EduVid-QA: An Annotated Question–Answering Dataset for Educational YouTube Videos

EduVid-QA is an annotated question–answering dataset built from publicly available educational YouTube videos. The dataset contains link to the videos with aligned transcripts, manually created multiple-choice and open-ended questions, and verified reference answers grounded in the video and transcript content. Only video links and derived annotations are provided; the videos themselves are not redistributed. EduVid-QA is intended for research on educational video understanding, multimodal reasoning, and retrieval-based question answering.

EduVid-QA是一款基于公开可获取的教育类YouTube视频构建的带标注问答数据集。该数据集包含带有对齐字幕的视频链接、人工撰写的选择题与开放式问题，以及经验证且依托视频及字幕内容生成的参考答案。本数据集仅提供视频链接与衍生标注，不重新分发视频本体。EduVid-QA旨在服务于教育视频理解、多模态推理以及基于检索的问答相关研究。