《计算机应用》唯一官方网站 ›› 2026, Vol. 46 ›› Issue (6): 1721-1727.DOI: 10.11772/j.issn.1001-9081.2025060727

• 人工智能 • 上一篇    下一篇

基于知识增强大语言模型架构的政务热线问答系统

熊龙雨, 杜圣东(), 史浩琛, 胡节, 杨燕, 李天瑞   

  1. 西南交通大学 计算机与人工智能学院，成都 611756
  • 收稿日期:2025-07-01 修回日期:2025-08-03 接受日期:2025-08-15 发布日期:2025-09-04 出版日期:2026-06-10
  • 通讯作者: 杜圣东
  • 作者简介:熊龙雨（2001—），女，贵州安龙人，硕士研究生，CCF会员，主要研究方向：大模型应用、数据挖掘
    史浩琛（1997—），男，四川绵阳人，博士研究生，主要研究方向：城市计算、时间序列分析、计算机视觉、深度学习
    胡节（1978—），女，四川成都人，副教授，博士，CCF专业会员，主要研究方向：人工智能、机器学习、数据挖掘、聚类分析、聚类集成
    杨燕（1964—），女，安徽合肥人，教授，博士，CCF杰出会员，主要研究方向：人工智能、大数据分析与挖掘、集成学习、多视图学习、聚类分析、时空挖掘
    李天瑞（1969—），男，福建莆田人，教授，博士，CCF杰出会员，主要研究方向：人工智能、数据挖掘、知识发现。
    第一联系人：杜圣东（1981—），男，重庆人，副教授，博士，CCF会员，主要研究方向：数据挖掘、人工智能、大数据分析、云计算、机器学习、深度学习
  • 基金资助:
    国家自然科学基金面上项目(62276215);国家自然科学基金联合基金资助项目(U2468207);四川省重大科技专项(2024ZDZX0012)

Government affairs hotline question answering system based on knowledge-enhanced large language model architecture

Longyu XIONG, Shengdong DU(), Haochen SHI, Jie HU, Yan YANG, Tianrui LI   

  1. School of Computing and Artificial Intelligence，Southwest Jiaotong University，Chengdu Sichuan 611756，China
  • Received:2025-07-01 Revised:2025-08-03 Accepted:2025-08-15 Online:2025-09-04 Published:2026-06-10
  • Contact: Shengdong DU
  • About author:XIONG Longyu， born in 2001， M. S. candidate. Her research interests include large model application， data mining.
    SHI Haochen， born in 1997， Ph. D. candidate. His research interests include urban computing， time series analysis， computer vision， deep learning.
    HU Jie， born in 1978， Ph. D.， associate professor. Her research interests include artificial intelligence， machine learning， data mining， cluster analysis， clustering ensemble.
    YANG Yan， born in 1964， Ph. D.， professor. Her research interests include artificial intelligence， big data analysis and mining， ensemble learning， multi-view learning， cluster analysis， spatio-temporal mining.
    LI Tianrui， born in 1969， Ph. D.， professor. His research interests include artificial intelligence， data mining， knowledge discovery.
    First author contact:DU Shengdong， born in 1981， Ph. D.， associate professor. His research interests include data mining， artificial intelligence， big data analysis， cloud computing， machine learning， deep learning.
  • Supported by:
    General Program of National Natural Science Foundation of China(62276215);Joint Fund of National Natural Science Foundation of China(U2468207);Major Science and Technology Project in Sichuan Province(2024ZDZX0012)

RichHTML

PDF

摘要：

针对当前政务热线问答（QA）系统中人工回复效率低以及传统检索增强生成（RAG）系统存在查询甄别机制不精准和意图差异识别不足等问题，提出一种基于知识增强大语言模型（LLM）架构的政务热线QA系统ChatGovt。首先，为了提高回复效率，设计整合智能问题分流和结构化反馈的系统架构，通过意图识别实现咨询类和投诉建议类等问题的分类处理；其次，为了提高系统检索知识的质量，提出多阶段语义增强的检索方法，包括历史对话总结检索、语义重排序和自我反思决策这3个阶段；最后，通过联网查询补充跨域知识，以形成政务咨询的服务闭环。实验结果表明，在检索质量上，相较于传统RAG系统，ChatGovt的查询-知识相关性、真实答案-知识相关性和知识支持度分别提升了15.0%、7.4%和24.6%；在系统整体性能上，ChatGovt的答案召回率相较于微调的GLM（General Language Model）4-9b-chat提升了55.4%，人工评价相较于商用系统“豆包”提升了27.3%。可见，该系统为政务热线QA系统的技术优化提供了可借鉴的架构与方法，能有效提升政务热线的响应效率与服务精准度，并推动政务服务的智能化转型。

关键词: 政务问答系统, 检索增强生成, 大语言模型, 语义检索, 知识增强

Abstract:

A government affairs hotline Question Answering （QA） system based on knowledge-enhanced Large Language Model （LLM） architecture， namely ChatGovt， was proposed to address the issues in current systems， such as low manual response efficiency， as well as inaccurate query screening mechanisms and insufficient intention difference recognition in traditional Retrieval-Augmented Generation （RAG） systems. Firstly， to improve response efficiency， a system architecture integrating intelligent question diverting and structured feedback was designed， thereby enabling classified processing of problems of consultation， complaint/suggestion， and other types with intent recognition. Then， to improve the system’s knowledge retrieval quality， a multi-stage semantic-enhanced retrieval method was proposed， including three stages： historical dialogue summary retrieval， semantic re-ranking， and self-reflective decision-making. Finally， cross-domain knowledge was supplemented through online queries， so as to form a service closed-loop for government consultation. Experimental results show that in terms of retrieval quality， compared with the traditional RAG system， ChatGovt has the query-knowledge relevance， real answer-knowledge relevance， and knowledge support improved by 15.0%， 7.4%， and 24.6% respectively； in terms of overall system performance， ChatGovt has the answer recall increased by 55.4% compared with the fine-tuned GLM （General Language Model）4-9b-chat model， and has the manual evaluation improved by 27.3% compared with the commercial system “Doubao”. It can be seen that this system provides a reference-worthy architecture and methodology for the technical optimization of government affairs hotline QA systems， can improve the response efficiency and service accuracy of government affairs hotlines effectively， and promotes the intelligent transformation of government services.

Key words: government affairs Question Answering (QA) system, Retrieval-Augmented Generation (RAG), Large Language Model (LLM), semantic retrieval, knowledge enhancement

中图分类号: 

版权所有 © 2021 四川计算机应用杂志社有限公司
蜀ICP备 05010208 号-3　新出网证(川)字026号
技术支持：北京玛格泰克科技发展有限公司
访问总数：
今日访问：