Network Dynamics: Modeling And Generation Of Very Large Heterogeneous Social Networks

One major achievement was the construction of redirection algorithms to efficiently generate large networks with prescribed degree characteristics. A hindered redirection algorithm was shown to reproduce sublinear preferential attachment. Conversely, enhanced redirection leads to highly-dispersed networks that contain multiple macrohubs (degree a finite fraction of the number of network nodes) and exhibit non-extensive scaling. The average number of distinct degrees that appear in a finite network was found to grow algebraically with network size and the underlying distribution is a universal Gaussian. A choice-driven network growth mechanism was formulated in which a new node first identifies a set of target nodes and attaches to either the target with the largest degree (greedy choice), or the target whose degree is not the largest (meek choice). The resulting network exhibits a non-universal power-law degree distribution.