Page 227 - 《软件学报》2026年第4期

1668 软件学报 2026 年第 37 卷第 4 期

5 总 结

根据自然语言描述的软件需求文本自动生成 UML 活动图对软件开发人员正确把握客户需求、提高需求分
析效率至关重要. 为进一步提高自动生成 UML 活动图的质量, 本文提出了一种融合中文需求文本语义特征的方
法 AutoADGC. 该方法通过语义多标签需求文本分类将需求文本的相关性语义特征和时序性语义特征融入 UML
活动图的自动生成过程, 并结合需求文本的语义特征和语法特征提取 UML 活动图图元素及其关系. 在 100 个企
业实际应用案例上的消融实验结果表明, 结合需求文本的语义特征设计 UML 活动图自动生成方法有助于提高自
动生成活动图的完整性和正确性, 降低活动图的冗余性. 此外, 与多种 UML 活动图自动生成方法的比较验证了本
文所提方法不仅可以在秒级时间内从需求文本中快速提取构建 UML 活动图所需的基本信息, 还可以有效避免需
求文本中无关信息的干扰, 准确识别需求文本包含的时序关系, 使得自动生成 UML 活动图的正确性最高提升
41.96%, 显著提高了 UML 活动图自动生成方法的性能和效率.
虽然目前本文方法仅适用于中文需求文本, 但只要有充足的需求文本数据和全面的语法知识, 本文方法可以
被扩展应用到任何语言的需求文本. 因此, 我们计划后续将 AutoADGC 方法用于分析英文软件需求文本. 此外, 本
文仅对 UML 活动图自动生成问题进行了初步探讨, 未来可着眼于将该方法推广至自动生成与需求分析和软件设
计相关的其他图表, 如 UML 用例图、数据库实体-关系图和用户界面等.

References
[1] Ahmad K, Abdelrazek M, Arora C, Bano M, Grundy J. Requirements engineering for artificial intelligence systems: A systematic
mapping study. Information and Software Technology, 2023, 158: 107176. [doi: 10.1016/j.infsof.2023.107176]
[2] Li BY, Smidts C. A zone-based model for analysis of dependent failures in requirements inspection. IEEE Trans. on Software
Engineering, 2023, 49(6): 3581–3598. [doi: 10.1109/TSE.2023.3266157]
[3] Ordoñez-Briceño K, Hilera JR, de-Marcos L, Otón-Tortosa S, Cueva-Carrión S. UML profile to model accessible Web pages. IEEE
Access, 2024, 12: 77181–77213. [doi: 10.1109/ACCESS.2024.3406688]
[4] Larman C. Applying UML and Patterns: An Introduction to Object-oriented Analysis and Design and Iterative Development. Pearson,
2004.
[5] Sharma R, Gulia S, Biswas KK. Automated generation of activity and sequence diagrams from natural language requirements. In: Proc. of
the 9th Int’l Conf. on Evaluation of Novel Approaches to Software Engineering. Lisbon: IEEE, 2014. 1–9.
[6] Nassif M, Robillard MP. Identifying concepts in software projects. IEEE Trans. on Software Engineering, 2023, 49(7): 3660–3674. [doi:
10.1109/TSE.2023.3265855]
[7] Gutiérrez JJ, Nebut C, Escalona MJ, Mejías M, Ramos IM. Visualization of use cases through automatically generated activity diagrams.
In: Proc. of the 11th Int’l Conf. on Model Driven Engineering Languages and Systems. Toulouse: Springer, 2008. 83–96. [doi: 10.1007/
978-3-540-87875-9_6]
[8] Yue T, Briand LC, Labiche Y. An automated approach to transform use cases into activity diagrams. In: Proc. of the 6th European Conf.
on Modelling Foundations and Applications. Paris: Springer, 2010. 337–353. [doi: 10.1007/978-3-642-13595-8_26]
[9] Iqbal U, Bajwa IS. Generating UML activity diagram from SBVR rules. In: Proc. of the 6th Int’l Conf. on Innovative Computing
Technology. Dublin: IEEE, 2016. 216–219. [doi: 10.1109/INTECH.2016.7845094]
[10] Yue T, Briand LC, Labiche Y. aToucan: An automated framework to derive uml analysis models from use case models. ACM Trans. on
Software Engineering and Methodology (TOSEM), 2015, 24(3): 13. [doi: 10.1145/2699697]
[11] Franch X, Palomares C, Quer C, Chatzipetrou P, Gorschek T. The state-of-practice in requirements specification: An extended interview
study at 12 companies. Requirements Engineering, 2023, 28(3): 377–409. [doi: 10.1007/s00766-023-00399-7]
[12] Kamarudin NJ, Sani NFM, Atan R. Automated transformation approach from user requirement to behavior design. Journal of Theoretical
and Applied Information Technology, 2015, 81(1): 73–83.
[13] Gulia S, Choudhury T. An efficient automated design to generate UML diagram from natural language specifications. In: Proc. of the 6th
Int’l Conf. Cloud System and Big Data Engineering. Noida: IEEE, 2016. 641–648. [doi: 10.1109/CONFLUENCE.2016.7508197]
[14] Alami N, Arman N, Khamyseh F. A semi-automated approach for generating sequence diagrams from Arabic user requirements using a
natural language processing tool. In: Proc. of the 8th Int’l Conf. on Information Technology. Amman: IEEE, 2017. 309–314. [doi: 10.
1109/ICITECH.2017.8080018]