Study On Design Pattern Detection Based On Empirical Research

Under the background of information age,software development is pursuing more and more efficient and high-quality software products.Therefore,in order to improve the productivity and quality of software and reduce the cost of software development and maintenance,people have introduced software reuse technology.The initial software reuse technology is mainly code-level reuse,that is,the reused knowledge refers to the program.Later,it evolved and expanded to include design scheme,design decision,architecture,development experience,requirement design and other related aspects.In 1995,the design pattern was formally proposed.It belongs to one of the reuse technologies,and it is a product of the combination of development experience and design.Design patterns play an important role in software development and maintenance,but their function is weakened because they are hidden in code.From previous studies,we know that design patterns can be detected,and the detection of design patterns in code can help later software maintenance and reconstruction,and improve the quality of software.This thesis proposes three purposes.First,find out which design patterns can detect,understand the current detection situation of design pattern and its variant,and comprehend design pattern detection methods.Second,find out the tools of design pattern detection;study the analysis types and detection strategies of the tools.Thirdly,based on the intermediate representation of XMI,we use graph isomorphism method to conduct a case study of composite pattern,which is a way better understand the process of design pattern recognition.Combining the experience of goal one and goal two,we propose a scheme of variant detection by graph isomorphism method.Based on the above three purposes,we first need a comprehensive understanding of the field of design pattern detection,so this thesis uses systematic literature review research methods for design pattern detection.This method based on existing empirical knowledge in a certain field,to organize,analyze,summarize and strive to excavate experience and rules.Therefore,systematic literature review can satisfy the first and second purpose of this thesis.For the third purpose,this thesis identifies design pattern by graph isomorphism method and chooses composite pattern for case study,because case study can show the detection process intuitively and clearly,and help us put forward the idea of detection variants based on our experience.Finally.applying the data synthesis method to present the above research results graphically.To sum up,the contribution of this paper has two points.Firstly,a detailed systematic review of the current situation in the field of design pattern detection is made,and three research questions related to design pattern detection tools are put forward and answered.The detection tools and methods of different types of design patterns are explored in depth,which provides empirical support for further proposing effective detection methods.It also explores the detection tools and methods of different types of design patterns,which provides empirical support for further proposing effective detection methods and selecting detection objects.Secondly,a design pattern detection method based on VF2 algorithm is proposed.This method regards the class or interface in source code as the vertex of the graph,the relationship between class and interface as the edge of the graph,and takes Composite design pattern as the case study object.It presents the process of identifying design pattern intuitively by graph isomorphism method.The results show that this method can effectively detect relevant design patterns.