Social Collaboration Process Recovery And Property Prediction In Open Source Software Development

Nowadays,open source software development is booming,and with a lot of social collaboration,the management of the collaboration process has gradually attracted everyone's attention.The high mobility of developers,the prior uncertainty of collaborative processes,and the looseness of project management in the process of social collaboration make traditional business process management no longer applicable.Team managers do not have an exact standard basis to arrange the human resources,time plans,etc.required for the unfinished collaborative process,and can only rely on previous management experience.Collaborative process management for open source software development has become difficult.In order to solve this problem,this topic takes the bug solving process of open source project software development as an example to carry out research.The specific research consists of the following three parts:(1)Give the mathematical definition of the social collaborative process of open source software development and the basic process of collaborative process recovery.The recovery process consists of five steps: collaborative behavior recognition,bug state division of the collaboration process,classification of collaborative behavior based on random forests,association relationship recognition based on heuristic rules,and classification of associations according to related collaborative behavior.It recoveries the social collaboration process from a flat collaborative behavior sequence to a rich stereoscopic flow chart.(2)The definition and measurement methods of the properties of social collaborative process of software development in open source projects are proposed,which helps the project manager to learn more about all aspects of the ongoing collaborative process and become more familiar with the project team's collaborative habits.In this paper,the properties of the collaborative process are divided into two categories: digital properties and collaborative patterns.Among them,the digital properties is divided into attention degree,difficulty degree,smoothness,each dimension contains a set of related sub-indicators;the collaborative process mode is a directed graph of nodes as local collaborative processes,each node can be obtained by frequent Subgraph mining.The similarity of digital properties and the frequent patterns of the collaborative process between different projects are calculated to facilitate future research on the nature of the collaborative process.(3)A social collaborative process property prediction algorithm for open source software development is proposed,which predicts the digital properties and the frequent patterns that may be included of the collaborative process that has not yet started and is in progress.Firstly,we carry out feature extraction and property calculation of the social collaborative process in the project and in similar projects,and then input into the random forest classifier for training,and finally predict the collaborative properties.The predicted results can help managers better understand the basic situation of the ongoing social collaborative process,in order to better support the evolution of open source software.This paper defines and recoveries the social collaboration process,then calculates the representative digital properties of the collaborative process,mines the frequent patterns of the collaborative process,and finally predicts the digital properties and frequent patterns of the unfinished collaborative process.