About the research
Search engines and recommendation systems use "machine learning": self-learning methods to learn from user behavior so that they automatically adapt to their preferences. Consider, for example, the system that recommends videos on services such as Netflix or YouTube. The challenge with learning from user behavior is that this behavior often says more about what recommendations or search results users were shown, than users' actual preferences. Harrie Oosterhuis' research focuses on separating these two factors. He has devised a new way that combines several approaches and is already accurate with much less data. As a result, small companies can also use a recommendation system and large tech companies can learn more efficiently from their customers.
