Um modelo de avaliação para projeto de arquitetura de LPS baseado em busca

• Autor(a) principal: Santos, Marcelo Carlos Benitez dos
• Data de Publicação: 2016
• 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/1/2592
• topic: Linha de produto de software (LPS); Arquitetura de linha de produto (ALP); Projeto de arquitetura de LPS; Arquitetura de linha produto (ALP); Modelo de avaliação; Brasil.; Software product line; Product line architecture design; Evaluation model; Brazil.; Ciências Exatas e da Terra; Ciência da Computação

The Product Line Architecture (ALP) is considered the most important artifact in a Software Product Line (LPS), as it includes a joint design to every LPS product. Thus, evaluate and modularize the ALP is essential to get a good LPS design. However, evaluation and modularization of ALP activities are difficult and require great human effort without the aid of an automated tool. It can be easy for an architect to recognize a good ALP design, however, it is difficult to get it. Therefore, MOA4PLA (Multi-Objective Approach for Product-Line Architecture Design), was proposed, which has the purpose to help the architect to evaluate and improve ALP design. This approach is multi-objective and uses an evaluation model composede by evaluation function to assess the ALP design. These functions are based on different metrics, which involve different architectural properties. Currently, the evaluation model presents problems in some evaluation functions, because these functions evaluate the desingn utilizing metrics with different magnitudes, so that the information resultant of the assessment is provided in a more general way. This work therefore proposes to restructure the evaluation model of MOA4PLA, separating the CM (pla) and FM (pla) by its architectural properties, generating new evaluation functions and increasing optimization options. Experimental studies were performed to investigate a possible correlation between the new evaluation functions. The results show that most of the functions are negatively correlated, so that when one function increases the other decreases. Thus the function has to be selected as an objective, if the architect wants to prioritize it in the optimization process.