Schedule a meeting
Back to projects making thousands of documents more accessible

A vast range of educational material uploaded by teachers in Flanders would benefit from semi-automatic labeling to make the website more user-friendly. At the heart: a self learning engine that is retrained every day, and a recommender engine.


Challenge is a highly successful website where teachers can upload educational material for the community to share and reuse. The wide range of material needs to be made more accessible and user-friendly by classifying it according to a large taxonomy of thousands of labels.


We built a machine learning pipeline that learns labeling from historical data and generates label suggestions for new educational materials.

The engine is retrained every day to make sure it stays continuously updated. We also developed a recommender engine that suggests educational materials based on a user’s behavior.


Check out our fully operational learning platform here.