Offensive Content: A Tale of Two Classes

This dataset contains 6000 rows evenly distributed between hate speech, trolling and normal. Data source primary from Twitter, Reddit and Wikipedia Talk pages. No metadata about posts is included. Posts have been anonymized to protect user privacy.本数据集共包含6000条数据记录，样本在仇恨言论（hate speech）、网络引战（trolling）与正常文本三类间均匀分布。数据主要来源于Twitter、Reddit及维基百科讨论页（Wikipedia Talk pages），未包含帖子相关的元数据。为保护用户隐私，所有帖子均已完成匿名化处理。