A Dataset of EMF Models from Eclipse Projects

Models are key artefacts in Model-driven software engineering. Data sets of models from practice are highly valuable as input for different modelling research areas, e.g., performance benchmarks for modelling tools and analysing model transformations, as well as in empirical research, e.g., understanding how models are designed and evolve over time. Unfortunately, there is a lack of data sets containing models, their meta models, and their evolution history. We present such a data set and describe our data collection method. The Eclipse modeling framework (EMF) is the major framework for developing and using EMF models providing a rich ecosystem developing many models and meta models. Thus, we mined meta models and their instances from git repositories associated with Eclipse projects (https://www.eclipse.org/projects/, accessed 2018-05-25), including their version history. Our data set was created on 2018-05-25 and contains 31799 models of which 4732 are meta models with a total of 101267 versions. These were mined from 247 repositories belonging to 130 projects hosted on Eclipse Projects