Recommending Collaborators Based on Co–Changed Files: A Controlled Experiment

Abstract

Active collaboration is essential for the success of software projects across the development life-cycle. However, social coding platforms, like GitHub, present challenges in finding potential collaborators with whom they could create new/stronger ties and enhance the quality of contributions. Thus, we conducted a controlled experiment with 35 participants. We asked participants to perform the experiment tasks to find collaborators with similar interests using a prototype recommendation tool and GitHub. We observed that results confirm that recommender based on co–changed files can provide suitable collaborator recommendations to developers of a specific project.