Collaborative filtering based on nonnegative/binary matrix factorization

Collaborative filtering generates recommendations by exploiting user-item similarities based on rating data, which often contains numerous unrated items. This paper proposes a nonnegative/binary matrix factorization (NBMF) algorithm modified for collaborative filtering and demonstrates that utilizing a low-latency Ising machine in NBMF is advantageous in terms of computation time. While previous studies have primarily applied NBMF to dense data, such as images, this study applies a modified NBMF to sparse data. Results show the benefits of using a low-latency Ising machine to implement the proposed method.