Spectral Decomposition for Optimal Graph Index Prediction

Recently, there has been ample of research on indexing for structural graph queries. However, as verified by our experiments with a large number of random graphs and scale-free graphs, the performances of indexes of graph queries may vary greatly. Unfortunately, the structures of graph indexes are too often complex and ad-hoc; and deriving an accurate performance model appears a daunting task. As a result, database practitioners may encounter difficulties in choosing the optimal index for their data graphs. In this paper, we address this problem by a spectral decomposition for predicting relative performances of graph indexes. Specifically, given a graph, we compute its spectrum. We propose a similarity function to compare the spectrums of graphs. We adopt a classification algorithm to build a model and a voting algorithm for the prediction of the optimal index. Our empirical studies on a large number of random graphs and scale-free graphs and four structurally distinguishable indexes demonstrate that our spectral decomposition is robust and almost always exhibits accuracies higher than 70%.