Finding collaborations based on co-changed files

• Ano de defesa: 2022
• Autor(a) principal: Kattiana Fernandes Constantino
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Minas Gerais; Brasil; ICX - DEPARTAMENTO DE CIÊNCIA DA COMPUTAÇÃO; Programa de Pós-Graduação em Ciência da Computação; UFMG
• Palavras-chave em Português: Open-source software projects; Collaborative software development; Distributed collaboration; Developer recommendation; Computação – Teses; Engenharia de software – Teses; Sistemas abertos (Computadores) – Teses; Software – Desenvolvimento – Sistemas colaborativos - Teses
• Link de acesso: http://hdl.handle.net/1843/63962; https://orcid.org/0000-0003-4511-7504

Software developers must collaborate at all stages of the software life-cycle to create successful software systems. However, for large projects with hundreds of dynamic developers, such as several successful open source projects, it can be very complex to find developers with the same affinity and thus gain suitable collaborations and new insights. Besides, in this project context, resources and efforts may be wasted, discouraging many developers from staying. Therefore, this doctoral thesis proposes an investigation of collaborative development based on similar code interests and tool-supported strategies to help developers find suitable collaborators. We performed five empirical studies: (1) we investigated how collaborations happen in open-source software development projects through an interview study. Some main findings from the interview study include that collaboration transcends coding and includes documentation and management tasks; (2) we designed and performed a survey study to investigate how open developers are to collaborate with others. Some analysis from the survey study revealed that most participants (85%) prefer to work collaboratively with the core team members and 30% prefer to work in independent tasks; (3) we provided two strategies based on co-changed files and a prototype tool, named CoopFinder, that support them; (4) we evaluated these two strategies to motivate collaborations based on changes of similar code of point of view of who receives the developer recommendations. As a result, the acceptance rates for them were greater than 65%. The joint strategy presented the best acceptance rate (81%); and, (5) we also evaluated these strategies and their supporting tool with GitHub users and non-GitHub users. About 86% of the participants answered that they could use or recommend this tool. Based on the results obtained in this doctoral thesis, it is possible that developers and maintainers can acquire the knowledge to foster collaborations in the project and, consequently, avoid emptying it.