ISCAS OpenIR  > 中科院软件所  > 中科院软件所
一种面向开发阶段的缺陷分类方法及度量模型
其他题名A Phase-oriented Defect Classification Method and Measurement Model
车美儒
2007-06-05
学位授予单位中国科学院软件研究所
学位博士
学位授予地点软件研究所
关键词软件缺陷分类 缺陷度量分析 开发阶段 过程改进
摘要软件缺陷分类是缺陷度量和分析的基础,是深入挖掘缺陷信息并用以评价和改进软件开发过程的重要前提。缺陷的度量分析可以帮助软件组织找到缺陷产生的根源,并有效地排除缺陷、改进过程。缺陷的分类和度量分析既是保证软件质量的必要手段,也是产品度量和过程度量的必要环节。 传统的缺陷分类方法其最大的局限性在于不能很好地获得缺陷关联的过程的语义信息,为过程改进提供的指导和帮助非常有限。相应的缺陷度量分析方法对软件过程控制和改进提供的支持也很有限。 本文着眼于将缺陷的采集、分类、度量以及分析等一系列工作和软件开发阶段相关联,通过挖掘缺陷在各个开发阶段上的详细语义信息,找到开发过程中潜在的问题和需要改进的地方。 具体地,本文主要在以下几方面进行研究和实践: 提出了面向开发阶段的软件缺陷分类方法Phase-DC(Phase-oriented Defect Classification)。该方法定义了缺陷的五个描述属性,属性信息有助于建立缺陷和开发阶段的关联关系,以及缺陷体现在各个阶段上的详细语义信息。通过对缺陷合理的分类,为找到实施过程改进的方向和理论依据提供支持。 提出了基于Phase-DC分类方法的缺陷度量模型MMDC(Measurement Model of Defect Classified)。该模型基于缺陷分类的结果,建立并计算相应的度量指标,得到各个开发阶段对应的过程引入及排除缺陷的量化能力,并结合详细的属性值分析缺陷存在和产生的原因,以及过程的执行情况和改进建议。 导出了基于Phase-DC和MMDC的软件缺陷度量方法和度量过程的算法,实现了软件缺陷度量的完整过程和主要步骤。为了支撑缺陷度量工作并不断地完善缺陷管理的功能,我们在目前中国科学院软件研究所开发的软件质量管理平台项目管理子系统中,对已有的缺陷管理功能实施了部分改进,开发了改进之后的页面原型。 最后,我们将Phase-DC方法和MMDC模型用于软件质量管理平台的缺陷分类和度量工作中,验证了本文提出的方法和模型的可用性以及对过程改进的重要指导作用。该平台是中科院软件所多年从事软件质量保证研究工作的科研成果,目前已经成功应用于国内多家软件组织。 实践结果表明,本文的研究为缺陷度量分析和软件过程改进提供了有力的支持,在分析缺陷存在根源、寻找过程改进目标和依据方面发挥了重要的作用。本文提出的分类方法和度量算法可以根据组织不同需求定制,对于软件组织的缺陷分类和缺陷度量分析工作提供了很好的指导和参考价值。
其他摘要Software defect classification is the basis of defect measurement and analysis, and also is the premise of software process evaluation and improvement. Defect measurement and analysis can help software organization to find the origin of defects, remove them effectively, and improve the process. Defect classification and measurement are not only necessary methods to assure software quality, but also essential steps of product measurement and process measurement. The main limit of existent defect classification methods is not convenient to obtain defect information with development process, and providing little support to process improvement. Besides, the defect measurement method is also limited to software process control and improvement. In this paper, we focus on combining defects collection, classification, measurement and analysis to software development phase. We find the potential problem and improvement of development process by identifying the detailed information of defects. Specifically, we do the major research and practice in the following aspects: Putting forward a phase-oriented software defect classification method: Phase-DC (Phase-oriented Defect Classification). This classification defines five attributes of defects, and its attribute information contribute to establishing relationship between defect and development phase, and embodying detailed semantic information of defects in various stages. The fundamental purpose of Phase-DC method is to find the direction and theoretical basis of process improvement through reasonable defect classification. On this basis, we bring forward a defect measurement model called MMDC (Measurement Model of Defect Classified). The model is based on the result of defect classification, establishes and calculates the corresponding defect metrics, in order to get the quantitative ability in injecting and removing defects for each phase. Then, we combine the detailed attribute to analyze the cause of defect and the implementation of process to obtain process improvement suggestions. Based on the Phase-DC and the MMDC, we conclude the software defect measurement method and measurement algorithms, and implement the integrated measurement process and steps. In order to constantly support and improve the function of defect management, we implement some improvement in our defect management module of the software quality management platform, and develop the page prototypes. At last, we apply Phase-DC and MMDC to our defect classification and measurement when we develop software quality management platform, and verify the usability of our defect measurement method and its distinct instructive effect. The platform is the research fruit of the Institute of Software, Chinese Academy of Sciences, and has already been successfully implemented in many software enterprises. Practice results show that the research in the thesis has provided strong support to defect measurement and software process improvement. Also it plays an important role in analyzing the root cause of defect and finding the goal and basis of process improvement. The classification method and measurement algorithm we put forward can be tailored according to different software organizations, and provide helpful guidance and valuable reference to software organizations.
页数106
语种中文
内容类型学位论文
URI标识http://ir.iscas.ac.cn/handle/311060/7142
专题中科院软件所_中科院软件所
推荐引用方式
GB/T 7714
车美儒. 一种面向开发阶段的缺陷分类方法及度量模型[D]. 软件研究所. 中国科学院软件研究所,2007.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
10001_20042801502901(3059KB) 限制开放--请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[车美儒]的文章
百度学术
百度学术中相似的文章
[车美儒]的文章
必应学术
必应学术中相似的文章
[车美儒]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。