An automatic restructuring approach preserving the behavior of object-oriented designs

Abstract

Work on restructuring object-oriented designs involves metrics for quality estimation, and automated transformations. However, these factors have been treated almost independently of each other. A long-term goal is to define behavior-preserving design transformations, and automate the transformations using metrics. This paper describes an automatic restructuring approach that preserves the behavior of an object-oriented design. Cohesion and coupling metrics, based on abstract models that represent design components and their relationships, are defined to quantify designs and provide criteria for comparing alternative designs. Primitive operations and semantics for restructuring are defined to validate the preservation of design behavior This approach devises a fitness function using cohesion and coupling metrics, and restructures object-oriented designs by applying a genetic algorithm using the fitness function. We empirically evaluate the capability of this approach by applying it to the designs of Java programs and compare our results with a simulated annealing-based approach. Results from our experiments demonstrate that this approach may be useful in improving object-oriented designs automatically.