Uma abordagem para a seleção de desenvolvedores baseada na reputação e esquecimento em projetos de software

Abstract

Maintenance tasks are essential to keep the software running during its life cycle. However, as software evolves, its source code tends to become increasingly complex and difficult to maintain, making it more likely to present defects to the end-user. As a result, choosing appropriate developers for software maintenance tasks becomes a complex, timeconsuming activity and susceptible to decision failures when performed manually. Because, for suitable developers for the functions, each candidate's expertise must be well known. It is also important to know how much these developers have worked on the knowledge (expertise) in technology required by the tasks, which is verified through their contributions to platforms that support software development, such as GitHub, StackOverflow, and TopCoder. These historical contributions make it possible to establish developers' reputation in their expertise and verify that this knowledge is not in disuse for a long time (forgetfulness). The proposed solution of this work was developed in order to support the selection of developers for software maintenance tasks considering their reputation and forgetfulness in the knowledge of technologies in software projects. For each software task, developers' suggestions are listed, considering reputation and forgetfulness in the knowledge required by the task. An evaluation was conducted to verify these suggestions' equivalence with the developers' historical attributions in software project tasks. Favorable evidence of the use of reputation and forgetfulness was found to support the selection of developers in software maintenance tasks.