Identificação de diretrizes para a construção de metamodelos na infraestrutura de MDA.

Abstract

MDA (Model-Driven Architecture) is an approach whose main objective is to shift the effort and time spent during the tests and implementation tasks of the software development lifecycle to modeling, metamodeling and transformations tasks. An entire infrastructure and a set of formalisms have been proposed in the MDA infrastructure, such as UML, OCL, MOF and QVT. By other hand, guidelines are directions to help people with little or none experience in carrying out certain activities. In the computing context, for instance, there are guidelines for best-practices programming and for requirements specification of a system. However, in the MDA infrastructure, the identification of guidelines that assist the elaboration of their artifacts (models, metamodels and transformations) is still incipient and little has been produced. We propose in this work a set of guidelines focused, specifically, on the construction of metamodels. These guidelines were identified by means of two approaches: (i) automatic, by applying the KDD (Knowledge Discovery in Databases) process. In this case, a tool was developed to support some KDD steps; and (ii) manual, by analyzing a set of metamodels. As a result of both approaches, we present a catalog with 13 guidelines documented according to a template based on that proposed by the GoF for design patterns. A tool support was developed to automatically apply the guidelines to existent metamodels. The adoption of these guidelines provides for the metamodels better understanding, construction, maintenance, development and reuse. The evaluation of this work was performed by applying the guidelines to six well-known metamodels in order to analyze the applicability of each guideline.