Similarity Functions for Collaborative Master Recommendations

A memory-based collaborative system for recommending Master programs has been recently developed for University College Maastricht (UCM). Given the academic profile of a Bachelor student, the system recommends Master programs for that student based on the similarity of her profile to the profiles of the alumni students. The system is operational since September 2011 and is already popular among the UCM students. This paper considers the question of how to improve the quality of Master recommendations. For that purpose we study several academic profile representations and similarity functions. We identify the best representation strategy and show how to combine recommender systems based on different similarity functions to achieve superior Master recommendations.