Prompt Tuning as User Inherent Profile Inference Machine

TL;DR

This paper introduces UserIP-Tuning, a prompt-based method for inferring user profiles in recommender systems using LLMs, which enhances efficiency, reduces noise, and outperforms existing algorithms in real-world deployment.

Contribution

It proposes a novel prompt-tuning approach with EM inference and profile quantization to improve user profile inference in LLM-based recommender systems.

Findings

Outperforms state-of-the-art recommendation algorithms.

Deployed in Huawei AppGallery, serving 2 million daily users.

Improves recommendation robustness and transferability.

Abstract

Large Language Models (LLMs) have exhibited significant promise in recommender systems by empowering user profiles with their extensive world knowledge and superior reasoning capabilities. However, LLMs face challenges like unstable instruction compliance, modality gaps, and high inference latency, leading to textual noise and limiting their effectiveness in recommender systems. To address these challenges, we propose UserIP-Tuning, which uses prompt-tuning to infer user profiles. It integrates the causal relationship between user profiles and behavior sequences into LLMs' prompts. It employs Expectation Maximization (EM) to infer the embedded latent profile, minimizing textual noise by fixing the prompt template. Furthermore, a profile quantization codebook bridges the modality gap by categorizing profile embeddings into collaborative IDs pre-stored for online deployment. This improves time efficiency and reduces memory usage. Experiments show that UserIP-Tuning outperforms state-of-the-art recommendation algorithms. An industry application confirms its effectiveness, robustness, and transferability. The presented solution has been deployed in Huawei AppGallery's Explore page since May 2025, serving 2 million daily active users, delivering significant improvements in real-world recommendation scenarios. The code is publicly available for replication at https://github.com/Applied-Machine-Learning-Lab/UserIP-Tuning.