ReCANet: A Repeat Consumption-Aware Neural Network for Next Basket Recommendation in Grocery Shopping

Retailers such as grocery stores or e-marketplaces often have vast selections of items for users to choose from. Predicting a user's next purchases has gained attention recently, in the form of next basket recommendation (NBR), as it facilitates navigating extensive assortments for users. Neural network-based models that focus on learning basket representations are the dominant approach in the recent literature. However, these methods do not consider the specific characteristics of the grocery shopping scenario, where users shop for grocery items on a regular basis, and grocery items are repurchased frequently by the same user.

In this paper, we first gain a data-driven understanding of users' repeat consumption behavior through an empirical study on six public and proprietary grocery shopping transaction datasets. We discover that, averaged over all datasets, over 54% of NBR performance in terms of recall comes from repeat items: items that users have already purchased in their history, which constitute only 1% of the total collection of items on average. A NBR model with a strong focus on previously purchased items can potentially achieve high performance. We introduce ReCANet, a repeat consumption-aware neural network that explicitly models the repeat consumption behavior of users in order to predict their next basket. ReCANet significantly outperforms state-of-the-art models for the NBR task, in terms of recall and nDCG. We perform an ablation study and show that all of the components of ReCANet contribute to its performance, and demonstrate that a user's repetition ratio has a direct influence on the treatment effect of ReCANet.