Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Dual Contrastive Transformer for Hierarchical Preference Modeling in Sequential Recommendation

Chengkai Huang, Shoujin Wang, Xianzhi Wang, Lina Yao

TL;DR

This paper proposes a novel hierarchical preference modeling framework to substantially model the complex low- and high-level preference dynamics for accurate sequential recommendation and demonstrates both the superiority of the proposed method over the state-of-the-art ones and the rationality of the design.

Abstract

Sequential recommender systems (SRSs) aim to predict the subsequent items which may interest users via comprehensively modeling users' complex preference embedded in the sequence of user-item interactions. However, most of existing SRSs often model users' single low-level preference based on item ID information while ignoring the high-level preference revealed by item attribute information, such as item category. Furthermore, they often utilize limited sequence context information to predict the next item while overlooking richer inter-item semantic relations. To this end, in this paper, we proposed a novel hierarchical preference modeling framework to substantially model the complex low- and high-level preference dynamics for accurate sequential recommendation. Specifically, in the framework, a novel dual-transformer module and a novel dual contrastive learning scheme have been designed to discriminatively learn users' low- and high-level preference and to effectively enhance both low- and high-level preference learning respectively. In addition, a novel semantics-enhanced context embedding module has been devised to generate more informative context embedding for further improving the recommendation performance. Extensive experiments on six real-world datasets have demonstrated both the superiority of our proposed method over the state-of-the-art ones and the rationality of our design.

Dual Contrastive Transformer for Hierarchical Preference Modeling in Sequential Recommendation

TL;DR

This paper proposes a novel hierarchical preference modeling framework to substantially model the complex low- and high-level preference dynamics for accurate sequential recommendation and demonstrates both the superiority of the proposed method over the state-of-the-art ones and the rationality of the design.

Abstract

Sequential recommender systems (SRSs) aim to predict the subsequent items which may interest users via comprehensively modeling users' complex preference embedded in the sequence of user-item interactions. However, most of existing SRSs often model users' single low-level preference based on item ID information while ignoring the high-level preference revealed by item attribute information, such as item category. Furthermore, they often utilize limited sequence context information to predict the next item while overlooking richer inter-item semantic relations. To this end, in this paper, we proposed a novel hierarchical preference modeling framework to substantially model the complex low- and high-level preference dynamics for accurate sequential recommendation. Specifically, in the framework, a novel dual-transformer module and a novel dual contrastive learning scheme have been designed to discriminatively learn users' low- and high-level preference and to effectively enhance both low- and high-level preference learning respectively. In addition, a novel semantics-enhanced context embedding module has been devised to generate more informative context embedding for further improving the recommendation performance. Extensive experiments on six real-world datasets have demonstrated both the superiority of our proposed method over the state-of-the-art ones and the rationality of our design.

Paper Structure

This paper contains 23 sections, 21 equations, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Sequential Recommendation
  4. Category-aware Preference Modeling
  5. Contrastive Sequential Recommendation
  6. Methodology
  7. Problem Formulation
  8. Framework Overview
  9. Embedding Layers
  10. Dual Transformer for Hierarchical Preference Modeling
  11. Semantics-enhanced Context Embedding Learning
  12. Dual Hierarchical Preference Contrastive Learning
  13. Training and Optimization
  14. Experiments
  15. Data Preparation
  16. ...and 8 more sections

Figures (5)

  • Figure 1: Alice's hierarchical preference dynamics through a sequence of purchased items. Alice's low-level preference indicated by item ID changes sharply while her high-level preference indicated by category changes smoothly.
  • Figure 2: (a)The overall framework of our Hierarchical Preference modeling (HPM), which is composed of three main components: Dual Transformer module for hierarchical preference modeling, Semantics-enhanced Context Embedding Learning (SCEL), and Dual-Contrastive Learning (DCL) scheme; (b) Semantics-enhanced Context Embedding Learning (SCEL) leverages the relations between context items and target items to enhance the representation of target item embedding.
  • Figure 3: Ablation study of our model (HR@5 and NDCG@5) (Upper left: Beauty, Upper right: Sports, Lower left: Cellphones, Lower right: Clothing).
  • Figure 4: Parameter setting's effect on the model performance. (HR@5) on Amazon Clothing dataset.
  • Figure 5: Parameter setting's effect on the model performance. (HR@5 and NDCG@5) on Amazon Cellphone dataset.