源代码分析注释的质量评价框架

在如今的软件开发中, 开源软件的使用越来越普遍, 但是对大型开源软件的理解和维护仍然是一项复杂的工作. 开源软件通常缺乏完善的文档和注释, 想要完整的理解开源系统难度较大, 研究界产生了一种通过分析大型开源软件的源代码, 进而深入理解系统, 发现和修复系统漏洞的软件分析型任务. 源代码分析注释是软件分析型任务的一项重要产出, 它是一种以注释形式存在的细粒度代码分析报告, 数量庞大, 难以快速做出质量评价. 在传统的软件质量评价中, 对注释的评价通常局限于覆盖度和文本长度, 不能满足源代码分析注释质量评价的要求. 为了更好的评价源代码分析注释的质量, 本文结合现有的对代码注释质量评价的研究以及信息质量领域的评价方法, 提出了一种综合考虑客观质量属性和主观质量属性的质量评价框架. 结合实际的项目数据分析, 本文的方法可以更有效的检测出注释中的冗余以及无关内容, 发现相关质量问题, 从而对源代码分析注释进行更全面的质量评价.

Quality Evaluation Framework of Source Code Analysis Comments

The application of open source software is more and more popular these days, but maintenance and full understanding of huge open source system is still a hard problem. The documentations and comments that are useful for open source system understanding is limited. A new type of software analysis task emerged in research field. The source code analysis comment is an important product of software analysis task. It is a type of code analysis documentation existing like code comments. Usually the amount of source code analysis comments is huge, and it is hard to evaluate its quality in limited time. Traditional quality evaluation method of code comments, which only measure comment length or comment ratio, is too simple. It's no longer suitable for source code analysis comments' quality evaluation. Based on existing code comment quality evaluation methods and some existing works in information quality, this paper propose a more complete quality evaluation framework for source code analysis comments, taking both objective quality and subjective quality into consideration. Data gathered from real projects show that the new framework is more suitable for finding quality issues and evaluating the quality of source code analysis comments.