COCOLOFA
收藏资源简介:
COCOLOFA是由威斯康星大学麦迪逊分校和宾夕法尼亚州立大学创建的最大的英文逻辑谬误数据集,包含7,706条新闻评论,每条评论都标注了是否存在逻辑谬误及其类型。数据集通过143名众包工作者在GPT-4辅助下生成,涵盖了648篇新闻文章。数据集的创建过程包括选择新闻文章、指定逻辑谬误类型、收集评论等步骤,旨在解决自动检测文本中逻辑谬误的问题,特别是在在线讨论中的应用。
COCOLOFA is the largest English logical fallacy dataset created by the University of Wisconsin–Madison and Pennsylvania State University. It contains 7,706 news comments, each annotated with the presence or absence of logical fallacies and their specific types. Developed by 143 crowdworkers with the assistance of GPT-4, the dataset covers 648 news articles. The dataset construction process includes steps such as news article selection, specification of logical fallacy categories, and comment collection. It aims to address the problem of automatic detection of logical fallacies in text, particularly for applications in online discussions.




