Research On The Role Of Developer In Open Source Software Based On Network Analysis

The research on the role of developer in the open source community has always been a hot research topic in the field of software engineering.Most of the current classification schemes for the role of developer are based on statistical counting,which has a sidedness and limitations.The network analysis in this topic originated from the development of network sciences,compared with the count classification method,it can be more comprehensive analysis of the developer ecosystem in the community.In the experiment,in order to prove the superiority of the network analysis method in this subject,this paper firstly according to the project data of the open source community,uses the counting classification and the network analysis to make a comprehensive comparison of the classification result of the developer role,and verifies the effectiveness and practicability of the network analysis in this subject.Then,in order to expand the application method for the problem of developer classification,this paper builds an interactive network between project files with the idea of network analysis,puts forward a new classification index of the developer role,and proves that it has an accurate advantage over the traditional method in the classification of developers.Finally,in order to get a universal theoretical framework for the problem of developer classification,this paper combines the features extracted by various methods,based on the support vector machine classification algorithm,and establishes a role classifier to analyze and compare the factors that affect the role classification of the developer from a comprehensive perspective.At the same time,according to the conclusion of the role classifier,the combination of multiple classification indexes can effectively improve the performance of the developer classification model.The main contents of this paper include the following parts:(1)The preprocessing of the open source community data and the extraction of related features are discussed,and the validity and reliability of the data are enhanced;(2)The research method of traditional counting classification is analyzed,which proves its usability and its limitations.(3)According to the method of network analysis,the classification performance and the traditional counting method are compared horizontally,proving its effectiveness and accuracy,and at the same time showing its simplicity in application.(4)Based on the idea of network analysis,a new network analysis index is proposed based on the interaction between project files,and its effectiveness is proved to further expand the application of network analysis in this topic.(5)Based on the support vector machine classification algorithm,the role classifier is established,and the model classification performance under different types of indexes is analyzed,and it is proved that the combination of various indexes can effectively improve the performance of the classification model.