[1]张中华,胡金先,刘宇哲.基于大模型与RAG技术的铁路智能旅服问答系统构建研究[J].高速铁路技术,2025,16(06):126-132.[doi:10.12098/j.issn.1674-8247.2025.06.018]
　ZHANG Zhonghua,HU Jinxian,LIU Yuzhe.On the Construction of an Intelligent Railway Passenger Service Q&A System Based on Large Language Models and RAG[J].HIGH SPEED RAILWAY TECHNOLOGY,2025,16(06):126-132.[doi:10.12098/j.issn.1674-8247.2025.06.018]
点击复制

基于大模型与RAG技术的铁路智能旅服问答系统构建研究()

《高速铁路技术》[ISSN:1674-8247/CN:51-1730/U]

卷:
16卷
期数:
2025年06期
页码:
126-132
栏目:
客运服务
出版日期:
2025-12-30

文章信息/Info

Title:
On the Construction of an Intelligent Railway Passenger Service Q&A System Based on Large Language Models and RAG
文章编号:
1674-8247(2025)06-0126-07
作者:
张中华; 胡金先; 刘宇哲
(中国铁路乌鲁木齐局集团有限公司, 乌鲁木齐 830011)
Author(s):
ZHANG Zhonghua;  HU Jinxian;  LIU Yuzhe
(China Railway Urumqi Group Co., Ltd., Urumqi 830011, China)
关键词:
RAG技术;  大语言模型;  智能问答;  铁路旅客服务;  信息检索
Keywords:
Retrieval-augmented Generation(RAG);  Large Language Models(LLMs);  intelligent Q&; A system;  railway passenger service;  information retrieval
分类号:
U293; TP391.3
DOI:
10.12098/j.issn.1674-8247.2025.06.018
文献标志码:
A
摘要:
随着铁路旅客服务需求的持续增长,传统信息查询系统因理解能力有限、上下文感知不足等问题,难以满足乘客与调度人员对高效、准确信息服务的需求。针对这一问题,提出一种融合大语言模型(Large Language Models, LLMs)与检索增强生成(Retrieval-Augmented Generation, RAG)技术的智能问答系统,专注于铁路旅客服务场景下的信息问答任务。该系统通过RAG机制融合语义检索与语言生成,提升对铁路专业知识的响应能力和答案准确性,同时构建了覆盖行业规则和乘客服务内容的知识库。基于500条真实问题设计实验,对比关键词匹配与无检索生成方法。结果表明,本系统在500条真实问题测试中取得了91.2%的回答准确率,较关键词检索方法提升近19%,较ChatGLM提升约10%; 平均响应时间为3.2 s,用户满意度评分达到4.6/5.0,各项指标均显著优于对比系统,验证了本方法在铁路旅客服务场景中的实用性与推广价值。
Abstract:
With the continuous growth of passenger service demands in the railway sector, traditional information query systems struggle to meet the needs for efficient and accurate information services for passengers and dispatchers, due to limitations in comprehension and context awareness. To address this issue, the paper proposed an intelligent question-answering(Q&A)system that integrates Large Language Models(LLMs)with Retrieval-augmented Generation(RAG)technology, specifically tailored for railway passenger service scenarios. The system leveraged the RAG framework to combine semantic retrieval with language generation, to enhance its capability to respond to domain-specific railway knowledge and improve answer accuracy, while also constructed a knowledge base covering industry regulations and passenger service content. Experiments were designed based on 500 real-world user queries, comparing the system with keyword matching and non-retrieval generation methods. The results show that the proposed system achieves an answer accuracy of 91.2% on the 500-question test set, outperforming keyword retrieval methods by nearly 19% and surpassing ChatGLM by approximately 10%. The system also achieves an average response time of 3.2 seconds and user satisfaction score of 4.6/5.0. These results validate the practicality and potential for broader application of the proposed method in railway passenger service scenarios.

参考文献/References:

