Self-Supervised Representation Learning with ID-Content Modality Alignment for Sequential Recommendation

2025-10-12

AI Summary

To address the performance degradation of ID-based sequential recommendation (SR) under sparse interaction histories, this paper proposes a self-supervised representation learning framework that aligns ID and multimodal content modalities. Methodologically, it (1) disentangles collaborative dependencies from content dependencies; (2) introduces an LLM-driven sample construction mechanism and a two-stage training strategy; and (3) incorporates a hybrid-modal sequence decoder. By unifying supervised fine-tuning, contrastive learning, and self-supervised learning, the framework jointly optimizes ID embeddings and multimodal content representations, thereby bridging semantic gaps across modalities and jointly modeling behavioral and content preferences. Extensive experiments on four video streaming datasets demonstrate substantial improvements over state-of-the-art methods: +8.04% in NDCG@5 and +6.62% in NDCG@10.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Personalized content ranking and presentation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Sequential recommendation (SR) models often capture user preferences based on the historically interacted item IDs, which usually obtain sub-optimal performance when the interaction history is limited. Content-based sequential recommendation has recently emerged as a promising direction that exploits items' textual and visual features to enhance preference learning. However, there are still three key challenges: (i) how to reduce the semantic gap between different content modality representations; (ii) how to jointly model user behavior preferences and content preferences; and (iii) how to design an effective training strategy to align ID representations and content representations. To address these challenges, we propose a novel model, self-supervised representation learning with ID-Content modality alignment, named SICSRec. Firstly, we propose a LLM-driven sample construction method and develop a supervised fine-tuning approach to align item-level modality representations. Secondly, we design a novel Transformer-based sequential model, where an ID-modality sequence encoder captures user behavior preferences, a content-modality sequence encoder learns user content preferences, and a mix-modality sequence decoder grasps the intrinsic relationship between these two types of preferences. Thirdly, we propose a two-step training strategy with a content-aware contrastive learning task to align modality representations and ID representations, which decouples the training process of content modality dependency and item collaborative dependency. Extensive experiments conducted on four public video streaming datasets demonstrate our SICSRec outperforms the state-of-the-art ID-modality sequential recommenders and content-modality sequential recommenders by 8.04% on NDCG@5 and 6.62% on NDCD@10 on average, respectively.

Problem

• Aligning item-level modality representations to reduce semantic gaps
• Jointly modeling user behavior and content preferences in sequences
• Developing training strategies for ID-content representation alignment

Innovation

• LLM-driven sample construction aligns modality representations
• Transformer-based model captures behavior and content preferences
• Two-step training strategy with contrastive learning aligns representations