基于LLM的UML类图评估与修复策略研究

AI 摘要

该研究评估了大型语言模型（LLM）在UML类图建模中的表现，发现与专家设计的类图相比，LLM生成的类图存在关键类识别不完整、关系混淆或遗漏、继承关系不足、未使用类和循环依赖等问题。研究提出了记忆强化、外部知识注入和检测引导的自动修复三种策略，实验显示这些策略能有效改善模型输出，关键类识别修复率达85%，耦合关系修复率46%，继承关系修复率69%，未使用类和循环依赖修复率均达100%。

正文节选

UML Class Diagram Evaluation and Repair Strategies based on LLMs Abstract UML class diagrams are a crucial tool for defining the structure of software systems, but designing accurate and comprehensive class diagrams is a challenging task. Traditionally, creating UML models relies on the expertise and experience of professionals. However, with the development of AI technologies, particularly large language models (LLMs), new opportunities for software modeling have emerged. Despite this, there ha

发布时间：2026-09-01 12:00

抓取时间：2026-09-01 14:12

来源机构：arXiv