The effect of algorithmic bias on recommender systems for massive open online courses

Most recommender systems are evaluated on how they accurately predict user ratings. However, individuals use them for more than an anticipation of their preferences. The literature demonstrated that some recommendation algorithms achieve good prediction accuracy, but suffer from popularity bias. Other algorithms generate an item category bias due to unbalanced rating distributions across categories. These effects have been widely analyzed in the context of books, movies, music, and tourism, but contrasting conclusions have been reached so far. In this paper, we explore how recommender systems work in the context of massive open online courses, going beyond prediction accuracy. To this end, we compared existing algorithms and their recommended lists against biases related to course popularity, catalog coverage, and course category popularity. Our study remarks even more the need of better understanding how recommenders react against bias in diverse contexts.