摘要
针对铁路客服系统面临的语义理解偏差与知识碎片化难题,提出融合大语言模型(LLM)与领域知识库的智能体协同架构。首先,结合现实中的典型业务需求分析提出智能体设计框架,包含大语言模型、规划、记忆、工具和知识图谱核心组件,设计智能体从用户多模态接入到意识识别、执行操作及结果响应的流程。然后基于知识图谱构建理论与铁路客服业务案例,分析领域知识构建与应用推理过程。最后在智能开发平台中实践了一个铁路客服原型以进一步理解设计思路,并验证方案的可行性。研究可以为铁路客服智能化技术提供更多解决思路,以提升乘客体验和运营效率。
The challenges of semantic comprehension deviations and fragmented knowledge are existed in railway customer service systems.This study proposes an intelligent agent collabora-tive architecture that integrates large language models(LLMs)with domain-specific knowledge bases.Firstly,an intelligent body design framework is proposed based on typical business needs in reality.It includes core components such as large language models,planning,memory,tools,and knowledge graphs.The framework designs the process of intelligent body from user multi-modal access to consciousness recognition,operation execution,and result response.Then,the process of domain knowledge construction and application reasoning is analyzed.It is based on the theory of knowledge graph construction and the case of railway customer service.Finally,a railway customer service prototype was implemented on an intelligent development platform to further comprehend the design approach and validate the feasibility of the solution.This research provides additional solutions for the intelligentization of railway customer service technologies,aiming to cnhance passenger experience and operational efficiency.
作者
段安勇
陈美才
路琦龙
DUAN An'yong;CHEN Meicai;LU Qilong(China Railway Nanchang Bureau Group Co.,Ltd.Safety supervision Office,Jiangxi Nanchang,330009)
出处
《长江信息通信》
2025年第8期86-89,93,共5页
Changjiang Information & Communications
关键词
铁路智能客服
领域知识
大语言模型
智能体
Railway Intelligent Customer Service
domain knowledge
Large Language Models
Intelligent Agents
