Figure 2.
Recall achieved across different mutation rates with increasing graph sizes for MG-Sketch (left) and MG-Align (right). We run MG-Sketch with t = 6, w = 16, s = 8, D = 14, and K = 40 neighbors. Query generation follows the same approach as explained in the section “Synthetic data generation.” We omit the 0% and 5% mutation rates, as both methods achieve almost perfect recall.
