Llm based recommendation improves user preference understanding and retrieval
P3Rec: Distilling Prior--Posterior Preference Reasoning for LLM-based Recommendation
Information Retrieval
Summary
Recommender systems try to figure out what you like to suggest things you might enjoy, but it can be tricky to balance long-term interests with what you want right now. This work introduces P3Rec, a new method that uses large language models to learn both your stable past preferences and your current detailed interests. It combines these two views to create a better user profile and then adjusts the recommendations based on how focused or varied your past interests are. The approach was tested on public datasets, showing that it helps make recommendations more accurate and efficient.
What this means in practice
- •For recommendation system engineers: Improve recommendation accuracy by integrating both stable and current user preferences distilled from large language models into lightweight retrieval models.
- •For ecommerce platform developers: Enhance product suggestion relevance by calibrating user profiles based on interest diversity before retrieval, capturing nuanced preferences more effectively.
Authors
Jinfei Chen, Weihai Lu, Jiawei Cheng
Abstract
Large language models (LLMs) exhibit strong semantic understanding and preference reasoning capabilities, offering new opportunities for user modeling in recommender systems. Existing LLM-as-Enhancer methods typically distill LLM-derived preference knowledge into lightweight recommenders to avoid costly online LLM inference. However, they often construct distillation knowledge from only one perspective. Prior preference captures users' stable and consistent interests but provides limited guidance for the current decision, whereas posterior preference reveals target-relevant fine-grained interests but may rely excessively on target clues. To address these limitations, we propose P$^3$Rec, a framework that jointly extracts and internalizes complementary prior and posterior preference reasoning knowledge. Specifically, P$^3$Rec first derives target-agnostic prior preferences and target-conditioned posterior preferences from the user side, while further extracting item-centric preference representations from item semantics and predecessor interactions. It then progressively internalizes prior and posterior knowledge into behavioral representations through prior preference absorption and posterior-guided preference distillation. Since the resulting comprehensive preference representation may not always provide an equally decisive retrieval direction, P$^3$Rec further characterizes historical interest dispersion with interest entropy and adaptively calibrates the user representation before contrastive retrieval optimization. In this way, P$^3$Rec achieves more complete preference reasoning while preserving efficient recommendation. Extensive experiments on multiple public datasets demonstrate its effectiveness.