Neural Autoregressive Collaborative Filtering for Implicit Feedback
This is an incremental improvement for recommendation systems using implicit feedback like clicks and watches.
The paper tackled collaborative filtering for implicit feedback by proposing implicit CF-NADE, a neural autoregressive model that converts user behaviors into like and confidence vectors to maximize a weighted negative log-likelihood, and it significantly outperformed implicit matrix factorization on a digital TV streaming dataset.
This paper proposes implicit CF-NADE, a neural autoregressive model for collaborative filtering tasks using implicit feedback ( e.g. click, watch, browse behaviors). We first convert a users implicit feedback into a like vector and a confidence vector, and then model the probability of the like vector, weighted by the confidence vector. The training objective of implicit CF-NADE is to maximize a weighted negative log-likelihood. We test the performance of implicit CF-NADE on a dataset collected from a popular digital TV streaming service. More specifically, in the experiments, we describe how to convert watch counts into implicit relative rating, and feed into implicit CF-NADE. Then we compare the performance of implicit CF-NADE model with the popular implicit matrix factorization approach. Experimental results show that implicit CF-NADE significantly outperforms the baseline.