Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Unified Framework for Adaptive Representation Enhancement and Inversed Learning in Cross-Domain Recommendation

Luankang Zhang, Hao Wang, Suojuan Zhang, Mingjia Yin, Yongqiang Han, Jiaqing Zhang, Defu Lian, Enhong Chen

TL;DR

The paper tackles data sparsity and cold-start in cross-domain recommendations by introducing AREIL, a unified framework that disentangles user representations into domain-shared and domain-specific factors and enhances them via adaptive intra- and inter-domain mechanisms. It fuses a LightGCN-based intra-domain encoder with a self-attention-driven inter-domain transfer, and enforces meaningful disentanglement with an Inversed Representation Learning Module that employs domain classifiers and a gradient reversal layer within a multi-task objective. Empirical results on three Amazon cross-domain datasets show AREIL consistently outperforms state-of-the-art baselines, with ablations confirming the contributions of graph-based enhancement and inversed learning. The work provides a principled approach to robust information transfer across domains and offers practical implications for improving recommender systems under data sparsity and cold-start constraints.

Abstract

Cross-domain recommendation (CDR), aiming to extract and transfer knowledge across domains, has attracted wide attention for its efficacy in addressing data sparsity and cold-start problems. Despite significant advances in representation disentanglement to capture diverse user preferences, existing methods usually neglect representation enhancement and lack rigorous decoupling constraints, thereby limiting the transfer of relevant information. To this end, we propose a Unified Framework for Adaptive Representation Enhancement and Inversed Learning in Cross-Domain Recommendation (AREIL). Specifically, we first divide user embeddings into domain-shared and domain-specific components to disentangle mixed user preferences. Then, we incorporate intra-domain and inter-domain information to adaptively enhance the ability of user representations. In particular, we propose a graph convolution module to capture high-order information, and a self-attention module to reveal inter-domain correlations and accomplish adaptive fusion. Next, we adopt domain classifiers and gradient reversal layers to achieve inversed representation learning in a unified framework. Finally, we employ a cross-entropy loss for measuring recommendation performance and jointly optimize the entire framework via multi-task learning. Extensive experiments on multiple datasets validate the substantial improvement in the recommendation performance of AREIL. Moreover, ablation studies and representation visualizations further illustrate the effectiveness of adaptive enhancement and inversed learning in CDR.

A Unified Framework for Adaptive Representation Enhancement and Inversed Learning in Cross-Domain Recommendation

TL;DR

The paper tackles data sparsity and cold-start in cross-domain recommendations by introducing AREIL, a unified framework that disentangles user representations into domain-shared and domain-specific factors and enhances them via adaptive intra- and inter-domain mechanisms. It fuses a LightGCN-based intra-domain encoder with a self-attention-driven inter-domain transfer, and enforces meaningful disentanglement with an Inversed Representation Learning Module that employs domain classifiers and a gradient reversal layer within a multi-task objective. Empirical results on three Amazon cross-domain datasets show AREIL consistently outperforms state-of-the-art baselines, with ablations confirming the contributions of graph-based enhancement and inversed learning. The work provides a principled approach to robust information transfer across domains and offers practical implications for improving recommender systems under data sparsity and cold-start constraints.

Abstract

Cross-domain recommendation (CDR), aiming to extract and transfer knowledge across domains, has attracted wide attention for its efficacy in addressing data sparsity and cold-start problems. Despite significant advances in representation disentanglement to capture diverse user preferences, existing methods usually neglect representation enhancement and lack rigorous decoupling constraints, thereby limiting the transfer of relevant information. To this end, we propose a Unified Framework for Adaptive Representation Enhancement and Inversed Learning in Cross-Domain Recommendation (AREIL). Specifically, we first divide user embeddings into domain-shared and domain-specific components to disentangle mixed user preferences. Then, we incorporate intra-domain and inter-domain information to adaptively enhance the ability of user representations. In particular, we propose a graph convolution module to capture high-order information, and a self-attention module to reveal inter-domain correlations and accomplish adaptive fusion. Next, we adopt domain classifiers and gradient reversal layers to achieve inversed representation learning in a unified framework. Finally, we employ a cross-entropy loss for measuring recommendation performance and jointly optimize the entire framework via multi-task learning. Extensive experiments on multiple datasets validate the substantial improvement in the recommendation performance of AREIL. Moreover, ablation studies and representation visualizations further illustrate the effectiveness of adaptive enhancement and inversed learning in CDR.
Paper Structure (27 sections, 12 equations, 4 figures, 3 tables)

This paper contains 27 sections, 12 equations, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Cross-Domain Recommendation
  4. Disentanglement Learning in CDR
  5. Problem Definition
  6. Methodology
  7. Disentanglement-based Embedding Layer
  8. Adaptive Representation Enhancement Module
  9. Intra-domain Enhancement.
  10. Inter-domain Enhancement.
  11. Inversed Representation Learning Module
  12. Prediction Layer and Multi-task Learning
  13. Experiments
  14. Experiments Settings
  15. Datasets.
  16. ...and 12 more sections

Figures (4)

  • Figure 1: The AREIL framework for adaptive representation enhancement and inversed representation learning.
  • Figure 2: Impact of the classification loss weight $\lambda_1$ in Sport(left) & Phone(right)
  • Figure 3: Impact of the fusion controlling weight $\gamma_s$ in Sport(left) & Phone(right)
  • Figure 4: Visualization of Disentangled Embeddings: (a) illustrates disentanglement in Elec&Cloth; (b) displays the distribution of domain-shared user embeddings.