A First Step Towards Exact Graph Edit Distance Using Bipartite Graph Matching
In recent years, a powerful approximation framework for graph edit distance computation has been introduced. This particular approximation is based on an optimal assignment of local graph structures which can be established in polynomial time. However, as this approach considers the local structural properties of the graphs only, it yields sub-optimal solutions that overestimate the true edit distance in general. Recently, several attempts for reducing this overestimation have been made. The present paper is a starting point towards the study of sophisticated heuristics that can be integrated in these reduction strategies. These heuristics aim at further improving the overall distance quality while keeping the low computation time of the approximation framework. We propose an iterative version of one of the existing improvement strategies. An experimental evaluation clearly shows that there is large space for further substantial reductions of the overestimation in the existing approximation framework.
KeywordsSearch Tree Edit Distance Tree Node Iterative Version Edit Operation
Unable to display preview. Download preview PDF.
- 12.Burkard, R.E., Dell’Amico, M., Martello, S.: Assignment Problems. SIAM (2009)Google Scholar
- 14.Riesen, K., Bunke, H.: IAM graph database repository for graph based pattern recognition and machine learning. In: da Vitoria Lobo, et al. (eds.) , pp. 287–297Google Scholar
- 15.da Vitoria Lobo, N., Kasparis, T., Roli, F., Kwok, J.T., Georgiopoulos, M., Anagnostopoulos, G.C., Loog, M. (eds.): SSPR&SPR 2008. LNCS, vol. 5342. Springer, Heidelberg (2008)Google Scholar