Relation generality - A metric for executable class model quality

In this paper, we investigated quality of class models from perspective of executability by focusing on relation between class model and processing code operating on that class model. We proposed new metric - relation generality - for measuring the quality of class models and related that metric to cyclomatic complexity of the processing code.We elaborated how overgeneralized class model leads to hidden, often unnecessary, complexities when processing the model. We applied our metric on three simple case study models modeling the same system in different ways, and have found out that the metric accurately captures the difference in quality between them.