ארנון שטורם

UML class diagrams play a central role in modeling activities, and it is essential that class diagrams keep their high quality all along a product life cycle. Correctness problems in class diagrams are mainly caused by complex interactions among class-diagram constraints. Detection, identification, and repair of such problems require background training. In order to improve modelers’ capabilities in these directions, we have constructed a catalog of anti-patterns of correctness and quality problems in class diagrams, where an anti-pattern analyzes a typical constraint interaction that causes a correctness or a quality problem and suggests possible repairs. This paper argues that exposure to correctness anti-patterns improves modeling capabilities. The paper introduces the catalog and its pattern language, and describes experiments that test the impact of awareness to modeling problems in class diagrams (via concrete examples and anti-patterns) on the analysis capabilities of modelers. The experiments show that increased awareness implies increased identification. The improvement is remarkably noticed when the awareness is stimulated by anti-patterns, rather than by concrete examples.