We plan to develop a novel, data-driven, student-centric “blended-learning platform”, wherethe advantages of m-learning platforms and human coaching are combined. In order to enable the envisioned learning platform, a matching module will be developed tomatch a student in need of assistance with a suitable coach, with a minimal latency.The goal of this project is to assess the feasibility and deliver a proof of concept for a matchingmodule, with following functionality:A. extracting and recognition of relevant information from an image of learning exercise,B. classifying the exercise in terms of learning subject and learning complexity,