UKPLab/HaluQuestQA

HaluQuestQA是一个针对长文档问题回答的英文数据集，包含1.8k个带有span-level错误注释的问题和答案对。数据集旨在帮助研究和改进问题回答系统，特别是针对长文本上下文的问题。它包括了事实性错误、不相关错误、不完整答案错误和问题误解错误的标注，以及相应的错误解释。数据集适用于问题回答和文本生成任务。

HaluQuestQA is an English dataset for long-document question answering, containing 1.8k question-answer pairs with span-level error annotations. The dataset is designed to aid research and improvement in question answering systems, especially for questions involving long text contexts. It includes annotations for factual errors, irrelevance errors, incomplete answer errors, and question misconception errors, along with corresponding error explanations. The dataset is suitable for question answering and text generation tasks.