Investigando abordagens para melhorias nas Escrita de relatórios de bugs.

Abstract

In software development, bug reports vary in quality, completeness, and accuracy in their descriptures. Typically, poorly written accounts are elaborated by usuaries without any technical instruction, which requires more time for bug compression, through discussions between developers and reporters. Meanwhile, despite the fact that there are platforms with tools to assist the publication of the reports, such as Bugzilla, the need for discussions is still frequent, which does necessary more studies to understand the root of this problem. Therefore, this work proposes an investigation in the description fields and the interactions of the users in bug reports, made on the Bugzilla platform, with the goal of identifying patterns in the posts that influence in the time period for solution or disposal of these. In that context, the research will be conducted by quantita-tive study with Machine Learning(ML) that seeks to relate resolution time, openness to closure, with a structuring of the fields belonging to reports, such as: descriptions, comments, priority, whether it was confirmed or no, number of comments, etc. As a result, he found the range of 8 to 25 comments as ideal for resolution of the reports and there was a limitation with the work with the structuring process, thus, for future work one expects to redo the study using a qualitative approach or use a tool external to the structuring.