Parallelization of Graph Transformation Based on Incremental Pattern Matching

Graph transformation based on incremental pattern matching explicitly stores all occurrences of patterns (left-hand side of rules) and updates this result cache upon model changes. This allows instantaneous pattern queries at the expense of costlier model manipulation and higher memory consumption. Up to now, this incremental approach has considered only sequential execution despite the inherently distributed structure of the underlying match caching mechanism. The paper explores various possibilities of parallelizing graph transformation to harness the power of modern multi-core, multi-processor computing environments: (i) incremental pattern matching enables the concurrent execution of model manipulation and pattern matching; moreover, (ii) pattern matching itself can be parallelized along caches.