Hierarchical Multi-Persona Induction from User Behavioral Logs: Learning Evidence-Grounded and Truthful Personas

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Choi, Nayoung, Jeong, Haeyu, Kim, Changbong, Lim, Hongjun, Choi, Jinho D.
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913071226683392
author Choi, Nayoung
Jeong, Haeyu
Kim, Changbong
Lim, Hongjun
Choi, Jinho D.
author_facet Choi, Nayoung
Jeong, Haeyu
Kim, Changbong
Lim, Hongjun
Choi, Jinho D.
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2604_26120
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Hierarchical Multi-Persona Induction from User Behavioral Logs: Learning Evidence-Grounded and Truthful Personas
Choi, Nayoung
Jeong, Haeyu
Kim, Changbong
Lim, Hongjun
Choi, Jinho D.
Artificial Intelligence
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.
title Hierarchical Multi-Persona Induction from User Behavioral Logs: Learning Evidence-Grounded and Truthful Personas
topic Artificial Intelligence
url https://arxiv.org/abs/2604.26120