Collaborative temporal order modeling

Alexandros Karatzoglou

Abstract: Past consumption of items affect current choices and influence the perceived quality. The order in which items are consumed can affect the score that a user might give to them. In this work we present two simple models that take advantage of the temporal order of choices and ratings by the user in order to improve the quality of the recommendation. Our model exploits the collaborative effects in the data while also taking into account the order in which items are seen by the users. Experiments show that our approach outperforms standard Matrix Factorization models.