A First Experimental Evaluation of Search Plan Driven Graph Pattern Matching

Abstract. With graph pattern matching the field of graph transformation (GT) includes an NP-complete subtask. But for real-life applications it is essential that graph pattern matching is performed as fast as possible. This challenge has been attacked by the approach of search plan driven, host-graph-sensitive (also known as model-sensitive) graph pattern matching. To our knowledge no experimental evaluation of this approach has been published yet. We performed first experiments regarding the runtime performance using the well-known GT benchmark introduced by Varr´o et al. as well as an example from compiler construction. Moreover we present an improved cost model and heuristics for search plans and their generation. Keywords. graph transformation, graph pattern matching, subgraph isomorphism problem, search plan driven, host-graph-sensitive, modelsensitive, heuristic optimization, experiment

• Type: Conference
• Year: 2007
• Where: AGTIVE
• Authors: Gernot Veit Batz, Moritz Kroll, Rubino Geiß
• Keywords: AGTIVE 2007, Formal Languages, Graph Pattern Matching, Graph Transformation, Search Plan