Ratings and Classification Data

R11 - Yahoo News Video dataset, version 1.0 (645MB)

The dataset is a collection of 964 hours (22K videos) of news broadcast videos that appeared on Yahoo news website's properties, e.g., World News, US News, Sports, Finance, and a mobile application during August 2017. The videos were either part of an article or displayed standalone in a news property. Many of the videos served in this platform lack important metadata, such as an exhaustive list of topics associated with the video. We label each of the videos in the dataset using a collection of 336 tags based on a news taxonomy designed by in-house editors. In the taxonomy, the closer the tag is to the root, the more generic (topically) it is.