Integrating meta-modelling aspects with graph transformation for efficient visual language definition and model manipulation

Abstract

Visual languages (VLs) play a central role in modelling various system aspects. Besides standard languages like UML, a variety of domain-specific languages exist which are the more used the more tool support is available for them. Different kinds of generators have been developed which produce visual modelling environments based on VL specifications. To define a VL, declarative as well as constructive approaches are used. The meta modelling approach is a declarative one where classes of symbols and relations are defined and associated to each other. Constraints describe additional language properties. Defining a VL by a graph grammar, the constructive way is followed where graphs describe the abstract syntax of models and graph rules formulate the language grammar. In this paper, we extend algebraic graph grammars by a node type inheritance concept which opens up the possibility to integrate both approaches by identifying symbol classes with node types and associations with edge types of some graph class. In this way, declarative as well as constructive elements may be used for language definition and model manipulation. Two concrete approaches, the GENGED and the AToM3 approach, illustrate how VLs can be defined and models can be manipulated by the techniques described above. © Springer-Verlag 2004.