nayohan/korean-hate-speech

该数据集包含韩语在线新闻评论，用于检测有毒言论。数据集的特征包括评论内容、是否包含性别偏见、偏见类型、仇恨言论类型和新闻标题。数据集分为训练集、验证集和测试集，分别包含7896、471和974个样本。数据集的下载大小为1172909字节，总大小为2008363字节。数据集的许可证为cc-by-sa-4.0，语言为韩语，标签为安全。

This dataset contains 9.4K manually labeled Korean entertainment news comments for identifying toxic speech in online news comments. The comments are annotated for social bias and hate speech, which are correlated. The dataset provides train, validation, and test sets for evaluating model performance on bias identification and hate speech detection.