The document summarizes Danai Koutra's presentation on techniques for understanding large graphs through summarization and measuring similarity. The presentation introduces VoG, a method for graph summarization that finds overlapping substructures to minimize description length. It also presents DeltaCon, an algorithm that measures graph similarity based on pairwise node influence values calculated through belief propagation. DeltaCon runs in time linear to the number of edges. Examples are given of applying these techniques to email networks, wiki graphs, and brain connectivity networks.