PRISM: Personalized Recommendation via Information Synergy Module

2026-01-16

AI Summary

This work addresses the limitations of existing multimodal sequential recommendation methods, which struggle to effectively model cross-modal synergistic information and rely on fixed fusion weights that lack personalization. To overcome these issues, we propose PRISM, a novel framework that, for the first time, explicitly disentangles multimodal sequences into unique, redundant, and synergistic components from an information-theoretic perspective. PRISM further introduces a plug-and-play adaptive fusion mechanism that dynamically weights and integrates multimodal signals according to individual user preferences. The framework is compatible with various backbone sequential recommendation models and demonstrates consistent and significant performance improvements across four benchmark datasets and three distinct backbone architectures, thereby validating its effectiveness and generalizability.

Technology Category

• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Multi-modal Vision

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Multimodal sequential recommendation (MSR) leverages diverse item modalities to improve recommendation accuracy, while achieving effective and adaptive fusion remains challenging. Existing MSR models often overlook synergistic information that emerges only through modality combinations. Moreover, they typically assume a fixed importance for different modality interactions across users. To address these limitations, we propose \textbf{P}ersonalized \textbf{R}ecommend-ation via \textbf{I}nformation \textbf{S}ynergy \textbf{M}odule (PRISM), a plug-and-play framework for sequential recommendation (SR). PRISM explicitly decomposes multimodal information into unique, redundant, and synergistic components through an Interaction Expert Layer and dynamically weights them via an Adaptive Fusion Layer guided by user preferences. This information-theoretic design enables fine-grained disentanglement and personalized fusion of multimodal signals. Extensive experiments on four datasets and three SR backbones demonstrate its effectiveness and versatility. The code is available at https://github.com/YutongLi2024/PRISM.

Problem

• multimodal sequential recommendation
• information synergy
• adaptive fusion
• modality interaction
• personalized recommendation

Innovation

• multimodal sequential recommendation
• information synergy
• adaptive fusion
• personalized recommendation
• information-theoretic decomposition