PRMNBR: Personalized Recommendation Model for Next Basket Recommendation Using User’s Long-Term Preference, Short-Term Preference, and Repetition Behaviour

Abstract

Next Basket Recommendation (NBR) tries to recommend items in a user's coming basket by understanding the user's characteristics from the past baskets of the user.The existing deep learning models for recommendation system (RS) are formulated by combining the long-term and short-term preferences of the user successfully.Recent statistical-based models highlight the importance of repeat purchase behavior, especially in the E-commerce industry, as most customer repeatedly purchases items.Including repeat behaviour dynamics can lead to a certain degree of improvement in the deep learning-based NBR models, as shown in a few recent statistical-based works.In this paper, we introduced a mechanism to extract the user's repetition behaviour along with the user's long-term preferences and short-term preferences.To capture the repetition behavior of the user, we introduced the encoded user's baskets as Repeat Aware Baskets, and to extract the correlation between items, we used a Correlation Sensitive Basket.Further, separate embedding is generated with respect to Repeat Aware and Correlation Sensitive Baskets.These embedding are fed parallel to two layered Long-Short Term Memory architecture for analyzing short-term preference.To evaluate the performance of the proposed model, we experimented on two data sets.Our proposed algorithm outperformed various recently developed models over various performance metrics.