A Metric-Based Approach for Anti-pattern Detection in UML Designs

Anti-patterns are poor solutions of recurring design problems, which decrease software quality. Numerous anti-patterns have been outlined in the literature as violations of various quality rules. Most of these anti-patterns have been defined in terms of code quality metrics. However, identifying anti-patterns at the design level would improve considerably the code quality and substantially reduce the cost of correcting their effects during the coding and maintenance phases. Within this context, we propose an approach that identifies anti-patterns in UML designs through the use of existing and newly defined quality metrics. Operating at the design level, our approach examines structural and behavioral information through the class and sequence diagrams. It is illustrated through five, well-known anti-patterns: Blob, Lava Flow, Functional Decomposition, Poltergeists, and Swiss Army Knife.