ssgyejin/KOTOX-classification

KOTOX是一个专为有毒文本脱雾和解毒稳健性研究设计的韩语数据集。它包含中性-有毒句子及其混淆对应物的成对样本，并基于17种语言学基础的转换规则构建。该数据集支持三个互补任务：混淆有毒文本分类、中性文本脱雾和混淆有毒文本清洗。数据集分为简单、正常和困难三个难度级别。KOTOX is a Korean toxic text dataset designed for robustness research in detoxification and deobfuscation. It contains paired neutral-toxic sentences and their obfuscated counterparts, constructed with 17 linguistically grounded transformation rules. The dataset supports three complementary tasks: obfuscated toxic text classification, neutral text deobfuscation, and obfuscated toxic text sanitization. It is divided into three difficulty levels: easy, normal, and hard.