大语言模型+检索增强方法的关键技术及其在情报任务中的应用流程  

Key Techniques of Large Language Models and Retrieval-Augmented Method with Their Application Process in Intelligence Tasks

作　　者：魏宏程 杨建林[1,2,3] Wei Hongcheng;Yang Jianlin(School of Information Management,Nanjing University,Jiangsu Nanjing 210023;National Security Development Research Institute of Nanjing University,Jiangsu Nanjing 210023;Key Laboratory of Data Engineering and Knowledge Services in Provincial Universities(Nanjing University),Jiangsu Nanjing 210023)

机构地区：[1]南京大学信息管理学院,江苏南京210023 [2]南京大学国家安全发展研究院,江苏南京210023 [3]江苏省数据工程与知识服务省高校重点实验室(南京大学),江苏南京210023

出　　处：《情报理论与实践》2025年第3期178-188,206,共12页Information Studies:Theory & Application

基　　金：国家社会科学基金重点项目“统筹科技发展和科技安全的情报学理论与方法研究”(项目编号:22ATQ007);南京大学新时代文科卓越研究计划“中长期研究专项”项目“‘数智赋能’背景下的情报学理论、方法与应用研究”(项目编号:010814914214);南京大学自主知识体系建构“引领工程”重大研究专项项目“中国自主的情报学知识体系构建研究”(项目编号:2024181013)的成果。

摘　　要：[目的/意义]探讨大语言模型+检索增强方法在情报实践中的应用路径,为利用其能力推动情报工作提供参考,进而提高科研情报工作者的效率。[方法/过程]梳理大语言模型与检索增强方法在情报实践中的应用探索,从敏感数据保护、检索方式、增强模式和提示工程4个部分讨论大模型+检索增强方法的关键技术,并阐述大语言模型结合检索增强方法在情报实践中的赋能环节以及在情报任务中的应用流程。[结果/结论]大语言模型+检索增强方法颠覆了传统情报研究范式,在情报环节与情报任务中发挥着重要作用。因此,应当顺应时代发展与战略需求,将大模型+检索增强方法应用于情报实践,开展高效的情报工作。[Purpose/significance]The paper aims to explore application pathways of the Large Language Model and retrievalaugmented method in intelligence practice,providing the guidance for leveraging their capabilities to advance intelligence tasks,thereby improving the efficiency of intelligence researchers.[Method/process]In this paper,the applications of the Large Language Model and retrieval augmentation method in intelligence practice are reviewed,and the key techniques of the two methods are discussed in four components,i.e.,sensitive data protections,retrieval methods,augmentation patterns,and prompt engineering.In addition,the paper describes the aspects of the Large Language Model and retrieval augmentation method to empower intelligence practice and the process of applying them to intelligence tasks.[Result/conclusion]The Large Language Model and retrieval augmentation method have reversed the traditional intelligence research paradigm,and play an important role in the intelligence process and tasks.Therefore,we should follow the development of the times and strategic needs,and apply the Large Language Model and retrieval augmentation method in the intelligence practice to carry out efficient intelligence work.

关 键 词：情报实践 情报方法 大语言模型 检索增强 敏感数据 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] G350[自动化与计算机技术—控制科学与工程]