Popping Bubbles: How to Categorize and Recommend Information Video

In this presentation, Mariah Martin Shein, Vubble’s Director of Machine Learning, talks about how Vubble leverages our massive information video dataset to develop automated tools for video categorization and recommendation.  For categorization, our goal is not just to produce high-quality predictions, but also to continuously provide support to our journalist editors, our journalists-in-the-loop, by offloading repetitive tasks to machines. For recommendation, Vubble’s algorithms suggest videos that balance personalization with showing serendipitous alternate viewpoints. This blend provides an engaging experience that helps ‘pop’ individuals’ news bubbles and encourages critical thinking. This talk was presented at the Waterloo Data Science and Data Engineering Meetup Group, Feb. 4, 2021.