Serendipity: Beyond Recommendation

In this talk Pedro will explain how the existing paradigms of search, category drill-down and social recommendation systems limit user's perceptions of existing data collections in a detrimental way. He will then propose a few strategies for introducing more serendipity and casual discovery into navigation systems, helping users find interesting content of which they were unaware.

In particular, he will present two case studies, one showing a serendipity engine whereby users remix keywords and filter content in a continuous contextual feedback loop, and another where the polysemy of images (they can have multiple meanings) creates the context for a way-finding application that enables high serendipity in the physical space. Pedro will finalise by suggesting a few other approaches.