Conditioned Variational Autoencoder for Top-N Item Recommendation

State-of-the-art recommender systems (RSs) generally try to improve the overall recommendation quality. However, users usually tend explicitly filter item set based on available categories, e.g., smartphone brands, movie genres. For this reason, an RS that can make step automatically is likely increase user’s experience. This paper proposes a Conditioned Variational Autoencoder (C-VAE) for constrained top-N where recommended items must satisfy given condition. The proposed model architecture similar standard VAE in which condition vector fed into encoder. ranking learned during training thanks new reconstruction loss takes input account. We show our generalizes state-of-the-art Mult-VAE collaborative filtering model. Experimental results underline potential of C-VAE providing accurate recommendations under constraints. Finally, performed analyses suggest be used other scenarios, such as context-aware recommendation.