通过用况聚类促进软件结构恢复的方法  

Approach for Facilitating Structural Recovery by Clustering Use Cases

作　　者：任武 

机构地区：[1]上海工业自动化仪表研究院,上海200233

出　　处：《电子学报》2013年第7期1412-1418,共7页Acta Electronica Sinica

摘　　要：为促进遗留软件系统的理解和演化任务,需要对系统的整体功能结构有一个总体认识.常见的方法主要是利用聚类技术对静态源代码进行分析.然而在面向对象的软件系统中,由于软件的复杂性如多态和动态绑定等因素的存在,因此需要考虑对象的运行时特性.用况和用况模型从用户的角度描述了软件系统的行为特点,反映了系统的功能特性,已经成为程序理解的一个关键所在.通过对用况的聚类分析,建立软件系统的结构框架,并与实际的设计结构进行比较,以验证软件结构质量好坏,是本文所提出的研究思路.该方法结合利用关联规则挖掘技术获取用况对应的执行事件,并通过静态结构的分析对用况和相关事件的调用依赖关系进行扩充.最后通过一个开源项目的实验分析进行评估,讨论方法的适用性和有效性.In order to promote the understanding and evolution of a legacy software system,it is necessary to have an overall view of the structure.The common solution is to analyze the static source code by using clustering technique.In object-oriented software systems,however,due to the complexity of the software itself,such as exists of polymorphism and dynamic binding,and other factors,object＇s behavior actives need to be considered also.Use cases reflect the functionality of a system and play a key role in program comprehension.By clustering to establish the structural framework of the system,and by comparison with the actual design of the structure,our approach combines association rule mining techniques to obtain the mapping between use cases and corresponding events,and through the analysis of the static structure of use cases and related events invoked dependencies expansion.Finally,the experimental analysis of an open source project is evaluated and discussed to account for the applicability and effectiveness of our approach.

关 键 词：动态分析 静态分析 用况 聚类 软件结构恢复 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]