Preference Discerning with LLM-Enhanced Generative Retrieval

2024-12-11

Trans. Mach. Learn. Res.

AI Summary

Existing sequential recommendation models leverage LLM-inferred user preferences only during training, relying solely on historical interactions during inference—leading to insufficient personalization, poor dynamic adaptability, and echo-chamber effects. Method: This paper introduces “preference identification” as a novel paradigm: (1) formally defining the capability, (2) constructing a comprehensive benchmark covering preference guidance and sentiment alignment, and (3) proposing Mender—a multimodal preference identifier that enables explicit preference modeling and real-time human instruction responsiveness via LLM-driven generative retrieval, context-conditioned sequence generation, and interpretable preference injection. Contribution/Results: Mender achieves state-of-the-art performance on our newly established benchmark and significantly advances zero-shot, preference-controllable recommendation—demonstrating superior generalization, interpretability, and adaptability to evolving user intent.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Knowledge Representation and Reasoning: Preferences
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Sequential recommendation systems aim to provide personalized recommendations for users based on their interaction history. To achieve this, they often incorporate auxiliary information, such as textual descriptions of items and auxiliary tasks, like predicting user preferences and intent. Despite numerous efforts to enhance these models, they still suffer from limited personalization. To address this issue, we propose a new paradigm, which we term preference discerning. In preference dscerning, we explicitly condition a generative sequential recommendation system on user preferences within its context. To this end, we generate user preferences using Large Language Models (LLMs) based on user reviews and item-specific data. To evaluate preference discerning capabilities of sequential recommendation systems, we introduce a novel benchmark that provides a holistic evaluation across various scenarios, including preference steering and sentiment following. We assess current state-of-the-art methods using our benchmark and show that they struggle to accurately discern user preferences. Therefore, we propose a new method named Mender ($ extbf{M}$ultimodal Prefer$ extbf{en}$ce $ extbf{d}$iscern$ extbf{er}$), which improves upon existing methods and achieves state-of-the-art performance on our benchmark. Our results show that Mender can be effectively guided by human preferences even though they have not been observed during training, paving the way toward more personalized sequential recommendation systems. We will open-source the code and benchmarks upon publication.

Problem

• Addresses limited dynamic adaptation to changing user preferences
• Mitigates echo chambers in sequential recommendation systems
• Enhances recommendation flexibility using natural language preferences

Innovation

• Generative model conditions on natural language preferences
• Multimodal preference discerner adapts to evolving user tastes
• LLM-enhanced retrieval dynamically steers recommendations via sentiment