mteb/KorHateSpeechMLClassification

KorHateSpeechMLClassification是一个韩语多标签仇恨言论数据集，包含109,692条来自韩国在线新闻评论的发言，标注有8种细粒度的仇恨言论类别（政治、起源、身体、年龄、性别、宗教、种族、咒骂）或非仇恨言论类别。每条发言可以标注一个到四个标签，以有效处理韩语语言模式。该数据集基于2018年至2020年间在Kaggle和Github上可用的韩国在线新闻评论。未标注的原始数据收集于2018年1月至2020年6月。语言生产者是2018年至2020年间在韩国在线新闻平台上留下评论的用户。

KorHateSpeechMLClassification is a Korean multi-label hate speech dataset consisting of 109,692 utterances from Korean online news comments, labeled with 8 fine-grained hate speech classes (Politics, Origin, Physical, Age, Gender, Religion, Race, Profanity) or Not Hate Speech class. Each utterance can be labeled with one to four labels to effectively handle Korean language patterns. The dataset is based on Korean online news comments available on Kaggle and Github from January 2018 to June 2020. The unlabeled raw data was collected between January 2018 and June 2020 from users who left comments on the Korean online news platform between 2018 and 2020.