GitHub Visualizations and their Usefulness for GitHub Users

Abstract: The most popular versioning system is undoubtedly Git, for which GitHub with its 40 million users provides a central location at which Git repositories can be discovered and contributed to. With all of the metadata that Git repositories contain, GitHub also holds a significant amount of data in its 44 million repositories which can be accessed freely with the GitHub API. To make use of this data, this report set out to develop and evaluate visualizations of GitHub repositories to find out how well these visualizations help GitHub users understand their projects. The report also tried to determine if there were any specific GitHub metadata that the current tools do not visualize and whether it would be useful to visualize that data. The currently available tools were first gathered in an iterative search process, which yielded three results: GitHub Visualizer by Artem Zubkov, Gource and GitHub Visualizer by Aw Young Qingzhuo. These tools were then evaluated using carefully developed evaluation criteria based on previous research. The report then developed a new tool called GitViz, which visualized more metadata than the current tools, in the form of GitHub issues. GitViz was then subjected to human evaluation so that it could be determined whether it was useful or not. The results showed that while GitViz had some use in providing an overview of the file structure of repositories, there was no apparent practical use for the tool that was significant enough that users would go through the trouble of using it. Furthermore, the addition of GitHub issues did not seem to appeal to the user base. The report concludes that GitHub visualizations have the ability to give a shallow overview of a repository’s commit history and file structure, and that visualizations of other metadata such as issues only appeals to a minimal part of the user base.