Induced Edge Samplings and Triangle Count Distributions in Large Networks.

This work focuses on distributions of triangle counts per node and edge, as a means for network description, analysis, model building and other tasks. The main interest is in estimating these distributions through sampling, especially for large networks. Suitable sampling schemes for this are introduced and also adapted to the situations where network access is restricted or streaming data of edges are available. Estimation under the proposed sampling schemes is studied through several methods, and examined on simulated and real-world networks.