Generating Large EMF Models Efficiently
Nebras Nassar, Jens Kosiol, Timo Kehrer, Gabriele Taentzer · Lecture notes in computer science · 2020
There is a growing need for the automated generation of instance models to evaluate model-driven engineering techniques. Depending on a chosen application scenario, a model generator has to fulfill different requirements: As a modeling language is usually defined by a meta-model, all generated models are expected to conform to their meta-models . For performance tests of model-driven engineering techniques, the efficient generation of large models should be supported. When generating several models, the resulting set of models should show some diversity . Interactive model generation may help in producing relevant models. In this paper, we present a rule-based, configurable approach to automate model generation which addresses the stated requirements. Our model generator produces valid instance models of meta-models with multiplicities conforming to the Eclipse Modeling Framework (EMF). An evaluation of the model generator shows that large EMF models (with up to half a million elements) can be produced. Since the model generation is rule-based, it can be configured beforehand or during the generation process to produce sets of models that are diverse to a certain extent.