Image Fusion for Cross-Domain Sequential Recommendation

2024-12-31

AI Summary

Cross-domain sequential recommendation (CDSR) faces the key challenge of jointly modeling intra-sequence and inter-sequence item interactions to accurately capture users’ dynamic cross-domain preferences. To address this, we propose the first CDSR framework incorporating a frozen CLIP image encoder, introducing an image-enhanced multi-attention mechanism. Specifically, fine-grained visual semantics of items are captured via CLIP’s visual embeddings; image–text feature fusion enables cross-domain visual semantic alignment; and a hierarchical cross-domain attention module unifies modeling of intra-sequence local dependencies and inter-sequence cross-domain associations. Extensive experiments on four reconstructed e-commerce datasets demonstrate that our method significantly outperforms state-of-the-art approaches, achieving a 12.6% improvement in Recall@10. These results validate both the effectiveness and advancement of leveraging visual information for modeling cross-domain user preferences.

Technology Category

• Computer Vision: Multi-modal Vision
• Application Domains: Humanities & Computational Social Science
• Machine Learning: Learning Preferences or Rankings

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems

Abstract

Cross-Domain Sequential Recommendation (CDSR) aims to predict future user interactions based on historical interactions across multiple domains. The key challenge in CDSR is effectively capturing cross-domain user preferences by fully leveraging both intra-sequence and inter-sequence item interactions. In this paper, we propose a novel method, Image Fusion for Cross-Domain Sequential Recommendation (IFCDSR), which incorporates item image information to better capture visual preferences. Our approach integrates a frozen CLIP model to generate image embeddings, enriching original item embeddings with visual data from both intra-sequence and inter-sequence interactions. Additionally, we employ a multiple attention layer to capture cross-domain interests, enabling joint learning of single-domain and cross-domain user preferences. To validate the effectiveness of IFCDSR, we re-partitioned four e-commerce datasets and conducted extensive experiments. Results demonstrate that IFCDSR significantly outperforms existing methods.

Problem

• Predict future user interactions across domains.
• Capture cross-domain user preferences effectively.
• Enhance recommendation with item image information.

Innovation

• Integrates CLIP model for image embeddings
• Uses multiple attention for cross-domain interests
• Enhances item embeddings with visual data