Deducing model metrics from meta models

Abstract

The use of model-based software development has become more and more popular because it aims to increase the quality of software development. Therefore, the number and the size of model instances are cumulatively growing and software quality and quality assurance consequently lead back to the quality and quality assurance of the involved models. For model quality assurance, several quality aspects can be checked by the use of dedicated metrics. However, when using a domain specific modeling language, the manual creation of metrics for each specific domain is a repetitive and tedious process. In this paper, we present an approach to derive basic model metrics for any given modeling language by defining metric patterns typed by the corresponding meta-meta model. We discuss several concrete patterns and present an Eclipse-based tool which automates the process of basic model metrics derivation, generation, and calculation.