Copying Subgraphs Within Model Repositories

The set of operations in state-of-the-art graph transformation tools allows one to conditionally create and remove nodes and edges from input graphs. Node attributes can be initialized or updated with information from other attributes, parameters or constants. These operations appear to be too restricted for expressing model refinements in a concise manner. More specifically, graph transformation lacks an operation for copying subgraphs (multiple connected nodes, including their attributes) to a new location in the host graph. This paper presents a case study that illustrates the need, a syntax and an informal semantics for such an operation. It also discusses how the operation was integrated in an existing graph transformation language. Finally, it indicates how our ongoing effort towards the implementation of a model transformation language based on graph transformation makes optimal reuse of evaluation code for existing language constructs.