Patch Hate Crime Dataset

该数据集来源于Patch网站，包含了超本地新闻，并对其进行了标注，以指明新闻报道的是否为仇恨犯罪事件以及仇恨犯罪的类型属性。该数据集总共包含5171个样本，其中1979个为正样本，3192个为负样本。此外，数据集被划分为训练集（3619个样本）、验证集（513个样本）和测试集（1039个样本）。规模上，该数据集包含5171个样本，其任务是对仇恨犯罪预测进行事件提取。

This dataset is sourced from the Patch website, which contains hyper-local news and has been annotated to specify whether a news report involves a hate crime incident and the type attribute of the hate crime. In total, this dataset comprises 5171 samples, with 1979 positive samples and 3192 negative samples. Furthermore, the dataset is split into a training set (3619 samples), a validation set (513 samples), and a test set (1039 samples). In terms of scale, this dataset includes 5171 samples, and its task is event extraction for hate crime prediction.