Go to ACL ARR 2025 February homepagePREMIUM: LLM Personalization with Individual-level Preference Feedback
Abstract: With an increasing demand for LLM personalization, various methods have been developed to deliver customized LLM experiences. However, most existing methods are not readily locally deployable, limited by the compute cost, privacy risks, and an inability to adapt to dynamic user preferences. Here, we propose utilizing a tag system to efficiently characterize user profiles, drawing inspiration from personality typology and recommendation systems. Based on the observation, we present a locally deployable LLM-agnostic personalization framework: $\textbf{PREMIUM}$, which obtains individual-level feedback by having users rank responses and continuously self-iterates optimization during the interaction process. Notably, a variant of PREMIUM, PREMIUM-Embed, can effectively capture user preferences while being deployable with laptop-level resources. Extensive experiments validate that PREMIUM remarkably outperforms various baselines, achieving a 15%-50% higher accuracy and a 2.5%-35% higher win rate on Ranking-TAGER, a valuable evaluation protocol for LLM personalization that we propose, as well as a 3%-13% higher accuracy and a 2%-7.5% higher F1 Score on LaMP-2. More importantly, we further demonstrate that PREMIUM can develop an effective strategy with minimal interactive data, adapt to dynamic user preferences, and demonstrate excellent scalability in both scale and functionality.
Paper Type: Long
Research Area: Human-Centered NLP
Research Area Keywords: human-AI interaction; user-centered design; prompting; phrase/sentence embedding; human-subject application-grounded evaluations
Contribution Types: NLP engineering experiment, Approaches to low-resource settings, Data resources
Languages Studied: English
Submission Number: 6314
Loading