Hierarchical Multi-Persona Induction from User Behavioral Logs: Learning Evidence-Grounded and Truthful Personas
For user modeling practitioners, this work provides a method to generate more reliable and interpretable personas from noisy behavioral logs, addressing the lack of direct persona quality assurance in prior work.
The paper proposes a hierarchical framework for inducing multiple evidence-grounded and truthful personas from user behavioral logs, formulating persona induction as an optimization problem over quality metrics and training with groupwise DPO. Experiments show improved coherence, evidence grounding, and trustworthiness of personas, as well as better future interaction prediction.
Behavioral logs provide rich signals for user modeling, but are noisy and interleaved across diverse intents. Recent work uses LLMs to generate interpretable natural-language personas from user logs, yet evaluation often emphasizes downstream utility, providing limited assurance of persona quality itself. We propose a hierarchical framework that aggregates user actions into intent memories and induces multiple evidence-grounded personas by clustering and labeling these memories. We formulate persona induction as an optimization problem over persona quality-captured by cluster cohesion, persona-evidence alignment, and persona truthfulness-and train the persona model using a groupwise extension of Direct Preference Optimization (DPO). Experiments on a large-scale service log and two public datasets show that our method induces more coherent, evidence-grounded, and trustworthy personas, while also improving future interaction prediction.