Recuperação de Componentes em Bibliotecas de Software: Uma Abordagem Conexionista

Abstract

This paper presents a reusable component retrieval approach that uses a neural network based on the Kohonen's self-organizing layer and the faceted classification. The network is trained with the objective of correcting similarity relationships between facet terms, enhancing the selection precision in an evolutive way (i.e., based on the results of previous selections). This technique was used on the implementation of the reusable components' search mechanisms of the Memphis environment.