[1] 王全蕊. 基于学习反馈的学科知识图谱智能问答系统研究[J]. 科技与创新, 2025(2): 54-57.
WANG Quanrui. Research on Intelligent Question Answering System of Subject Knowledge Map Based on Learning Feedback[J]. Science and Technology & Innovation, 2025(2): 54-57.
[2] 海佳丽, 汪润, 袁良志, 等. 基于检索增强的中医药标准知识问答系统构建探索与实践[J]. 数据分析与知识发现, 2025, 9(7): 165-174.
HAI Jiali, WANG Run, YUAN Liangzhi, et al. Constructing a Retrieval-augmented Question-answering System for Traditional Chinese Medicine Standards[J]. Data Analysis and Knowledge Discovery, 2025, 9(7): 165-174.
[3] 李博. 基于对话文本的FAQ知识库构建技术研究[D]. 西安: 西安电子科技大学, 2020.
LI Bo. Research on FAQ Knowledge Base Construction Technology Based on Dialogue text[D]. Xi'an: Xidian University, 2020.
[4] 吴璇, 付涛. 检索增强生成技术研究综述[J]. 计算机工程与应用, 2025, 61(20): 19-35.
WU Xuan, FU Tao. Comprehensive Review of Retrieval-augmented Generation[J]. Computer Engineering and Applications, 2025, 61(20): 19-35.
[5] 鞠炜刚, 汪鹏, 王佳. 基于大语言模型和RAG的持续交付智能问答系统[J]. 计算机技术与发展, 2025, 35(2): 107-114.
JU Weigang, WANG Peng, WANG Jia. Continuous Delivery Intelligent Question-answering System Based on Large Language Models and RAG[J]. Computer Technology and Development, 2025, 35(2): 107-114.
[6] 丁宁, 宋雨欣, 单泽田, 等. 基于检索增强生成(RAG)技术的医学教学辅助智能问答系统的构建探索[J]. 中国医学教育技术, 2025, 39(1): 1-5.
DING Ning, SONG Yuxin, SHAN Zetian, et al. Exploration of the Construction of a Medical Teaching Assistant Intelligent Question-answering System Based on Retrieval-augmented Generation Technology[J]. China Medical Education Technology, 2025, 39(1): 1-5.
[7] JING LI, OSKAR BARTOSZ, CHENGYU WANG, et al, Shared Neural Space: Unified Precomputed Feature Encoding for Multi-task and Cross Domain Vision [EB/OL].(2025-09-24)[2025-07-03]. https://arxiv.org/abs/2509.20481.
[8] 王亮. 检索增强生成(RAG)驱动的知识服务: 原理、范式及评估[J]. 科技与出版, 2025(4): 37-46.
WANG Liang. Retrieval-augmented Generation(RAG)-driven Knowledge Service: Principles, Paradigms, and Evaluation[J]. Science-Technology & Publication, 2025(4): 37-46.
[9] 吴頔, 陆海峰, 漆艺. 基于生成式人工智能的开放教育问答系统设计[J]. 智能物联技术, 2024, 7(6): 69-72.
WU Di, LU Haifeng, QI Yi. Design of an Open Education Question-answering System Based on Generative Artificial Intelligence[J]. Technology of IoT & AI, 2024, 7(6): 69-72.
[10] 高雅奇. 基于大语言模型和RAG技术的高校知识库智能问答系统构建与评价[J]. 电脑知识与技术, 2024, 20(29): 18-20, 38.
GAO Yaqi. Construction and Evaluation of Intelligent Question Answering System of University Knowledge Base Based on Large Language Model and RAG Technology[J]. Computer Knowledge and Technology, 2024, 20(29): 18-20, 38.
[11] 塔凌夫. 基于12306客服中心咨询数据的铁路客运满意度评价[D]. 北京: 北京交通大学, 2021.
TA Lingfu. Evaluation of Railway Passenger Satisfaction Based on the Consulting Data of 12306 Customer Service Center[D]. Beijing: Beijing Jiaotong University, 2021.
[12] 贾臻, 罗智文, 王一全. 智能技术在铁路运输中的应用与展望[J]. 中国航务周刊, 2025(21): 56-58.
JIA Zhen, LUO Zhiwen, WANG Yiquan. Application and Prospect of Intelligent Technology in Railway Transportation[J]. China Shipping Gazette, 2025(21): 56-58.
[13] 王辉, 李昊光, 傅家琪, 等. 铁路客运服务质量监督监察管理系统设计与实现[J]. 铁路计算机应用, 2025, 34(1): 39-43.
WANG Hui, LI Haoguang, FU Jiaqi, et al. Railway Passenger Transport Service Quality Supervision and Supervision Management System[J]. Railway Computer Application, 2025, 34(1): 39-43.
[14] 蔡振华, 王燕贞, 杨净. 基于LangChain和ChatGLM的高校财务问答系统研究与实现[J]. 现代计算机, 2024, 30(15): 104-110.
CAI Zhenhua, WANG Yanzhen, YANG Jing. Research and Implementation of a Financial Question Answering System for Universities Based on LangChain and ChatGLM[J]. Modern Computer, 2024, 30(15): 104-110.
[15] KARPUKHIN V, OGUZ B, MIN S, et al. Dense Passage Retrieval for Open-domain Question Answering[C]//Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing(EMNLP). Online. Stroudsburg, PA, USAACL, 2020.

备注/Memo

备注/Memo:
收稿日期:2025-07-03
作者简介:张中华(1978-),男,高级工程师。
基金项目:中国国家铁路集团有限公司科技研究开发计划(P2024S001)
更新日期/Last Update: 2025-12-30