The FAIRview project proposes an approach for adaptive news video summarisation. Each video can have practically an infinite number of video summaries, each adapted to a specific set of user needs. For example, a given news video can generate a summary capturing all female anchors, or a summary capturing all the agricultural news items, or a summary capturing all the science news in this video, etc. Each of these video summaries will contain a portion of the original video content. In FAIRview, the selection of this portion is guided by a set of personalisation features provided by users.
There’s a growing demand from consumers to watch instead of reading news. To stay in tune with these changing viewing habits, news broadcasts are distributed in smaller units (atomised news). FAIRview presents a timely solution to this manual effort, by generating adaptive news video summaries suitable for distribution on social media, as well as on the archive’s own video search environment. Secondly, as “fake videos” are on the rise as digital video manipulation techniques advance, FAIRview develops metrics to assess the quality of the video summaries and their potential for misinformation.
The FAIRview project is made possible by the Google Digital News Innovation Fund.