Multi-agents based User Values Mining for Recommendation

TL;DR

The paper tackles the instability of recommender systems by grounding recommendations in users' long-term values. It introduces ZOOM, a zero-shot, multi-LLM framework with evaluators and supervisors to mine user values from historical interactions using Schwartz's ten universal values, aided by text summarization to handle long histories and debate to reduce hallucinations. It demonstrates two fusion strategies—direct concatenation and a contrastive learning auxiliary task—for integrating mined values into state-of-the-art, language-model-based recommenders, with CL delivering the strongest gains. Extensive experiments on PENS and MovieLens-1M show improved recommendation performance and high alignment between mined values and human judgments, validating both the mining framework and the integration approach. The work advances practical, stable, and more personalized recommendations by leveraging structured value information extracted from user history.

Abstract

Recommender systems have rapidly evolved and become integral to many online services. However, existing systems sometimes produce unstable and unsatisfactory recommendations that fail to align with users' fundamental and long-term preferences. This is because they primarily focus on extracting shallow and short-term interests from user behavior data, which is inherently dynamic and challenging to model. Unlike these transient interests, user values are more stable and play a crucial role in shaping user behaviors, such as purchasing items and consuming content. Incorporating user values into recommender systems can help stabilize recommendation performance and ensure results better reflect users' latent preferences. However, acquiring user values is typically difficult and costly. To address this challenge, we leverage the strong language understanding, zero-shot inference, and generalization capabilities of Large Language Models (LLMs) to extract user values from users' historical interactions. Unfortunately, direct extraction using LLMs presents several challenges such as length constraints and hallucination. To overcome these issues, we propose ZOOM, a zero-shot multi-LLM collaborative framework for effective and accurate user value extraction. In ZOOM, we apply text summarization techniques to condense item content while preserving essential meaning. To mitigate hallucinations, ZOOM introduces two specialized agent roles: evaluators and supervisors, to collaboratively generate accurate user values. Extensive experiments on two widely used recommendation datasets with two state-of-the-art recommendation models demonstrate the effectiveness and generalization of our framework in automatic user value mining and recommendation performance improvement.

Figures (10)

• Figure 1: ZOOM Overview
• Figure 2: Overview of Schwartz's Theory of Basic Values
• Figure 3: Text Summarization Prompt: $\mathcal{P}_{sum}$
• Figure 4: Evaluator Profile Prompt: $\mathcal{P}_{eval}$
• Figure 5: Supervisor Review Prompt: $\mathcal{P}_{review}$