textdetox/multilingual_toxic_spans

这是一个多语言的可解释毒性数据集，包含英语、德语、西班牙语、俄语、乌克兰语、阿拉伯语、阿姆哈拉语、中文和印地语等九种语言的句子及其负面含义。数据集用于研究文本的毒性程度，并提供了每种语言毒性关键词的提取结果。数据集旨在为可解释的毒性缓解提供支持，并已应用于多语言文本的去毒性研究。

This is a multilingual explainable toxicity dataset containing sentences and their negative connotations in nine languages: English, German, Spanish, Russian, Ukrainian, Arabic, Amharic, Chinese, and Hindi. The dataset is used for researching the toxicity level of text and provides extracted results of toxic keywords for each language. It aims to support explainable toxicity mitigation and has been applied in research on multilingual text detoxification.