Page 274 - 《软件学报》2025年第10期

软件学报 ISSN 1000-9825, CODEN RUXUEW E-mail: jos@iscas.ac.cn
2025,36(10):4671−4694 [doi: 10.13328/j.cnki.jos.007294] [CSTR: 32375.14.jos.007294] http://www.jos.org.cn
©中国科学院软件研究所版权所有. Tel: +86-10-62562563

*
结合大语言模型和领域知识库的证券规则规约方法

李靓果 1 , 薛志一 1 , 陈小红 1 , 张 民 1 , 陈良育 1 , 李萍萍 2 , 姜婷婷 2

1
(上海高可信计算重点实验室 (华东师范大学), 上海 200062)
2
(国泰君安证券股份有限公司 数据中心, 上海 201201)
通信作者: 陈小红, E-mail: xhchen@sei.ecnu.edu.cn; 姜婷婷, E-mail: jiangtingting@gtjas.com

摘 要: 业务规则在证券领域至关重要, 它们是证券交易系统的需求的来源. 鉴于业务规则的易变性, 如何提升从
业务规则交易文档中规约出软件需求的效率, 成为一个核心的问题. 证券业务规则文档具有与软件不相关描述多、
专业术语多、上下文相关表述多和抽象表示多等特性, 其自动化规约需要领域相关知识的支持. 如何将领域相关
知识融入自动化过程中, 成为规约的关键问题. 提出了一种结合大语言模型和领域知识库的证券领域业务规则自
动规约方法, 对大语言模型通过微调、上下文学习等嵌入领域知识执行规则分类和需求信息提取等自然语言处理
任务. 此外, 还通过领域知识库提供专业领域知识, 进行需求的可操作化和关系识别, 最终形成数据流形式的需求
规约. 评估结果显示, 该方法能够处理各种证券交易领域的业务规则文档, 在评估数据集上的平均功能点识别率为
91.97%, 达到甚至超越了领域专家的水平, 与人类参与者相比, 效率平均提高了 10 倍.
关键词: 证券领域; 业务规则; 需求规约; 大语言模型; 领域知识; 软件需求
中图法分类号: TP311

中文引用格式: 李靓果, 薛志一, 陈小红, 张民, 陈良育, 李萍萍, 姜婷婷. 结合大语言模型和领域知识库的证券规则规约方法. 软件
学报, 2025, 36(10): 4671–4694. http://www.jos.org.cn/1000-9825/7294.htm
英文引用格式: Li LG, Xue ZY, Chen XH, Zhang M, Chen LY, Li PP, Jiang TT. Specification Method of Securities Rules Integrating
Large Language Models and Domain Knowledge Base. Ruan Jian Xue Bao/Journal of Software, 2025, 36(10): 4671–4694 (in Chinese).
http://www.jos.org.cn/1000-9825/7294.htm

Specification Method of Securities Rules Integrating Large Language Models and Domain
Knowledge Base
1
1
1
2
1
1
LI Liang-Guo , XUE Zhi-Yi , CHEN Xiao-Hong , ZHANG Min , CHEN Liang-Yu , LI Ping-Ping , JIANG Ting-Ting 2
1
(Shanghai Key Laboratory of Trustworthy Computing (East China Normal University), Shanghai 200062, China)
2
(Data Center, Guotai Junan Securities Co. Ltd., Shanghai 201201, China)
Abstract: Business rules are crucial for the securities domain and serve as the source of requirements for securities trading systems. Due
to the variability of these business rules, how to improve the efficiency of specifying software requirements from business rule trading
documents has become a core problem. The securities business rule documents feature numerous software-unrelated descriptions, abundant
professional terms, and many context-related expressions and abstract representations, which necessitate the support of domain-specific
knowledge for automatic specification. As a result, how to integrate the domain-related knowledge into the automatic process becomes a
key problem for specification. This study proposes an automatic specification method for securities domain businesses integrating large
language models and the domain knowledge base. It leverages the large language models, employing techniques such as fine-tuning and in-
context learning to embed domain knowledge for natural language processing tasks such as rule classification and requirement information
extraction. Additionally, this study also employs the domain knowledge base to provide professional knowledge and assist in the
operationalization and relationship extraction of requirements. Finally, requirement specification in the form of data flow is formed. The

* 基金项目: 国家自然科学基金 (62161146001, 62372176, 62272166); 上海市可信工业互联网软件协同创新中心项目; 上海市“数字丝路”
可信智能软件国际联合实验室项目 (22510750100)
收稿时间: 2024-01-31; 修改时间: 2024-04-18, 2024-08-03; 采用时间: 2024-09-19; jos 在线出版时间: 2025-06-27
CNKI 网络首发时间: 2025-06-30