arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2512.05065cs.CR
收藏

通过逻辑蕴含个性化代理隐私决策

Personalizing Agent Privacy Decisions via Logical Entailment

University of Southern California(南加州大学);Google Research(谷歌研究院)

James Flemings, Ren Yi, Octavian Suciu, Kassem Fawaz, Murali Annavaram, Marco Gruteser

AI总结:

本文提出ARIEL框架,结合LLM与规则逻辑,实现结构化个性化隐私决策,实验显示其在适当判断的F1误差率上降低40.6%。

摘要:

大型语言模型(LLM)代理越来越多地执行需要访问用户数据的任务,引发了关于适当数据披露的担忧。我们证明仅依赖LLM进行数据共享决策是不足的。通过提示LLM使用通用隐私规范无法捕捉个体用户的隐私偏好,而通过上下文学习(ICL)提供先前用户数据共享决策会导致不可靠和不透明的推理。为解决这些限制,我们提出了ARIEL(具有个性化蕴含逻辑的代理推理)框架,结合LLM与基于规则的逻辑,以实现结构化、个性化的隐私推理。ARIEL的核心机制确定用户在数据共享请求上的先前决定是否逻辑蕴含新请求的相同决定。使用先进模型和公共数据集的实验评估显示,与基于ICL的标准推理相比,ARIEL将适当判断的F1误差率降低了40.6%,表明其在用户会批准数据共享的请求中能够正确判断。这些结果表明,将LLM与逻辑蕴含结合提供了一种有效且可解释的方法,用于自动化个性化隐私决策。

英文摘要:

Personal large language model (LLM) agents increasingly perform tasks that require access to user data, raising concerns about appropriate data disclosure. We show that relying solely on LLMs to make data-sharing decisions is insufficient. Prompting LLMs with general privacy norms fails to capture individual users' privacy preferences, while providing prior user data-sharing decisions through in-context learning (ICL) leads to unreliable and opaque reasoning. To address these limitations, we propose ARIEL (Agentic Reasoning with Individualized Entailment Logic), a framework that combines LLMs with rule-based logic to enable structured, personalized privacy reasoning. The core mechanism of ARIEL determines whether a user's prior decision on a data-sharing request $\textit{logically entails}$ the same decision for a new request. Experimental evaluations using advanced models and public datasets show that ARIEL reduces the F1 error rate for appropriate judgments by $\textbf{40.6%}$ compared to standard ICL-based reasoning, indicating that ARIEL is effective at correctly judging requests where the user would approve data sharing. These results demonstrate that integrating LLMs with logical entailment provides an effective and interpretable approach for automating personalized privacy decisions.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