Improving Domain-Specific Languages by Analyzing, Constraining and Enhancing Metamodels

Abstract

We present an approach for improving domain-specific modeling languages (DSML) by automatically revealing unintended models and subsequently introducing constraints to disallow these. One purpose with domain-specific modeling is to raise the level of abstraction by restricting application models to be within a domain. A metamodel, describing the concepts of the language, will typically restrict the type of concepts and how they are connected. However, these restrictions are not sufficient since the number of possible illegal models can still be large. Using a formal definition of the static semantics, we generate arbitrary models of a DSML. Based on these models, we show how to incrementally constrain the language to prohibit unintended models. We provide a prototype implementation of the approach, and we apply this prototype to an example in the train domain to illustrate the approach.

Oppdragsgiver: Research Project MoSiS and Research Product VERDE