Recommenders with a mission: towards a more diverse internet

In the current age of information overload, we rely increasingly on recommender systems that help us filter the information available down to those items that are interesting or relevant to us. Typically, these systems are based on general popularity of the items or similarity to items we have interacted with before. This approach works well in domains such as product or movie recommendation, but could be problematic in the context of news recommendation. Recently concerns have been raised over this approach inducing filter bubbles and echo chambers, where people primarily see items that are in line with their own views and preferences. Simultaneously, when done well, these news recommender systems could also be used to expose users to viewpoints different from what they are used to. This session will facilitate a discussion on what diversity in the context of news recommenders would look like.