考虑多种特征因素的设计模式自动识别

Automatic Design Pattern Detection on the Consideration of Multiple Characteristic Factors

现有的设计模式自动识别方法大多只考虑设计模式的结构特征,识别准确率不高.为此,提出了一种考虑多种特征因素的设计模式自动识别方法.首先,提出了一种基于特征矩阵的待考查系统和设计模式的形式化描述方法;然后,给出了设计模式自动识别的基本流程,并详细讨论了综合考虑多种特征因素的设计模式识别算法;最后,实现了该方法的支撑工具,并使用该工具对一个开源项目进行了设计模式的识别.实验结果表明,相对于只考虑结构特征的设计模式识别方法,该方法的识别准确率较高.对于结构特征不够明显或者与其他模式具有相似结构特征的设计模式,识别准确率明显提升.

Most of the existing methods for automatic design pattern detection only consider structural characteristics of design patterns,so the detection accuracy rate is not high enough. Therefore,a method for automatic design pattern detection on the consideration of multiple characteristic factors was proposed.First,a formal description method for system under study and design patterns based on characteristic matrix was proposed. Then the basic flow of automatic design pattern detection was given,and the design pattern detection algorithm considering multiple characteristic factors was discussed in detail. Finally,a support tool for this method is implemented,and design patterns in an open source project were detected by using this tool. The experimental results show that compared with the design pattern detection methods which only consider structural characteristics,detection accuracy rate of this method is higher. For design patterns whose structural characteristics are not obvious or who have similar structure characteristics with other design patterns,accuracy rate is improved obviously.