摘要
为了解决软件可维护性的定量评估,提出基于神经网络的支持向量机工作原理,构造一种软件可维护性的定量分析模型。以16种可维护性为基础,建立软件可维护性评估模型,把以往每个软件的16种可维护性指标看作一个1×16维行矢量,并作为支持向量机的训练矢量,对其进行聚类分析,最终把软件可维护性水平分为:可维护性低,可维护性中等,可维护性高等3个类别,并对软件可维护性水平做出预测。
Based on neural network a support vector machine is presented, and then a new software maintainability evaluation model is built. Based on sixteen kinds ofmaintainability, the model considers the maintainability ofevery project in the past as a 1 ×16 dimension row vector and them as the training vectors of SVM, then has a cluster analysis about the training vectors, By classifying the analysis, project maintainability is divided into three levels： low maintainability, middle maintainability and high maintainability, the application in software maintainability is predicted.
出处
《计算机工程与设计》
CSCD
北大核心
2008年第3期566-569,616,共5页
Computer Engineering and Design
基金
湖南省自然科学基金项目(05JJ40098)
湖南省教育厅科研基金项目(05C720)
关键词
软件项目
模型
可维护性
支持向量机
聚类
预测
sottware project
model
maintainability
support vectormachine
classify
predict
