Data mining algorithms have been recently applied to software repositories to help on the maintenance of evolving software systems. In the past, information about what classes changed together, obtained by mining software repositories, were used to guide future changes. We use this information to measure the possible impacts of a proposed change. In this paper we propose and compare two approaches for sorting impact analysis results that use two different data mining algorithms: Apriori and DAR. Even though Apriori is a classic and largely used algorithm, the case study shows that the approach with DAR is less complex and more suitable for measuring the impacts of a change.
http://www.lbd.dcc.ufmg.br/colecoes/sbbd/2008/015.pdfCaso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web