An evidence-based study on issue labeling in Github-based repositories (Supplementary material)

The open-source software community has grown in size and importance over the years. As a consequence, the number of project contributors has increased considerably. The capability of open-source project repositories to accommodate issue reports is essential. An issue report encompasses a large set of data that describes the necessary changes a software should handle. As developers need detailed information to reproduce and find them, incomplete information is a severe problem that may influence triage and defect detection leading to delays in project maintenance. Issue trackers commonly use the labeling method to add extra details to issues. Knowing the importance of labels, this dissertation focus on investigating the usage, creation, and similarities in the context of the issue lifecycle in both maintenance and evolution in the repository issue trackers of the largest and most popular code hosting platform, Github. In addition, it analyzes the number of labeled and unlabeled issues in the repository and the connection between the issues' components, with an analysis focused on the lifecycle. The results indicate a significant correlation between repositories with many issues and the creation of labels, but not all repositories use them. 64.58% of the repositories insert new labels as the project evolves. 73.14% repositories applied on issues the Github standard labels. We also found an influence of primary issue fields such as title, description, and comments in most issue labels, impacting the creation and labeling issues. These numbers show that issue labeling is of prominent relevance for project maintenance and evolution. It provides developers with an easy and convenient way to inform about an incoming issue reported by systems users.