A3NCF: An Adaptive Aspect Attention Model for Rating Prediction

Zhiyong Cheng, Ying Ding, Xiangnan He, Lei Zhu, Xuemeng Song, Mohan Kankanhalli

Abstract: Current recommender systems consider the various aspects of items for making accurate recommenda- tions. Different users place different importance to these aspects which can be thought of as a pref- erence/attention weight vector. Most existing rec- ommender systems assume that for an individual, this vector is the same for all items. However, this assumption is often invalid, especially when con- sidering a user’s interactions with items of diverse characteristics. To tackle this problem, in this pa- per, we develop a novel aspect-aware recommender model named A3NCF, which can capture the vary- ing aspect attentions that a user pays to different items. Specifically, we design a new topic model to extract user preferences and item characteristics from review texts. They are then used to 1) guide the representation learning of users and items, and 2) capture a user’s special attention on each as- pect of the targeted item with an attention network. Through extensive experiments on several large- scale datasets, we demonstrate that our model out- performs the state-of-the-art review-aware recom- mender systems in the rating prediction task.