hltcoe/wikivideo

Wikivideo数据集是一个与Wikipedia文章相关的视频数据集，旨在用于总结任务。它包含了人类编写的关于视频数据的文章、原始Wikipedia文章、与文章相关的声明，以及音频、视频、OCR文本等模态的布尔值支持列表。数据集中的视频和音频文件被压缩为tar格式，需要特定工具进行下载和解压。数据集还包含了视频的元数据信息，如视频ID、匿名尺度ID、语言、视频类型和相关性。

The Wikivideo dataset is a video dataset associated with Wikipedia articles, designed for summarization tasks. It includes human-written articles on video data, the original Wikipedia articles, claims related to the articles, and boolean support lists for audio, video, OCR text, and other modalities. The videos and audio files in the dataset are compressed in tar format and require specific tools for downloading and decompressing. The dataset also contains metadata information for the videos, such as video ID, anonymous scale ID, language, video type, and relevance.