Controllability of an interaction network constructed from an event stream

Abstract

Event streams provide temporal data that can be represented by networks. Here, we construct an interaction network from an event stream taken from the Twitter public API stream, where nodes represent users and edges represent activity between users. We characterize the network based on the component sizes and degree distribution and show both follows power-law distributions. We also measure its structural controllability based on the minimum driver node density. The construction and analysis of the network allows us to analyze an event stream structurally and provide insights that can only be derived from its network representation.