Towards a multilevel model transformation engine

Abstract

Domain specific modelling languages (DSML) are usually defined through fixed level meta modelling tools such as EMF. While this is sufficient for defining languages that has no overlap with other languages, the approach struggles to reuse overlapping parts of an existing language when defining a new language, especially when it comes to the definition of behaviour. Many domain specific languages have a significant overlap with eachother in terms of concepts and behaviour. Multilevel meta modelling is a promising approach to define a family of DSMLs. In this thesis, we aim to define behaviour once on a higher level of abstraction, and reuse it on every DSML which share that behaviour. We use multilevel coupled transformations (MCMT) to define the behaviour, and we present a multilevel transformation engine capable of transforming MCMTs into traditional two-level rules which can be run by existing transformation engines.