Finding structural patterns in complex networks

Complex networks are widely used to model various real-world systems. However, traditional tools from graph theory are only suitable for the study of individual networks, but not the relationship between them. In this work, we utilize tools from data mining and pattern recognition to study the similarity and difference between large-scale networks. After transforming networks into data clouds, structural patterns of the networks can be discovered by standard data analysis tools. Using this method, we studied Internet and yeast protein-protein interaction networks. We further discussed the network similarity and difference on the mined structural patterns. Network mining methods in this work provide us novel ways for network analysis, which is valuable for future research.