Diversely Regularized Matrix Factorization for Accurate and Aggregately Diversified Recommendation

Jongjin Kim, Hyunsik Jeon, Jaeri Lee, U Kang

Abstract: When recommending personalized top-k items to users, how can we recommend them diversely while satisfying users’ needs? Aggregately diversified recommender systems aim to recommend a variety of items across whole users without sacrificing the recommendation accuracy. They increase the exposure opportunities of various items, which in turn increase the potential revenue of sellers as well as user satisfaction. However, it is challenging to tackle aggregate-level diversity with matrix factorization (MF), one of the most common recommendation models, since skewed real-world data lead to the skewed recommendation results of MF.