Personalization Matters: Long-Horizon Conversation Agent with User-Centric Information in Online Shopping Interactions

2026-10-08

AI Summary

This study addresses the challenges of maintaining preference consistency and the limitations of static user profiles in controlling long-horizon interactions within multi-turn dialogues for online shopping. To this end, we propose a multi-agent multimodal Retrieval-Augmented Generation (RAG) framework. By leveraging a role decomposition mechanism and a user-centric retrieval strategy, the framework integrates product metadata, reviews, and image descriptions to enable dynamic state tracking and personalized reasoning. Furthermore, we construct a trajectory-level evaluation protocol encompassing global preference consistency. Experimental results demonstrate that the proposed method significantly outperforms baseline models on automatic metrics and achieves a score of 4.60 in real-user studies, effectively validating its advantages in enhancing personalized experiences during long-horizon interactions.

Abstract

Personalized conversational shopping requires maintaining preference consistency over multi-turn interactions, where users reveal constraints gradually. Existing approaches often rely on static profiles and do not explicitly control long-horizon interaction behavior. We propose a multi-agent, multimodal Retrieval-Augmented Generation (RAG) framework that decomposes dialogue state tracking, recommendation retrieval, preference-aware reasoning, and response generation, while integrating product metadata, product reviews, image-derived descriptions, and user historical reviews. To evaluate interaction-level quality, we adopt a trajectory-level protocol with four dimensions: Global Preference Consistency, Cumulative Information Synthesis, Interaction Trajectory, and Tone Consistency. On an Amazon Reviews 2023 benchmark, retrieval-enabled variants outperform a no-RAG baseline on automatic trajectory metrics (average 4.82 vs. 3.74). In a small real-user study ($n{=}5$), the Full variant achieves the highest mean overall rating (4.60 vs. 2.20 for Baseline), providing exploratory evidence that role decomposition plus user-centric retrieval improves perceived personalization.\footnote{Code and dataset are available at: https://github.com/RenaGao/Multimodel_RAG_Indexing

Problem

• Personalized conversational shopping
• Long-horizon conversation
• Preference consistency
• User-centric information
• Multi-turn interactions

Innovation

• Multi-agent RAG
• Multimodal Retrieval-Augmented Generation
• Long-Horizon Conversation
• Personalized Recommendation
• Trajectory-level Evaluation