Assisting human understanding to aid the targeting of necessary reengineering work

Abstract

Reengineering is a necessary but expensive process. Therefore if ways of reducing the cost of reengineering can be found, then this can help to direct limited budgets to where they are most needed. This paper evaluates a number of ways in which early predictions of the benefits of performing reengineering work can be made. The early predictions estimate whether potential reuse candidates can be obtained from the reengineering process. In particular, this paper is concerned with evaluating approaches to predict the numbers of potential reuse candidates which are available within legacy applications. A number of differing approaches are proposed, each is evaluated. These evaluations range from fully automatic to fully manual tasks. The accuracy of each approach is tested against the results of the application of the full reuse reengineering method. The method used is the IDENT method. The IDENT method is a reuse reengineering approach for the identification and extraction of reuse candidates from existing COBOL legacy systems. An overview of the method is provided within this paper. By comparison of the two sets of results, the accuracy of the early estimation techniques are obtained.