[1]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,(02):107-114.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0347]
　JU Wei-gang,WANG Peng,WANG Jia.Continuous Delivery Intelligent Question-answering System Based on Large Language Models and RAG[J].,2025,(02):107-114.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0347]
点击复制

基于大语言模型和RAG的持续交付智能问答系统()

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2025年02期
页码:
107-114
栏目:
人工智能
出版日期:
2025-02-10

文章信息/Info

Title:
Continuous Delivery Intelligent Question-answering System Based on Large Language Models and RAG
文章编号:
1673-629X(2025)02-0107-08
作者:
鞠炜刚1; 汪鹏2; 王佳1
1. 中兴通讯股份有限公司,江苏 南京 210012;
2. 东南大学,江苏 南京 210096
Author(s):
JU Wei-gang1; WANG Peng2; WANG Jia1
1. ZTE Corporation,Nanjing 210012,China;
2. Southeast University,Nanjing 210096,China
关键词:
持续交付; 智能问答; 大语言模型; 检索增强生成; 提示词工程
Keywords:
continuous delivery; intelligent question and answer; large language model; retrieval-augmented generation; prompt engineer-ing
分类号:
TP391
DOI:
10.20165/j.cnki.ISSN1673-629X.2024.0347
摘要:
持续交付是一种持续的将各类变更快速、高质量地落实到生产环境的方法和技术,对提升产品竞争力越来越重要。 因此迫切需要对持续交付进行规划、建设和应用,但其知识范围广、专业性强、更新快,难以有效及时获取指导和帮助,影响实施效果。 针对该问题,提出了一种基于大语言模型和检索增强生成(RAG)的持续交付智能问答系统构建方法。该方法通过高质量语料处理形成数据集,采用高效微调技术训练领域大模型,使用改进的向量知识检索并结合提示词工程的多场景提示词模板技术增强生成效果,实现了一种持续交付智能问答系统。 实验结果表明,该系统对持续交付各环节的知识问答覆盖场景范围广,能有效提升回答的准确性,降低幻觉率,效果明显,从而极大帮助了持续交付的规划、实施和应用。 提出的方法和技术具备很强的通用性,可以向更多领域的智能问答推广应用。
Abstract:
Continuous Delivery (CD) is a method and technology for continuously implementing various changes rapidly and with high quality into production environments,which is becoming increasingly important for enhancing product competitiveness. Therefore,there is an urgent need to plan,build,and apply CD practices,but its broad knowledge base,specialized nature,and rapid updates make it difficult to effectively and timely obtain guidance and assistance,affecting implementation effect. To address this issue,we propose a con-struction method for a CD intelligent question-answering system based on large language models and Retrieval-Augmented Generation (RAG). The method involves forming a dataset through high-quality corpus processing,training domain-specific large models using efficient fine-tuning techniques,and employing improved vector knowledge retrieval combined with multi-scenario prompt templates enhanced by prompt engineering to improve the generation effect. The implementation resulted in a CD intelligent question-answering system. Experimental results show that the system covers a wide range of scenarios for knowledge questions and answers related to CD,effectively improves the accuracy of responses,reduces hallucination rates,and demonstrates significant effects,thereby greatly assisting in the planning,implementation,and application of CD. The proposed methods and technologies possess strong generalizability and can be extended to intelligent question-answering applications in more fields.

相似文献/References:

[1]张晓帆,刘 宁,潘 帆.持续集成系统可视化设计研究[J].计算机技术与发展,2020,30(02):58.[doi:10. 3969 / j. issn. 1673-629X. 2020. 02. 012]
　ZHANG Xiao-fan,LIU Ning,PAN Fan.Research on Visualization Design of Continuous Integration System[J].,2020,30(02):58.[doi:10. 3969 / j. issn. 1673-629X. 2020. 02. 012]
[2]张冬松,胡秀云,邬长安,等.面向 DevOps 的政务大数据分析可视化系统[J].计算机技术与发展,2020,30(08):1.[doi:10. 3969 / j. issn. 1673-629X. 2020. 08. 001]
　ZHANG Dong-song,HU Xiu-yun,WU Chang-an,et al.Analysis Visualization System of Government Big Data for DevOps[J].,2020,30(02):1.[doi:10. 3969 / j. issn. 1673-629X. 2020. 08. 001]
[3]黄东晋,秦 汉,郭 昊.基于 BERT-CNN 的电影原声智能问答系统[J].计算机技术与发展,2020,30(11):158.[doi:10. 3969 / j. issn. 1673-629X. 2020. 11. 029]
　HUANG Dong-jin,QIN Han,GUO Hao.Movie Soundtrack Intelligent Question and Answer System Based on BERT-CNN[J].,2020,30(02):158.[doi:10. 3969 / j. issn. 1673-629X. 2020. 11. 029]
[4]卢 琪,谢艺菲,谢 钧,等.知识图谱在智能问答中的应用研究[J].计算机技术与发展,2021,31(07):13.[doi:10. 3969 / j. issn. 1673-629X. 2021. 07. 003]
　LU Qi,XIE Yi-fei,XIE Jun,et al.Research on Application of Knowledge Graphs in Intelligent Question Answering[J].,2021,31(02):13.[doi:10. 3969 / j. issn. 1673-629X. 2021. 07. 003]
[5]刘 鹏,程浩然,王 莹,等.煤矿工种知识图谱智能问答研究[J].计算机技术与发展,2024,34(03):185.[doi:10. 3969 / j. issn. 1673-629X. 2024. 03. 027]
　LIU Peng,CHENG Hao-ran,WANG Ying,et al.Research on Intelligent Question Answering Based on Knowledge Graph of Coalmine Occupation[J].,2024,34(02):185.[doi:10. 3969 / j. issn. 1673-629X. 2024. 03. 027]

更新日期/Last Update: 2025-02-10