IDD - Information Diffusion Detection
-
Contact: Clemens Deußer
Description
Hypes as defined in this project are stories that emerge from social networks and rapidly gain popularity. They can be as harmless as disagreeing on the true color of a dress and they can ruin lives, they can become tremendously popular all over the world or stay smaller than a town. The reasons for why one thing becomes a viral event and another doesn't are diverse and fuzzy; the time of day, a slight change in tone and a random visitor are just some of the near coincidental factors that can determine a hype. The ability to predict hypes, however, as difficult a task as that is, could be a valuable tool. Law enforcement for instance could use such information to quickly react to mob events or companies could react to brand damaging hypes (such as unintentionally offensive advertisement) before they reach a wider audience.
In this project we will explore the idea that hypes are caused or magnified by influential individuals. We will identify characteristics of such influencers and correlate them with graph metrics that we can access at a large scale. Besides exploring the nature of such hypes, our contributions will primarily be toward the efficient calculation of metrics on dynamic graphs, since any practical monitoring solution of social networks will necessarily involve dynamic graph calculations.
This project is funded by the "Deutsche Forschungsgemeinschaft" (DFG) and is conducted in cooperation with the electronic markets chair of the TU Darmstadt.
© DFG
© emarkets
© TU Darmstadt