Structural Analysis of Social Streams
Social networks influence our lives from information to entertainment up to elaborate remote social relations. Social networks are characterized by two dimensions: 1) information about content such as author, message, location, time and 2) structural information such as user connections and paths of information diffusion.
Our goal is to extend the limited operations and algorithms of streaming systems to more complex ones in order to be able to handle and analyze social streams. More specifically, our aim is to develop algorithms that can handle the computational state of social streams and analyze structural information.
We analyze data from popular social media like Twitter and utilize state-of-the-art stream processing platforms like Storm.
A particular focus is on the analysis information cascades, tracing how information is spreading.
- Innovationsfond der Universität Freiburg: "Methoden und Infrastrukturen zur Analyse sozialer Medien in Echtzeit", January - December 2013
- Metrics and Algorithms to evaluate information cascades
- Quantitative and Qualitative: What do the metrics reveal about information diffusion, specific users, events etc
- Develop incremental and distributed techniques for real-time computation
- Cascade lifetime calculation and prediction
- Inherently difficult problem: no direct indicator exists, cascade vary by orders of magnitude in size and duration
- Important for many metrics, large-scale problem (100Ks of active cascades in Twitter)
- Determine indicators (strucutural, spatio-temporal) to indicate the remaining lifetime out of the already observed information
- Implement a scalable detection technique
- Distributed Cascade Reconstruction
- Large scale of available social network data and message streams overwhelms traditional, centralized methods
- Design a distributed algorithm out of an existing centralized approach
- Investigate partitioning and replications approaches
- Management of Large-Scale, Evolving Graphs
- Investigate state-of-the-art solution for graph management
As usual: Please ask us for more specific topics!
- Team Project Social Media Analysis: Reconstruction of Information Cascades
- Waleed Butt: Infrastructures for Social Stream Analysis (Master thesis, July 2012- January 2013)