An evolutionary and graph-rewriting based approach to graph generation

摘要: This paper describes an evolutionary computation based graph rewriting approach to generating classes of graphs that exhibit a set desired global features. A rules are used generate, in constructive manner, graphs. Each rule represents transformation from one another. these transformations causes local changes the graph. Probabilities can be assigned which govern frequency with they will applied. By assigning probabilities correctly, generate exhibiting desirable However, choosing correct probability distribution is not easy task for certain and finding settings may represent difficult search space algorithms. In order features, algorithm find suitable assign rules. The fitness function rewards properties. We show, using small base, how range generated.