Specification of Graph Translators with Triple Graph Grammars

A. Schürr

	  Data integration is a key issue for any integrated set of
	  software tools where each tool has its own data structures
	  (at least on the a conceptual level), but where we have many
	  interdependencies between these private data structures. A
	  typical CASE environment, for instance, offers tools for
	  manipulation of requirements and software design documents and
	  provides more or less sophisticated assistance for keeping these
	  documents in a consistent state. Up to now almost all of these
	  data consistency observing or preserving integration tools
	  are hand-crafted due to the lack of generic implementation
	  frameworks and the absence of adequate specification
	  formalisms. Triple graph grammars, a proper superset of pair
	  grammars, are intended to fill this gap and to support the
	  specification of interdependencies between graph-like data
	  structures on a very high level. Furthermore, they form
	  a solid fundament for a new machinery for the production
	  of batch-oriented as well as incrementally working data
	  integration tools.