A Partition Method For GitHub Software Development Process Log

With the popularity of the Internet,software products are going deep into People's daily life,which puts forward higher requirements for the quality of software products.How to develop high-quality software quickly is becoming a difficult problem in modern software development,and open source software provides great convenience.GitHub is the largest open source software community.Many software projects rely on discussions between software engineers and developers to determine the future direction of software,without the guidance of fixed software process model.Software process model plays an important role in guiding and managing the development of high quality and efficient software products.At present,software process modeling has changed from the traditional method of artificial modeling to the method of automatically mining the software process model from the massive software process data.However,GitHub open source software development process data has certain complexity and inevitable single instance problem,so it is difficult to mine.The existing process mining technology cannot be directly applied to GitHub software development process log mining.This thesis adds a label to the log submitted in the software development process on GitHub,and extracts the log data as activity sequence according to the semantic information.The log is divided from the perspective of global sequence,and a method based on the relationship between global computing activities is proposed to determine the log partition activities.The multi-instance partition is taken as the case information,and the process mining algorithm is used to mine the divided log to find the software development process model.The experiment used the real GitHub open source software to submit log data,and demonstrated in detail the acquisition,format and transformation of single instance event sequence into multiple instances of GitHub open source software development process log.The experiment shows that the global partition method for the event sequence of log formation is feasible,and the method is feasible and correct.The main work in this thesis is as follows:(1)developed new ideas and methods for the research on GitHub open source software development process mining,and takes log as a starting point to analyze data and extract information from log hierarchically.;(2)solved the complexity problem of GitHub software development process log.Through the structured and serialized processing of log,the log can effectively carry out relational calculation,and meanwhile ensure the feasibility of activity mining;(3)judged log partition activity from a global perspective increases global calculation,reduces partition attempts,improves partition efficiency and effectively solves the single instance problem.