图书情报知识 ›› 2025, Vol. 42 ›› Issue (4): 88-101.doi: 10.13366/j.dik.2025.040.088

• 学术聚焦(2)·生成式 AI 大模型结合外部工具和知识库的场景化应用 • 上一篇    下一篇

生成式AI大模型结合知识库与AI Agent开展知识挖掘的探析

赵浜1, 曹树金2   

  1. 1.中山大学信息管理学院,广州,510006;
    2.山东理工大学信息管理学院,淄博,255000
  • 出版日期:2025-07-10 发布日期:2025-08-16
  • 通讯作者: 曹树金(ORCID:0000-0003-1855-4522),博士,教授,研究方向:信息组织与信息检索,Email:caosj@mail.sysu.edu.cn。
  • 作者简介:赵浜(ORCID:0000-0001-9618-1042),博士研究生,研究方向:知识组织与文本挖掘,Email:zhaob55@mail2.sysu.edu.cn。
  • 基金资助:
    本文系国家社会科学基金社科学术社团主题学术活动资助课题研究类项目“中国共产党历史知识图谱与知识索引构建研究”(21STA028)的研究成果之一。

Exploring Knowledge Mining Using Large Language Model Combined with Knowledge Base and AI Agent

ZHAO Bang1, CAO Shujin2   

  1. 1. School of Information Management, Sun Yat-sen University, Guangzhou, 510006;
    2. School of Information Management, Shandong University of Technology, Zibo, 255049
  • Online:2025-07-10 Published:2025-08-16
  • Contact: Correspondence should be addressed to CAO Shujin, Email:caosj@mail.sysu.edu.cn, ORCID:0000-0003-1855-4522
  • Supported by:
    This is an outcome of the project "Research on the Construction of Historical Knowledge Graph and Knowledge Index of the Communist Party of China"(21STA028)supported by National Social Science Foundation of China for the Thematic Academic Activity Funding Program for Social Science Academic Associations.

摘要: [目的/意义] 探索生成式AI大模型结合知识库与AI Agent开展知识挖掘这一情报领域典型业务的方法、工具、技术框架与应用实践,为深入探索大模型在情报领域的专业化、场景化应用提供参考。[研究设计/方法]系统调研分析大模型结合知识库与AI Agent相关技术与工具,开展针对科技文献的知识挖掘及测试。[结论/发现] 大模型作为逻辑中枢结合知识库与AI Agent链接领域知识与特定工具,可自主细分知识挖掘任务,更有全流程自主化、智能化完成的能力。[创新/价值] 从概念、方法、技术框架以及开发应用等角度较为系统地探析基于大模型开展知识挖掘任务的智能手段,为未来情报领域相关实践和研究提供一定的启示。

关键词: 生成式AI, 大模型, 知识库, AI Agent, 知识挖掘

Abstract: [Purpose/Significance] The purpose of this paper is to explore the methods, tools, technical frameworks, and application practices on how to use Large Language Model(LLM)combined with knowledge base and AI agent for knowledge mining, which is a typical business in the field of intelligence, and to provide a reference for in-depth explorations of the professional and scenario-based applications of LLMs in the field of intelligence. [Design/Methodology] This study systematically investigates and analyzes LLM combined with knowledge base and AI Agent related technologies and tools, and carries out knowledge mining and testing for scientific and technological literature. [Findings/Conclusion] As a logical hub, LLM combines knowledge base and AI Agent to link domain knowledge with specific tools, enabling independent subdivision of knowledge mining tasks, with the ability to complete the entire process autonomously and intelligently. [Originality/Value] This paper systematically and comprehensively explores the intelligent means for carrying out knowledge mining tasks based on LLM from the perspectives of concepts, methods, technological frameworks, and applications, providing certain insights for future practices and research in the field of intelligence.

Keywords: Generative AI, Large language model, Knowledge base, AI agent, Knowledge mining