An artificial neural network based software reengineering tool for extracting objects

Given the current level of interest in software reengineering and object-oriented methodologies, the possibility of a tool that could reengineer non object-oriented source code (COBOL) to a object-oriented formal specification and then into an implementation might attract widespread commercial interest. In this paper the authors address the issues in reengineering to an object-oriented form; in particular the extraction of objects from existing source code. The paper is divided into three sections. The first section introduces software reengineering and the object-oriented methodology while the reader is referred to key papers in these areas. The second section describes the three components of the software reengineering tool, which are, (1) the decomposer, (2) the information base, and (3) the composers, and a detailed description of the artificial neural network is presented. The third section describes results obtained from attempting the extraction of objects from existing systems