基于复杂网络的软件测度研究

随着软件系统规模的增大，软件各组成元素之间的交互日益复杂，导致软件开发经常处于不稳定状态，软件质量问题逐步突出．借鉴复杂网络的发展成果，将软件结构描述为加权有向图，使用节点重要度定义了出度系数，以此来查找和判断软件结构中缺陷概率比较高的类，并利用无标度网络的缺陷发生特性来研究软件结构的缺陷分布．实例结果表明：该度量方法对查找软件缺陷类是有效的，有利于在软件开发初期就发现缺陷，提高软件开发质量．

Research on software measurement based on complex networks

With the augmentation of the scale of software systems, the interaction between become increasingly complicated, which leads the development of software to be instable, the software elements has making problems in software quality gradually prominent. Based on the achievements obtained in complex networks, the software structure is made into a graph of weighted directed networks ; then, defining the Out-degree coefficient by a corresponding index of node important degree, to find and judge the classes with high defect probability, the flaw occurrence characteristics of scale - free networks are utilized to study the defect distribution of software structure. The last results show that the measurement method is effective in checking for software defective classes, which is helpful to finding out defects in the early phase of developing software and reducing the cost of software development and testing so as to improve the quality of software developed.