Enregistré dans:
Détails bibliographiques
Auteurs principaux: Li, Yingxin, Zhao, Jianbo, Ren, Xueyu, Tang, Jie, You, Wangjie, Chen, Xu, Zhou, Kan, Feng, Chao, Ran, Jiao, Meng, Yuan, Wang, Zhi
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2509.18864
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866916964313595904
author Li, Yingxin
Zhao, Jianbo
Ren, Xueyu
Tang, Jie
You, Wangjie
Chen, Xu
Zhou, Kan
Feng, Chao
Ran, Jiao
Meng, Yuan
Wang, Zhi
author_facet Li, Yingxin
Zhao, Jianbo
Ren, Xueyu
Tang, Jie
You, Wangjie
Chen, Xu
Zhou, Kan
Feng, Chao
Ran, Jiao
Meng, Yuan
Wang, Zhi
contents User profiling, as a core technique for user understanding, aims to infer structural attributes from user information. Large Language Models (LLMs) provide a promising avenue for user profiling, yet the progress is hindered by the lack of comprehensive benchmarks. To bridge this gap, we propose ProfileBench, an industrial benchmark derived from a real-world video platform, encompassing heterogeneous user data and a well-structured profiling taxonomy. However, the profiling task remains challenging due to the difficulty of collecting large-scale ground-truth labels, and the heterogeneous and noisy user information can compromise the reliability of LLMs. To approach label-free and reliable user profiling, we propose a Confidence-driven Profile reasoning framework Conf-Profile, featuring a two-stage paradigm. We first synthesize high-quality labels by leveraging advanced LLMs with confidence hints, followed by confidence-weighted voting for accuracy improvement and confidence calibration for a balanced distribution. The multiple profile results, rationales, and confidence scores are aggregated and distilled into a lightweight LLM. We further enhance the reasoning ability via confidence-guided unsupervised reinforcement learning, which exploits confidence for difficulty filtering, quasi-ground truth voting, and reward weighting. Experimental results demonstrate that Conf-Profile delivers substantial performance through the two-stage training, improving F1 by 13.97 on Qwen3-8B.
format Preprint
id arxiv_https___arxiv_org_abs_2509_18864
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Conf-Profile: A Confidence-Driven Reasoning Paradigm for Label-Free User Profiling
Li, Yingxin
Zhao, Jianbo
Ren, Xueyu
Tang, Jie
You, Wangjie
Chen, Xu
Zhou, Kan
Feng, Chao
Ran, Jiao
Meng, Yuan
Wang, Zhi
Artificial Intelligence
User profiling, as a core technique for user understanding, aims to infer structural attributes from user information. Large Language Models (LLMs) provide a promising avenue for user profiling, yet the progress is hindered by the lack of comprehensive benchmarks. To bridge this gap, we propose ProfileBench, an industrial benchmark derived from a real-world video platform, encompassing heterogeneous user data and a well-structured profiling taxonomy. However, the profiling task remains challenging due to the difficulty of collecting large-scale ground-truth labels, and the heterogeneous and noisy user information can compromise the reliability of LLMs. To approach label-free and reliable user profiling, we propose a Confidence-driven Profile reasoning framework Conf-Profile, featuring a two-stage paradigm. We first synthesize high-quality labels by leveraging advanced LLMs with confidence hints, followed by confidence-weighted voting for accuracy improvement and confidence calibration for a balanced distribution. The multiple profile results, rationales, and confidence scores are aggregated and distilled into a lightweight LLM. We further enhance the reasoning ability via confidence-guided unsupervised reinforcement learning, which exploits confidence for difficulty filtering, quasi-ground truth voting, and reward weighting. Experimental results demonstrate that Conf-Profile delivers substantial performance through the two-stage training, improving F1 by 13.97 on Qwen3-8B.
title Conf-Profile: A Confidence-Driven Reasoning Paradigm for Label-Free User Profiling
topic Artificial Intelligence
url https://arxiv.org/abs/2509.18864