Analysis of a Greedy Method for Sampling Social Networks

Investigating the characteristics and parameters in social networks can pave the way for finding important results in different areas of science and reveal various characteristics of societies. An important issue in the study of social networks is the rapid growth of these networks. Networks with millions of members and the connections among them are so huge that it has made their whole simultaneous analysis difficult; hence, the use of the concept of sampling in these networks has been proposed. There are many sampling methods which attempted to find the best sample in which the sampled subnetwork has the most similarity to the original network; however, the similarity is not desirable. In this research, an existing sampling method has been evaluated, which has a greedy approach in finding network vertices. The vertices are chosen based on their degrees for which higher-degree vertices are preferable. The similarity between the sampled networks in this method and the original network is evaluated by two criteria, i.e. KS distance and Normalized Root Mean Square Error (NMSE). The results prove outstanding similarity between the sampled and the original network .