通信学报2026，Vol.47Issue(1)：54-73,20.DOI:10.11959/j.issn.1000−436x.2026014
大语言模型赋能的网络配置分析技术研究
Research on network configuration analysis technology empowered by large language models
摘要
Abstract
Due to the limited intelligence,complex operation and reliance on specialized expertise of the existing net-work configuration analysis tools,leading to significant human effort,a network configuration analysis framework em-powered by large language model(LLM)was proposed.By integrating LLM with formal verification tools,the frame-work interpreted the semantics of network configurations to automatically construct a network knowledge graph,and per-formed reasoning and completion based on this graph to reconstruct the network state.It further guided the LLM to ap-ply domain knowledge,enabling automated analysis for tasks such as network configuration error detection and key node identification.Experiments based on configuration files from the EVE-NG,Cisco CCNA training cases,and real-world Internet topologies were conducted from multiple perspectives,including network modeling accuracy,application effectiveness,runtime efficiency,and robustness.Results indicate that the proposed framework can quickly and accu-rately restore the network state based on configurations,adapt to the analysis requirements of diverse network scenarios,and achieve one-click analysis,thus providing important technical support for network management.关键词
网络管理/网络配置/大语言模型/知识图谱Key words
network management/network configuration/large language model/knowledge graph分类
信息技术与安全科学引用本文复制引用
李鹏飞,刘宇靖,苏金树,喻波..大语言模型赋能的网络配置分析技术研究[J].通信学报,2026,47(1):54-73,20.基金项目
国家自然科学基金资助项目(No.62372462) （No.62372462）
国家重点研发计划基金资助项目(No.2022YFB3104800) （No.2022YFB3104800）
"芙蓉计划"青年英才基金资助项目(No.2024RC3136)The National Natural Science Foundation of China(No.62372462),The National Key Research and Development Program of China(No.2022YFB3104800),The Science and Technology Innovation Program of Hunan Province(No.2024RC3136) （No.2024RC3136）
