nayohan/KOLD
收藏Hugging Face2024-07-07 更新2024-07-22 收录
下载链接:
https://hf-mirror.com/datasets/nayohan/KOLD
下载链接
链接失效反馈官方服务:
资源简介:
KOLD(Korean Offensive Language Dataset)是一个用于研究韩语中冒犯性语言的数据集。数据集包含多个特征,如唯一标识符(guid)、来源(source)、日期(date)、标题(title)、评论(comment)等。其中,OFF表示是否包含冒犯性内容,TGT和GRP分别表示目标对象和目标群体。数据集还包含注释信息,如注释者ID、冒犯性内容的起止索引、目标对象的起止索引等。数据集分为训练集,包含40429个样本,文件大小为20683839字节。数据集的语言为韩语,标签为safety。
KOLD (Korean Offensive Language Dataset) is a dataset designed for studying offensive language in Korean. The dataset includes various features such as guid, source, date, title, comment, etc. The OFF feature indicates whether the content is offensive, while TGT and GRP represent the target and target group, respectively. The dataset also contains annotation information, such as annotator ID, start and end indices of offensive content, and start and end indices of the target. The dataset is divided into a training set containing 40,429 samples, with a file size of 20,683,839 bytes. The language of the dataset is Korean, and it is tagged with safety.
提供机构:
nayohan
原始信息汇总
数据集概述
数据集信息
-
特征:
guid: 字符串类型source: 字符串类型date: 时间戳类型title: 字符串类型comment: 字符串类型OFF: 布尔类型TGT: 字符串类型GRP: 字符串类型OFF_span: 字符串类型TGT_span: 字符串类型raw_labels: 列表类型,包含以下子特征:annotator_id: 整数类型off_end_idx: 整数序列类型off_start_idx: 整数序列类型offensiveness: 布尔类型target: 字符串序列类型target_group: 字符串序列类型tgt_end_idx: 整数序列类型tgt_start_idx: 整数序列类型
-
分割:
train: 包含40429个样本,占用20683839字节
-
下载大小: 8176787字节
-
数据集大小: 20683839字节
配置
- 默认配置:
data_files:train: 路径为data/train-*
语言
- 韩语 (ko)
标签
- 安全性 (safety)



