Graph Mining Techniques: Focusing on discriminating between real and synthetic graphs

Abstract

Graphs appear in several settings, like social networks, recommendation systems, computer communication networks, gene/protein biological networks, among others. A large amount of graph patterns, as well as graph generator models that mimic such patterns have been proposed over the last years. However, a deep and recurring question still remains: “What is a good pattern?” The answer is related to finding a pattern or a tool able to help distinguishing between actual real-world and fake graphs. Here we explore the ability of ShatterPlots, a simple and powerful algorithm to tease out patterns of real graphs, helping us to spot fake/masked graphs. The idea is to force a graph to reach a critical (“Shattering”) point, randomly deleting edges, and study its properties at that point.