Graph Edit Distance Benchmark

The Graph Edit Distance Benchmark was part of my Ph.D dissertation. The objective of the benchmark is to precisely evaluate the accuracy and the scalability of Graph Edit Distance (GED) methods. Thus, in this benchmark, well-known and public datasets are divided into subsets each of which represents a certain number of vertices. To build the Ground-Truth (GT), exact and approximate GED methods were run under a time constraint that is equal to 300 seconds. In the repository, the provided GT consists of the matching and the distances obtained by the most precise algorithm for each pair of graphs.