What do we learn from correlations of local and global network properties?

Abstract

In complex networks a common task is to identify the most important or "central" nodes. There are several definitions, often called centrality measures, which often lead to different results. Here we study extensively correlations between four local and global measures namely the degree, the shortest-path-betweenness, the random-walk betweenness and the subgraph centrality on different random-network models like Erdos-Renyi, Small-World and Barabasi-Albert as well as on different real networks like metabolic pathways, social collaborations and computer networks. Correlations are quite different between the real networks and the model networks questioning whether the models really reflect all important properties of the real world.