How README files are structured in open source Java projects

doi:10.1016/j.infsof.2022.106924

Return

How README files are structured in open source Java projects

2022-08-01

PRE

Yuyang Liu

Ehsan Noei *

Kelly Lyons

DOI：10.1016/j.infsof.2022.106924

Original

Original request for help

Save

Abstract

En 中文

Context: Recent studies on open source platforms, such as GitHub, provide insights into how developers engage with software artifacts such as README files. Since README files are usually the first item users interact with in a repository, it is important that README files provide users with the information needed to engage with the corresponding repository.Objective: We investigate and compare README files of open source Java projects on GitHub in order to (i) determine the degree to which README files are aligned with the official guidelines, (ii) identify the common patterns in the structure of README files, and (iii) characterize the relationship between README file structure and popularity of associated repositories.Method: We apply statistical analyzes and clustering methods on 14,901 Java repositories to identify structural patterns of README files and the relationship of README file structure to repository stars. Results: While the majority of README files do not align with the GitHub guidelines, repositories whose README files follow the GitHub guidelines tend to receive more stars. We identify 32 clusters of common README file structures and the features associated with each structure. We show that projects with README files that contain project name, usage information, installation instructions, license information, code snippets, or links to images tend to get more stars. Conclusion: README file structure shares a statistically significant relationship with popularity as measured by number of stars; however, the most frequent README file structures are associated with less popular repositories on GitHub. Our findings can be used to understand the importance of README file structures and their relationship with popularity.

Keywords:

Empirical study

README files

Software popularity

Clustering