AI编排软件演化中的软件变更请求：文献综述与研究议程
原标题：Software Change Request in AI-Orchestrated Software Evolution: Literature Review and Research Agenda
AI 摘要
该论文对移动应用评论（MAR）中软件变更请求（SCR）的识别与分类研究进行了系统性文献综述，梳理了机器学习与深度学习在该任务中的应用现状。研究发现常用数据集规模从5千条到超1800万条评论不等，随机森林与支持向量机占主导，深度学习采用有限，F1分数在0.08至1之间波动，SURF、ChangeAdvisor等现有工具仅实现部分自动化。作者据此提出比较分类法、研究缺口与标注框架，主张统一分类方案、共享数据集和标注框架是推进SCR研究、改善移动应用维护的关键。
正文节选
[Page 1] Pakistan Journal of Scientific Research (Vol. 6 No. 1 March, 2026) SOFTWARE CHANGE REQUEST IN AI-ORCHESTRATED SOFTWARE EVOLUTION: LITERATURE REVIEW AND RESEARCH AGENDA Muhammad Nadeem a, b, Khurram Shahzad c and Muhammad Nadeem Majeed c a Department of Computer Science, FCIT, University of the Punjab, Lahore, Pakistan b Department of Artificial Intelligence, SST, University of Management and Technology, Lahore, Pakistan