Correlation Properties and Self-similarity of Renormalization Email Networks

A degree-thresholding renormalization method is recently introduced to find topological characteristics of some complex networks. As a matter of fact, the applicability of these characteristics depends on the level or the type of complex networks. Here, a modified version of this original algorithm is presented to unravel ubiquitous characteristics of observed email networks and obtain correct understanding of underlying evolutionary mechanism. Some topology metrics of the email networks under renormalization were analyzed. The results show that renormalization email networks have the power-law distribution with double exponents, are disassortative and become assortative after half of total renormalization steps, have high-clustering coefficients and rich-club phenomena. These characteristics are self-similar both before and after renormalization until half of total renormalization steps, otherwise are self-dissimilar.