Extracting Requirements Patterns from Software Repositories

Open Source Software (OSS) hosted in Repositories such as GitHub can be valuable as a source of information for requirements engineers, especially in the apprentice phase of a new application. In this context, we propose a strategy to speed up the discovery of valuable information, since manual search may be time consuming in the vast dataset of GitHub projects. Our strategy is based on the identification of requirements patterns as a way to summarize requirements-related information. Initial work has been performed using Music Application as a domain exemplar, and focusing on the Read-me perspective of GitHub projects. We explore some of our earlier findings, as well as present an overall architecture where patterns will be central in organizing information.