HyperTrace traces user preferences for better personalized responses
HyperTrace: Hypothesis-Based Preference Tracing for Online LLM Personalization
Computation and Language
Summary
Personalizing language models is hard because people’s likes and needs change and aren’t obvious right away. The authors created HyperTrace, which keeps track of what a user wants both in the short and long term using clear natural language ideas. It updates these ideas by watching how users interact and guessing their preferences without needing to retrain the model. This method helps the system respond more accurately to individuals over time.
What this means in practice
- •For conversational ai developers: Improve chatbot responses by online adapting to individual user preferences without retraining the language model.
- •For customer support platform teams: Enable support bots to remember and reflect user preferences across interactions for more consistent assistance.
Authors
Jianzhi Shen, Keyu Mao, Minghao Shao, Chuanyang Jin, Yusong Wang, Ailiang Lin, Kotaro Funakoshi, Manabu Okumura, Tianmin Shu, Muhammad Shafique
Abstract
Personalized language models aim to adapt responses to individual users, whose preferences are often latent and revealed gradually through interaction. Existing training-free methods rely on stored histories or retrieved memories, but they often struggle to reconcile long- term preferences with short-term topic-specific needs. To address this issue, we propose HyperTrace, a training-free framework that formulates online personalization as latent preference tracing. HyperTrace maintains interpretable natural-language hypotheses over short-term intent and long-term preferences, and updates them through an SMC-style reweight process using an LLM-based surrogate choice model. By updating these hypotheses across turns and sessions, HyperTrace enables personalization without parameter updates. Experiments on PRISM and PersonaMem-v2 show that HyperTrace improves response alignment, preference prediction, and profile consistency over strong online baselines, demonstrating the effectiveness of tracing latent user preferences for robust personalization. Code and scripts are available in the repository: https://github.com/jiseshen/HyperTrace.