1: \begin{abstract}
2: We consider the problem of graph generation guided by
3: network statistics, i.e., the generation of graphs which have given
4: values of various numerical measures that characterize networks,
5: such as the clustering coefficient and the number of cycles of
6: given lengths.
7: Algorithms for the generation of synthetic graphs are often based on
8: graph growth models, i.e., rules of adding (and sometimes removing)
9: nodes and edges to a graph that mimic the processes present in real-world
10: networks. While such graph generators are desirable from a
11: theoretical point of view, they are often only able to reproduce a
12: narrow set of properties of real-world networks,
13: resulting in graphs with otherwise unrealistic properties.
14: In this article, we instead evaluate common graph generation
15: algorithms at the task of reproducing the numerical statistics of
16: real-world networks, such as the clustering coefficient, the degree
17: assortativity, and the connectivity.
18: We also propose an iterative algorithm, the Guided Graph Generator,
19: based on a greedy-like procedure that recovers realistic values over a
20: large number of commonly used graph statistics, while at the same time
21: allowing an efficient implementation based on incremental updating of
22: only a small number of subgraph counts. We show that the proposed algorithm
23: outperforms previous graph generation algorithms in terms of the error
24: in the reconstructed graphs for a large number of graph statistics
25: such as the clustering coefficient, the assortativity, the mean node
26: distance, and also evaluate the algorithm in terms of precision, speed
27: of convergence and scalability, and compare it to previous graph
28: generators and models. We also show that the proposed algorithm generates
29: graphs with realistic degree distributions, graph spectra, clustering
30: coefficient distributions, and distance distributions.
31: \end{abstract}
32: