Personalization Matters: Long-Horizon Conversation Agent with User-Centric Information in Online Shopping Interactions
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.
ProofPaper ↗
Key points
- 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.
- 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).
Sources (1)
- [1]Personalization Matters: Long-Horizon Conversation Agent with User-Centric Information in Online Shopping InteractionsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 07:07 AM
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.
Personalized conversational shopping requires maintaining preference consistency over multi-turn interactions, where users reveal constraints gradually.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 7, 2026[AINews] Claude Haiku 5.5 — better than GPT-6 Luna at the same pricing
- Oct 7, 2026Rethinking access control for RAG with Amazon Quick and Amazon Bedrock
- Oct 7, 2026Automate remediation post AWS DevOps Agent investigation
- Oct 6, 2026EmbeddingGemma 2: an open, lightweight multimodal embedding model
- Oct 6, 2026Manage Amazon SageMaker HyperPod Spaces directly from SageMaker Studio
- Oct 6, 2026Build a voice travel concierge with Amazon Bedrock AgentCore, Managed Knowledge Base and Nova Sonic