A Google-inspired error-correcting graph matching algorithm

Graphs and graph algorithms are applied in many different areas including civil engineering, telecommunications, bio-informatics and software engineering. While exact graph matching is grounded on a consolidated theory and has well known results, approximate graph matching is still an open research subject. This paper presents an error tolerant approximated graph matching algorithm based on tabu search using the Google-like PageRank algorithm. We report preliminary results obtained on 2 graph data benchmarks. The first one is the TC-15 database [14], a graph data base at the University of Naples, Italy. These graphs are limited to exact matching. The second one is a novel data set of large graphs generated by randomly mutating TC-15 graphs in order to evaluate the performance of our algorithm. Such a mutation approach allows us to gain insight not only about time but also about matching accuracy.