基于特征的领域构件高内聚控制  

Variability Control of High-cohesion Domain Component Based on Feature

作　　者：王映辉[1] 张世琨[1] 王立福[1] 

机构地区：[1]北京大学软件工程国家工程研究中心

出　　处：《计算机科学》2005年第7期237-241,共5页Computer Science

基　　金：本文得到国家"八六三"高技术研究发展计划项目(2001AA113171);国家"九七三"重点基础研究发展规划项目(2002CB312006);国家博士后基金(20040350251)

摘　　要：构件的高内聚性是基于构件复用研究领域常见的话题,也是软件开发者一直追求的目标。无论是对软件体系结构的获取,还是对软件演化的探究,甚至遗留系统的现代化改造,都与高内聚性构件的确定有着不可分割的天然联系。通过特征跟踪矩阵,建立了领域构件与特征之间的联系;给出了基于特征的n维向量空间中领域构件向量的表示,进而获取了反应领域构件间密切关系的领域构件向量距离矩阵;基于领域构件向量距离矩阵,结合等级簇聚合算法建立了领域构件等级簇树;最后在领域构件等级簇树的基础上,提出了一种高内聚领域构件获取和控制策略。It's a general topic to achieve component cohesion in the research filed based on component reuse. High-co- hesion of module (component is a special module) is always an aim of software developer. There is inherent relation- ship between All, which we get software architecture and explore software evolution and modernize legacy system, and high-cohesion component decision. The relationship between domain component and feature is established based on feature track matix. Expression of domain component vector within n-dimension space based feature is represented, then a distance matrix of domain component vector, which represent the relationship within domain components, is achieved. The domain component rank cluster tree is established combine rank cluster algorithm with the distance ma- trix of domain component vector. At last, the granularity decision strategy of high-cohesion domain component is demonstrated based on the domain component rank cluster tree.

关 键 词：领域构件 基于特征 内聚 软件体系结构 n维向量空间 距离矩阵 软件开发者 现代化改造 研究领域 软件演化 遗留系统 特征跟踪 控制策略 等级 分割 算法 簇 

分 类 号：TP311.52[自动化与计算机技术—计算机软件与理论] TP391.72[自动化与计算机技术—计算机科学与技术]