Motifs Analysis And Research In Task-oriented Network

Since the middle of 1990 s, the open-source software has become an important part of computer software engineering. Open-source software has gradually become large, mature, stable and more and more welcomed over time. The growing size and complexity of software projects result in the growing of development teams. Understanding the patterns of development, collaboration, and communication behaviors of programmers can provide important insights and lessons for software engineering researchers and programmers.In recent years, social network analysis has become a popular method to analyze the self-organization, the coordination and relationships between programmers in large-scale software development teams. Software development is a social behavior, the researchers constructed several kinds of social networks in software systems based on various social relationships in software development. They utilized the structural properties of social network to analyze the various aspects of software project teams. This thesis focus on Apache software projects, studies the commit, communication and collaboration behaviors of programmers based on social network motif analysis. The programmers’ commit data is exploited in version control repository to build a programming network. The definition and recognition of programming motif is proposed to capture the commit behavior of a programmer, based on which this thesis finds that the files committed by a programmer at a time are usually dependent on each other. The programming network is expended to be a programming and communication network, by integrating programmers’ communication data into programming network, based on which temporal motifs including two types of nodes are proposed. This kind of motifs can be used to reveal programmers’ collaboration and communication patterns. Furthermore, the structural properties in the networks reconstructed by the temporal motifs can be used to establish better statistical models to predict individual and group productivity than those in the original aggregate ones.