Seleção automatizada de componentes de software orientada por métricas estruturais e informações de reúso

• Autor(a) principal: Alexandre Segundo, Jailton Maciel
• Data de Publicação: 2014
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UFPB
• Texto Completo: https://repositorio.ufpb.br/jspui/handle/tede/6131
• topic: DBC; ESBB; seleção de componentes; métricas de software; CBD; SBSE; component selection; software metrics; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO

The great difficulty of selecting software components is still an obstacle to achieving the success of Component-Based Development (CBD). With the growing market for components, the tendency is always to increase the number of options for assembling applications in different contexts, making it impractical to manual selection. Dealing with a problem of gigantic and complex search space, it is required automation performed by optimization techniques. The proposed approach aims to automate the process of selecting components using techniques of Search-Based Software Engineering (SBSE), whose optimization technique is driven by structural metrics (i.e., connections between components of a software architecture) and information reuse (i.e., aggregated values to the component itself). The metrics used in this component selection context are intended to assess the structural perspective of an architectural instance, since they predict possible integration problems between implementations of components produced by third parties. Note that other proposals ignore this perspective and focus only on the component itself. In addition, reuse information can bring an alternative to represent the perception of the developers about the quality attributes of the software components in a reuse scenario as: the degree of consumers' satisfaction who have already purchased this component and the number of downloads of it. The proposed evaluation is carried out through experiments, which are validated by applying statistical tests.