| 白佳欣,谭玉珊,胡文鹏.面向情报研究的多智能体协同处理方法[J].情报工程,2026,(1):003-016 |
| 面向情报研究的多智能体协同处理方法 |
| A Multi-Agent Collaborative Processing Method for Intelligence Research |
| |
| DOI： |
| 中文关键词: 情报智能处理；AI 智能体；大语言模型；提示工程；人在回路 |
| 英文关键词: Intelligent Intelligence Processing; AI Agents; Large Language Models; Prompt Engineering; Human-in-the-Loop |
| 基金项目: |
| 作者 | 单位 | | 白佳欣 | 军事科学院军事科学信息研究中心 北京 100142 | | 谭玉珊 | 军事科学院军事科学信息研究中心 北京 100142 | | 胡文鹏 | 军事科学院军事科学信息研究中心 北京 100142 |
|
| 摘要点击次数: 3 |
| 全文下载次数: 2 |
| 中文摘要: |
| [目的/意义]情报研究面临着大量的情报处理需求。随着大数据与人工智能技术的发展，借助算法和工具开展情报处理，实现高价值情报的深度挖掘成为必然。然而，情报研究人员普遍缺乏技术基础，算法与代码的复杂性增加了技术赋能情报研究的难度。[方法/过程]提出了一种面向情报研究的多智能体协同处理方法，通过构建一个多智能体协同框架，实现自动解析用户指令和需求，利用大模型强大的理解和生成能力，自动解析可用算法模型的README文件及代码，自动执行算法并反馈情报处理结果。[结果/ 结论]情报研究人员仅需通过简单的界面操作或自然语言交互，即可轻松实现对各种情报处理工具算法的调用，显著提升情报处理效率和智能化水平，大大降低情报处理对情报人员的技术要求，框架技术路线具有广泛的普适性应用价值。 |
| 英文摘要: |
| [Objective/Significance] Intelligence research faces the growing demand for processing large volumes of intelligence data. With the advancement of big data and artificial intelligence technologies, leveraging algorithms and tools to process intelligence and achieve in-depth mining of high-value intelligence has become imperative. However, intelligence researchers generally lack a technical foundation, and the complexity of algorithms and code increases the difficulty of empowering intelligence research through technology. [Methods/Processes] To address this issue, this paper proposes a multi-agent collaborative processing method for intelligence research. By constructing a multi-agent collaborative framework, it enables automatic parsing of user instructions and requirements, leverages the powerful comprehension and generation capabilities of large language models (LLMs), automatically analyzes README files and code of available algorithmic models, and autonomously executes algorithms to return intelligence processing results. [Results/Conclusions] Intelligence researchers can effortlessly invoke various intelligence processing tools and algorithms through simple interface operations or natural language interactions. This significantly enhances the efficiency and intelligence level of intelligence processing, greatly reduces the technical skill requirements for intelligence personnel. The framework’s technical approach offers broad applicability and universal value. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
