The Edge of Graph Transformation - Graphs for Behavioural Specification

The title of this paper, besides being a pun, can be taken to mean either the frontier of research in graph transformation, or the advantage of using graph transformation. To focus on the latter: Why should anyone not already educated in the field adopt graph transformation-based methods, rather than a mainstream modelling language or a process algebra; or vice versa, what is holding potential users back? These questions can be further refined by focusing on particular aspects like usability (available tools) or power (available theory). In this paper, we take a fresh and honest look at these issues. Our perspective is the use of graph transformation as a formalism for the specification and analysis of system behaviour. There is no question that the general nature of graphs is at once their prime selling point (essentially everything can be specified in terms of graphs) and their main drawback (the manipulation of graphs is complex, and many properties that are useful in more specialised formalisms no longer hold for general graphs). The outcome of this paper is a series of recommendations that can be used to outline a research and development programme for the coming decade. This may help to stimulate the continued and increasing acceptance of graph transformation within the rest of the scientific community, thereby ensuring research that is relevant, innovative and on the edge.