UTSNLPGroup/PCR-ToxiCN
收藏官方服务：
资源简介：
PCR-ToxiCN是一个用于测试模型识别通过音译伪装隐藏攻击性语言的中文数据集，包含500个例子，涵盖同音词和近音词伪装的攻击性语言。数据集包括原始评论、是否攻击性的标签以及伪装策略等信息。
PCR-ToxiCN is a Chinese dataset with 500 examples designed to test models ability to detect offensive language hidden by phonetic cloaking, including homophones and near-homophones. It includes original comments, offensive labels, and cloaking strategies.
提供机构：
UTSNLPGroup
