Figure 6
From: Uncovering nodes that spread information between communities in social networks
Boundary node detection over time from real time monitoring of events in FearneHolly Twitter discussion. The Twitter activity of a TV show, FearneHolly, is monitored and the Tweets are gathered. The top plot shows the number of boundary nodes that produce a Tweet between time points in blue, and in red is shown the number of Tweets from a random set of nodes of equal size to the boundary node set of the Twitter network. In the bottom plot is the total volume of Tweets over time and we see a single dominant spike in the conversation activity with a decay trailing afterwards. The black dashed line in each subplot is a threshold for spike detection which is one standard deviation from the average Tweet count over time. We can see that the boundary nodes produce a single dominant spike mirroring that of the total conversation activity. The random set of nodes is selected to see whether the boundary node activity simply reflects the total number of Tweets, but we can see a spike in the activity occurs at the start of the monitoring which is not present in either of the other trajectories.