Analysis of (social) streams and graphs

Modern social media like micromessaging, web logs or social networks allow us to perform a large number of social media analyses, requiring architectures and systems for distributed realtime computations on (social) streams and graphs. The large number of participants (more than 2 billion active users of Facebook) and interactions (around 500 million messages on Twitter per day) as well as the public, often instantaneous availability of data allows us (but also require us) to analyze current events in “live” manner and also perform predictions.

Our research focuses on Information Diffusion, tracing and understanding on which ways a particular piece of information (like a message, a picture or a term) is spreading, which roles particular users are playing and how popularity is affected (“virality”).

Computing diffusion pathways in realtime enables new applications in areas such as online journalism since both sources and paths of information become traceable and the own influence can be understood precisely. In turn, this allows users to assess trust and relevance of information, which so far to be done manually and thus could not keep up with the volume of information and the speed of existing detection tools.