Modeling Temporal Positive and Negative Excitation for Sequential Recommendation

TL;DR

A novel Static-Dynamic Interest Learning (SDIL) framework featured with a novel Temporal Positive and Negative Excitation Modeling (TPNE) module for accurate sequential recommendation and extensive experiments show that SDIL can effectively capture both static and dynamic interest and outperforms state-of-the-art baselines.

Abstract

Sequential recommendation aims to predict the next item which interests users via modeling their interest in items over time. Most of the existing works on sequential recommendation model users' dynamic interest in specific items while overlooking users' static interest revealed by some static attribute information of items, e.g., category, or brand. Moreover, existing works often only consider the positive excitation of a user's historical interactions on his/her next choice on candidate items while ignoring the commonly existing negative excitation, resulting in insufficient modeling dynamic interest. The overlook of static interest and negative excitation will lead to incomplete interest modeling and thus impede the recommendation performance. To this end, in this paper, we propose modeling both static interest and negative excitation for dynamic interest to further improve the recommendation performance. Accordingly, we design a novel Static-Dynamic Interest Learning (SDIL) framework featured with a novel Temporal Positive and Negative Excitation Modeling (TPNE) module for accurate sequential recommendation. TPNE is specially designed for comprehensively modeling dynamic interest based on temporal positive and negative excitation learning. Extensive experiments on three real-world datasets show that SDIL can effectively capture both static and dynamic interest and outperforms state-of-the-art baselines.

Figures (7)

• Figure 1: An example of recommendations via modeling positive excitation only (existing methods) and modeling both positive and negative excitation (our proposal). Clearly, the latter achieves better performance via ranking the ground-truth next item AirPods at the Top-1 position in the recommendation list.
• Figure 2: The framework of our proposed SDIL framework. SDIL is composed of three main modules: (a) user Dynamic Interest Modeling module (DIM), (b) user Static Interest Modeling module (SIM), and (c) Next-item Prediction module. DIM captures a user's evolutionary dynamic interest by carefully modeling the time-sensitive positive and negative excitation of the user's historical interactions on the user's current interest. The SIM captures the user's relatively stable and high-level interest (e.g., interest in item category, brand) from the attribute information of interacted items. Finally, the prediction module well integrates both dynamic and static interest to obtain more precise interest for next-item prediction.
• Figure 3: Ablation study on the model performance (HR@5 and NDCG@5) on different datasets.
• Figure 4: Embedding size setting's effect on the model performance. (HR@5 and NDCG@10).
• Figure 5: Different transformer layers setting's effect on the model performance. (HR@10 and NDCG@10).