOxygen: Number of oxygen atoms. class: Class label indicating whether the nanoparticles are toxic or non-toxic.

Kaggle toxic dataset annotated with gpt-4o-mini with the same prompt used to annotate Toxic-Commons Celadon

This version of the CivilComments Dataset provides access to the primary seven labels that were annotated by crowd workers, the toxicity and other tags are a value between 0 and 1 indicating the fraction of annotators that assigned these attributes to the comment text.

The other tags are only available for a fraction of the input examples. They are currently ignored for the main dataset; the CivilCommentsIdentities set includes those labels, but only consists of the subset of the data with them. The other attributes that were part of the original CivilComments release are included only in the raw data. See the Kaggle documentation for more details about the available features.

The comments in this dataset come from an archive of the Civil Comments platform, a commenting plugin for independent news sites. These public comments were created from 2015 - 2017 and appeared on approximately 50 English-language news sites across the world. When Civil Comments shut down in 2017, they chose to make the public comments available in a lasting open archive to enable future research. The original data, published on figshare, includes the public comment text, some associated metadata such as article IDs, publication IDs, timestamps and commenter-generated "civility" labels, but does not include user ids. Jigsaw extended this dataset by adding additional labels for toxicity, identity mentions, as well as covert offensiveness. This data set is an exact replica of the data released for the Jigsaw Unintended Bias in Toxicity Classification Kaggle challenge. This dataset is released under CC0, as is the underlying comment text.

For comments that have a parent_id also in the civil comments data, the text of the previous comment is provided as the "parent_text" feature. Note that the splits were made without regard to this information, so using previous comments may leak some information. The annotators did not have access to the parent text when making the labels.

To use this dataset:

import tensorflow_datasets as tfds

ds = tfds.load('civil_comments', split='train')
for ex in ds.take(4):
 print(ex)

See the guide for more informations on tensorflow_datasets.

tcapelle/kaggle-toxic-annotated-filtered dataset hosted on Hugging Face and contributed by the HF Datasets community

Preporcessed Toxic Comments Classification Dataset

The obstacle I faced in Toxic Comments Classification Challenge was the preprocessing part. One can easily improve their LB performance if the preprocessing is done right.

This is the preprocessed version of Toxic Comments Classification Challenge dataset. The code for preprocessing: https://www.kaggle.com/fizzbuzz/toxic-data-preprocessing

A cleaned up version of train dataset from kaggle, the Toxic Comment Classification Challenge

https://www.kaggle.com/competitions/jigsaw-toxic-comment-classification-challenge https://www.kaggle.com/competitions/jigsaw-toxic-comment-classification-challenge/data?select=train.csv.zip the alt_format directory contains an alternate format intended for a tutorial.

What was done:

Removed extra spaces and new lines Removed non-printing characters Removed punctuation except apostrophe… See the full description on the dataset page: https://huggingface.co/datasets/vluz/Tox.

Dataset

This dataset was created by x2022gmv

Contents

Dataset

This dataset was created by Roshan Velpula

Contents

ToxicConversationsClassification An MTEB dataset Massive Text Embedding Benchmark

Collection of comments from the Civil Comments platform together with annotations if the comment is toxic or not.

Task category t2c

Domains Social, Written Reference https://www.kaggle.com/competitions/jigsaw-unintended-bias-in-toxicity-classification/overview

  How to evaluate on this task

You can evaluate an embedding model on this dataset using the following code: import… See the full description on the dataset page: https://huggingface.co/datasets/mteb/toxic_conversations_50k.

Dataset

This dataset was created by Gaurav Dutta

Contents

Dataset

This dataset was created by Juntao Liang

Released under CC0: Public Domain

Contents

This classifier is a fine-tuned Roberta-base model using the simpletransformers toolkit for classifying toxic language. We use the Kaggle Toxic Comments Challenge dataset as training data. This dataset contains Wikipedia comments classified as toxic or non-toxic.

Dataset

This dataset was created by Ilya Ezepov

Contents

FinToxicityClassification An MTEB dataset Massive Text Embedding Benchmark

This dataset is a DeepL -based machine translated version of the Jigsaw toxicity dataset for Finnish. The dataset is originally from a Kaggle competition https://www.kaggle.com/c/jigsaw-toxic-comment-classification-challenge/data.
The original dataset poses a multi-label text classification problem and includes the labels identity_attack, insult, obscene, severe_toxicity, threat and toxicity.
Here… See the full description on the dataset page: https://huggingface.co/datasets/mteb/FinToxicityClassification.

This dataset is a customized and re-labeled version of the original Jigsaw Toxic Comment Classification Challenge dataset.

Instead of toxic behavior categories, the comments are now annotated with depression severity levels, aiming to support mental health research and AI-based early detection of psychological distress.

🗂️ Label Categories: Each comment has been carefully annotated into one of the following classes: - psychotic_depression - severe_depression - moderate_depression - mild_depression - toxic_depression - major_depression

These labels help transform the original problem into a multi-class depression severity classification task. 👨‍💻 Project Contributors: - Muhammad Mugees Asif — Lead Annotator & AI Researcher - Dr. Arfan Ali Nagra — Computational Intelligence Expert - Sana Asif — Mental Health Research Support & Dataset Coordination

This dataset was created with the intention to help data scientists, researchers, and students work on AI solutions for mental health support.

⚠️ Acknowledgement: The original dataset was sourced from the Jigsaw Toxic Comment Classification Challenge hosted on Kaggle. Full credit to the creators of the original dataset. This re-labeled version is shared for educational and research purposes only.

Dataset Summary

This dataset is a DeepL -based machine translated version of the Jigsaw toxicity dataset for Finnish. The dataset is originally from a Kaggle competition https://www.kaggle.com/c/jigsaw-toxic-comment-classification-challenge/data. The dataset poses a multi-label text classification problem and includes the labels identity_attack, insult, obscene, severe_toxicity, threat and toxicity.

  Example data

{ "label_identity_attack": 0, "label_insult": 0… See the full description on the dataset page: https://huggingface.co/datasets/TurkuNLP/jigsaw_toxicity_pred_fi.

Misalignment Toxic Comments Dataset

A curated collection of toxic comments and tweets for LLM misalignment research.

  Dataset Description

This dataset contains only toxic comments and tweets, drawn from two established sources:

Hate Speech and Offensive Language Datasethttps://www.kaggle.com/datasets/mrmorj/hate-speech-and-offensive-language-dataset/data Wikipedia Talk Labels: Personal… See the full description on the dataset page: https://huggingface.co/datasets/Masabanees619/toxic_tweets_and_comments.

Dataset

This dataset was created by DataNerd

Released under CC0: Public Domain

Contents

All thing toxic about the datasets and more.