Arch-KDM 2.0: checagem de conformidade arquitetural em projetos de modernização dirigida a arquitetura
LANDI, André de Souza. Arch-KDM 2.0: checagem de conformidade arquitetural em projetos de modernização dirigida a arquitetura. 2018. Dissertação (Mestrado em Ciência da Computação) – Universidade Federal de São Carlos, São Carlos, 2018. Disponível em: https://repositorio.ufscar.br/handle/ufscar/10296.
Landi, André de Souza
Architecture-Driven Modernization (ADM) is a software reengineering process based on Object Management Group (OMG) concepts and yours ISO standards. There are several types of modernization projects, such as language-to-language conversion, platform migration, application integration, and application improvements. One necessary activity in scenario of application improvements is architectural reconciliation. An important step of the architectural reconciliation is the Architectural Conformance Checking (ACC) whose objective is to identify architectural drifts. This master's project consists of three main parts, the evolution of an ACC approach already presented in another project called Arch-KDM; a new computational support for the approach; and a formal definition of architectural drifts and violations. The original Arch-KDM approach has three steps that support the ACC proccess. They are the specification of a planned architecture; the extraction of the current architecture; and the accomplishment of the conformance checking. In the course of this project, there were evolutions related to several points of this approach like bug fixes, erroneously use of elements and a new step. Thus, the approach is re-presented in this document with a tool called Arch-KDM 2.0 that includes all evolutions and fixes. Two evaluations were carried out, one in order to evaluate the first step of the approach by means of an empirical study evaluating as an appropriate stage for the specification of planned architectures. The second one was an empirical study evaluating the precision, recall and f-measure. The results obtained were very promising and satisfactory in both evaluations. It was verified that the first step of the approach is appropriate for the specification of planned architectures and for the second evaluation the computational support was able to obtain an accuracy of 78.99% against the manual accuracy, this value is considered a high accuracy.