An Efficient Subgraph Isomorphism Solver for Large Graphs

Abstract

For a given pair of pattern and data graphs, the subgraph isomorphism finding problem locates all instances of the pattern graph into the data graph. For a given subgraph isomorphic image of the pattern graph in a data graph, the set of all ordered pairs of the pattern graph’s vertices and their respective images data graph is called an embedding. Many solvers, such as $\mathrm {Turbo_{ISO}}$ , Glasgow , and VF3 exist in the literature for subgraph isomorphism finding problem. Though each solver aims to minimize computing costs in its own way, computational efficiency is still a central issue for the subgraph isomorphism finding problem. In this paper, we present the development of an efficient solver, SubGlw , for subgraph isomorphism finding which first decomposes data graph into small-size candidate subgraphs using a ranking function and then searches the embeddings of the pattern graph in each of them separately. The ranking function is designed in such a way that it minimizes both number and size of the candidate subgraphs. The performance of SubGlw is empirically evaluated and compared with two state-of-the-art subgraph isomorphism solvers – SubISO and Glasgow over three benchmark datasets – Yeast , Human , and Hprd . The experimental findings reveal that SubGlw performs significantly better in terms of both embedding count and execution time . We have also presented an analysis for identifying saddle point , which is a timeout at which our solver achieves maximum embeddings in least execution time. This analysis provides a better understanding for parameter settings. The source codes of SubGlw can be downloaded from https://github.com/ZubairAliIgraph/SubGlw-master .