A Model for Generating Random Networks with Clustering Coefficient Corresponding to Real-World Network Graphs

In this paper, we propose that when a random network graph model that prefers to close two-hop chains and transform them to triangles as part of link formation. We hypothesize that such a model would generate random network graphs with high clustering coefficients and a larger variation in node degree, compared to that of the well-known Erdos-Renyi (ER) model. We refer to the proposed model as two-hop neighbor preference (THNP)-model that prefers to pair a node with any of its two-hop neighbors rather than to an arbitrary node. The probability of link formation is still governed by the formulation used to generate links in the ER model. We observe the THNP-model to generate random network graphs wherein the clustering coefficient of a node decreases with increase in node degree (resembling closely to several of the realworld network graphs), and the graphs still exhibit a Poisson-style distribution for the node degree and path length.