ExHalder Dataset
收藏arXiv2025-09-30 收录
下载链接：
https://bit.ly/exhalder-dataset
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了6270个人工精选的新闻文章和标题示例，这些示例被标记用于检测虚构内容。其中，有1934个示例被标记为“虚构”，4336个被标记为“蕴含”，还有2074个示例附有评分员撰写的额外评论。数据集按规模分为5190个训练示例、349个验证示例和731个测试示例，其任务是标题虚构内容检测。
This dataset comprises 6,270 manually curated news article and headline examples annotated for fictional content detection. Of these, 1,934 examples are annotated as "fictional", 4,336 as "entailed", and an additional 2,074 examples are accompanied by extra comments written by annotators. The dataset is split into 5,190 training examples, 349 validation examples, and 731 test examples, with the core task being fictional content detection for news headlines.
提供机构：
Authors of the paper
