Data-driven study: Augmenting prediction accuracy of recommendations in social learning platforms (abstract)
Soude Fazeli, Hendrik Drachsler, Peter Sloep · Research Repository (Delft University of Technology) · 2013
This study aims to develop a recommender system for a social learning platform to be provided by EU FP7 Open Discovery Space (ODS) project by taking into account social data of users to make recommendations. In this paper, we investigate which recommender algorithm can best fits social learning platforms like ODS platform. We conducted an experiment to test a set of different classical collaborative filtering algorithms on representative educational datasets similar to the future ODS dataset, as well as on the MovieLens dataset as a reference for studies on recommender systems. In addition to the classical collaborative filtering algorithms, we evaluated a graph-based recommender approach called T-index. We compare performance of the used algorithms in terms of F1 score. We also show how T-index approach can provide a balanced distribution of users degree centrality.