Research On Measurement Method In Open Source Software Ecosystem Based On Complex Network

Software metrics exist at all stages of the entire software development life cycle,and are important activities for understanding,evaluating,predicting,controlling and improving software development.The information obtained by software metrics is a very important resource for the software organization.Software management can effectively use the information to make informed decisions;these decisions have a positive impact on the quality of the software.Software metrics,as a basic software engineering practice,have long been included in the software process standards and become the most active areas of software engineering research.With the advance in Internet of things,cloud computing big data,and other technologies,software technology has undergone major changes.In order to handle the massive information more conveniently,satisfy the personalized and diversified user requirements and serve the public better,collective software engineering emerges as the times require.The practice of collective software engineering has greatly changed the traditional software engineering practice.It makes important changes of software development organization from factories to communities,developers from elite to the public,development processes from closed to open environment,development methods from machine engineering to social engineering.This open-source software development model based on the open source community is increasingly being recognized and adopted.A large amount of related data is generated during the collective software development of open community,which provides a rich data base for open source software metrics.Mature open source software development organizations should make full use of these metrics and related analysis results to develop short-or long-term decisions.The relevant data of the open source software and open source community for effective analysis and mining is the foundation to realize the accurate measurement and the recommendation is conducive to the improvement of community development and maintenance process of the open source and open source software quality and efficiency.Effective analysis and mining of open source software and open source community data is the foundation to realize the accurate measurement and the recommendation,helps improving the efficiency of software development and maintenance,and the quality of open source software in open source community.This thesis takes open source software and open source community as the object,analyze and mine the data of source code and version logs of open source software,interactions between developers and projects,and collaboration between developers in open source community,verify the application of H index in software measurement from the technical(project)and social(developer)level,which provides new guidance for software engineering measurement and recommendation.The specific content and achievements of this paper are mainly embodied in the following aspects:(1)Measurement method of important class in software system based on h indexNew members involved in software project development and maintenance system,often need to spend a lot of time to understand the structure and function of the system.In order to accelerate the understanding of the new members of the system,we usually recommended some key classes to them in system.By constructing undirected weighted software network model,four important classes of metrics measurement software network based on h index are proposed,and compared with the commonly used indexes and some new indexes with better measurement effect are found.The research results have some guiding significance for software development and maintenance.(2)A method for measuring important classes of software system based on variant index of h indexOn the basis of weighted software network model,combined with the degree of neighbor node and the weight of edge,a measurement method of important classes in software network based on variant index of h index is proposed.The relationship and differences between the three proposed indexes and the commonly used indexes in the measurement of important classes are compared and analyzed.It is found that the h index,which combines the degree of neighbor nodes and the weights of edges,is more accurate than the existing indexes in identifying key classes.(3)A method of developer measurement based on H index and analysis of developer cooperation behaviorUnderstanding the collaboration behavior among developers is an important way to understand the collective software development and the organization process,and is also the key to promoting collaborative development of open source communities.We analyze the relationship between the cooperation preference of new developers and the h index of original developers,find that new members choose to cooperate with the original developers who have high value of h index,and analyze the cooperation between original developers in the open source community.The findings will help to develop relevant open source community recommendation tools to promote the stability and sustainable development of the open source community.(4)An approach based on internal link to project recommendation in collective softwareThe rapid development of open source software ecosystem provides a new model for software development,and the coordination of developers and projects is directly related to the efficiency and quality of development.The original developers with cooperative experience in open source community are more likely to cooperate again,and the internal link just conforms to this nature.By analyzing the interaction between developers and projects,we can find out the internal link and recommend the project to the developer.Compared with the collaborative filtering method,the recommendation method based on the internal link is better.This result provides support for improving the efficiency of open source software development.