大语言模型在档案开放审核中的应用框架与路径探究  

Research on the Application Strategies of Large Language Models in the Open Audit of Archival Records

作　　者：刘越男[1,2,3] 张茜雅 杨建梁 LIU Yuenan;ZHANG Xiya;YANG Jianlianog(School of Information Resource Management,Renmin University of China,Beijing 100872,China;Electronic Records Management Research Center,Renmin University of China,Beijing 100872,China;Key Laboratory of the National Archives Administration for Multimodal Archives Protection and Development,Beijing 100872,China)

机构地区：[1]中国人民大学信息资源管理学院,北京100872 [2]中国人民大学电子文件管理研究中心,北京100872 [3]多模态档案保护与开发国家档案局重点实验室,北京100872

出　　处：《档案学通讯》2025年第2期31-38,共8页Archives Science Bulletin

基　　金：国家社会科学基金青年项目“基于人工智能的分层多粒度档案开放鉴定研究”(21CTQ030)。

摘　　要：大语言模型的出现掀起了自然语言处理领域的一场新科技革命,也为档案领域的智能化转型提供了新的机遇。面向当前较为急迫的档案开放审核需求,探讨大语言模型在其中的应用策略具有重要的现实意义。现有智能档案开放审核面临智能开放审核规则构建的依据不清、训练数据的可获取性不强、审核结论的可解释性不足等问题,而大语言模型则在语义理解、无监督学习、可解释性等方面具备技术优势。提出基于大语言模型的档案开放审核任务框架,并设计具体实施的技术路径,强调通过提示工程引导大模型完成复杂多样的档案开放审核任务。The emergence of large language models has triggered a technological revolution in the field of natural language processing,offering new opportunities for the intelligent transformation of the archival domain.Given the pressing need for the open audit of archival records,exploring the application strategies of large language models in this context holds significant practical value.Existing intelligent open audit of archival records faces issues such as unclear basis for constructing intelligent audit rules,limited availability of training data,and insufficient interpretability of audit conclusions.However,large language models possess technical advantages in semantic understanding,unsupervised learning,and interpretability.This study proposes a task framework for the open audit of archival records based on large language models and designs specific technical implementation paths,emphasizing the use of prompt engineering to guide large models in performing complex and diverse open audit tasks.

关 键 词：大模型 人工智能 无监督学习 档案开放利用 

分 类 号：G270.7[文化科学—档案学]