Metadados para apoio à recuperação de componentes de software baseada em agrupamento por rede neural

Abstract

New software development approaches have been proposed in order to permit the increasing of productivity and application quality. Among these approaches, the software development based on components can be evidenced. This approach is characterized by the software component reuse. In order to permit this reuse, it is necessary a repository which is responsible for storing, searching and retrieving software components. In this context, this work aims to identify and model component metadata on databases, allowing the definition of eficient search strategies to retrieve such components. For reaching these goals, information retrieval concepts and the ART-2A self-organizing artificial neural network architecture were adopted. Experiments were conduced to conrm the viability of this proposal.