Avaliação experimental da abordagem SMarty para gerenciamento de variabilidade em linhas de produto de software baseadas em UML

• Autor(a) principal: Marcolino, Anderson da Silva
• Data de Publicação: 2014
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da Universidade Estadual de Maringá (RI-UEM)
• Texto Completo: http://repositorio.uem.br:8080/jspui/handle/1/2499
• topic: SMarty - Estudos empíricos; Linha de produto de software variabilidade; UML (Linguagem de modelagem unificada); Gerenciamento de variabilidade de software; Ciências Exatas e da Terra; Ciência da Computação

The Software Product Line (SPL) approach allows the reuse of different software artifacts. In order to such artifacts generate specific products from a central repository of a SPL, several elements of these artifacts which vary from one product to another, called variabilities, are managed by the variability management activity. The importance of this activity in SPL refiects the diversity of existing variability management approaches. Stereotype-based Management of Variability (SMarty) is a variability management approach that, by means of its UML 2.0 profile and a systematic process, allows the identification and representation of variability in different UML models. However, SMarty and other existing approaches presented in the literature need effectiveness evidence by means of empirical studies to provide support to the adoption of such approaches in academic and industrial contexts. This work proposes the extension of SMarty 4.0, the SMarty 5.0, for UML sequence models, aiming to represent variability in dynamic levei, and a set of empirical studies that evaluate the SMarty effectiveness compared with other significant existing approaches in the literature. Thus, four studies were conducted with subjects from several universities in Brazil. The results for the experimental studies of use case and sequence mo deis demonstrate the effectiveness of SMarty 5.0 in identifying and representing variability, as the empirical study for class mo deis could not provide any evidence on the Smarty effectiveness. Therefore, Smarty 5.0 evolved to Smarty 5.1, and another experiment was carried out. This new study increased the effectiveness calculated for SMarty, but still below the effectiveness of the compared approach. Further analysis and considerations were draw. By means of correlation techniques, it became evident that the levei of knowledge of the subjects refiected on a smaller infiuence on the application of SMarty 5.1, compared with other approaches, suggesting that its guidelines were crucial for the application of the approach. Thus, SMarty 5.1 shows effective to manage variability in general, increasing the levei of quality and guarantee of generating specific products on SPLs.