A toolkit for real-time analysis of dynamic large-scale networks

Abstract

Networks are used in many research domains to model the relationships between entities. We present a publicly available toolkit to extract graphs from datasets or data streams and to analyse their properties. The graph extraction is based on a set of rules that define the links between entities in a set or stream of self-contained events involving sets of entities. As the extracted graph is dynamic and, moreover, can be spread over multiple machines, we include the class of gossip algorithms to analyse them. In addition, the toolkit also contains algorithms to compute metrics of static snapshots of the dynamic graph.