ETHOS (multi-labEl haTe speecH detectiOn dataSet)

ETHOS：多标签仇恨语音检测数据集。该存储库包含一个用于在社交媒体平台上检测仇恨言论的数据集，称为 Ethos。数据集有两种变体： Ethos_Dataset_Binary.csv[Ethos_Dataset_Binary.csv] 包含数据集中的 998 条评论以及关于仇恨言论存在与否的标签。其中 565 篇不包含仇恨言论，其余 433 篇包含。 Ethos_Dataset_Multi_Label.csv [Ethos_Dataset_Multi_Label.csv] 包含 8 个标签，用于 433 条带有仇恨言论内容的评论。这些标签是暴力（如果它煽动（1）或不（0）暴力），directed_vs_general（如果它是针对一个人（1）或一个群体（0）），以及6个关于仇恨言论类别的标签，比如，性别、种族、国籍、残疾、宗教和性取向。ETHOS: A Multi-label Hate Speech Detection Dataset. This repository contains a dataset named Ethos for detecting hate speech on social media platforms. The dataset has two variants: Ethos_Dataset_Binary.csv contains 998 comments from the dataset along with labels indicating the presence or absence of hate speech. Of these, 565 comments do not contain hate speech, while the remaining 433 comments do. Ethos_Dataset_Multi_Label.csv includes 8 labels applied to 433 comments with hate speech content. These labels are: violence (whether it incites violence (1) or not (0)), directed_vs_general (whether it is targeted at an individual (1) or a group (0)), and 6 labels for hate speech categories including gender, race, nationality, disability, religion, and sexual orientation.

搜集汇总

背景与挑战

背景概述

ETHOS是一个多标签仇恨言论检测数据集，包含998条社交媒体评论，分为二元分类（仇恨言论存在与否）和多标签分类（8个具体标签）两种变体。该数据集由塞萨洛尼基亚里士多德大学于2022年发布，适用于仇恨言论检测研究。