Recommender systems: the case of repeated interaction in matrix factorization

This work presents a new matrix factorization recommender system approach, that takes repeated interaction into account. We analyze if and how users' repeated interaction behavior---such as repeat purchases---can be integrated into a recommender system. We develop a method that takes advantage of this additional data dimension that is studied in many other fields to derive useful conclusions. Furthermore, we empirically test our method on real-life retailer data and on the Last.fm dataset. We compare our algorithm with popular matrix factorization approaches. Results indicate that our method manages to slightly outperform the existing methods.