Predicting Network Evolution through Temporal Twitter Snapshots for Paris Attacks of 2015

Abstract

As technology advances, modern networks rapidly evolve. Capturing the dynamic nature of networks and predicting their evolution has been a common focus in network science. This research investigates a social network’s temporal evolution, and how metrics and descriptors during its creation compare a snap shot in time during the network’s growth to the known state of the final network. As social media is a primary way of communication, Twitter data collection provide real traces for this study that focuses on the ability to determine if knowing network’s early metrics provide an accurate prediction of the this final network. This can then be extended to monitor other similar events as they are happening. However, this does not generalize arbitrary social evens. Specifically, this research utilizes data from Twitter feeds regarding the Paris terrorist attacks (#ParisAttacks) in November 2015, and focuses on the analysis of k-Core, Betweenness centrality, and community comparison as the network grows. The topology of the overall network after 24 hours from the time of the first post provides the known “end-state” that we compare against.