摘要
针对传统软件工程知识分类方法效率低下的问题,提出一种改进的软件工程知识分类方法。依据软件工程知识体系(SWEBOK)对构件行为进行聚类,确定关联系数、最佳聚类数和模糊关联矩阵,基于K-NN算法和结构建模方法生成软件知识分类系统,并根据训练先验知识将新知识归入到SWEBOK的对应类别下。实验结果表明,该方法具有较好的分类效果。
Aiming at the problems that traditional means towards software engineering knowledge classification is inefficient,this paper presents an improved software engineering knowledge classification method.Regarding the architecture of Software Engineering Body Of Knowledge(SWEBOK),it goes on clustering for component behavior,determines the behavior correlation coefficient,the best clustering numbers,fuzzy correlation matrix,applies K-NN Algorithm and structure modeling method to generate a knowledge classification system.It can train the system with experience knowledge to classify new knowledge into corresponding SWEBOK categories.Experimental results show that the method has good classification efficiency.
出处
《计算机工程》
CAS
CSCD
北大核心
2011年第9期110-111,114,共3页
Computer Engineering
基金
国家自然科学基金资助项目(60874120)
关键词
软件工程知识体系
接口自动机
构件行为聚类
聚类构造器
Software Engineering Body Of Knowledge(SWEBOK)
Interface Automata(IA)
component behavior clustering
clustering constructor