Predictive analysis of massive streaming graphs

Predictive analysis of massive streaming graphs
Data Analytics


Predictive analysis of massive streaming graphs

Rich interaction data is everywhere. Computer networks carry massive data flows that may hide illicit traffic or people interacting through online and other social networks and valuable information that impacts safety and security concerns may be missed. Static analysis of a network at one point in time does not suffice for many security applications. Current methods investigate historical data and build models on what already has been seen. Adapting to new behavior by re-analyzing old data introduces a delay in response.

Streaming graph data represents the changing relationships in social networks, financial transactions, communication and data transfers. Using a streaming graph framework, STINGER, the research team at CAOE is developing new methods that will provide critical information as the data changes rather than waiting for forensic analysis.

Rapidly analyzing, detecting and predicting unusual behavior in these rich and interrelated data sets gives us a real-time window to predict outcomes and respond to rapidly changing situations.

Research Leadership Team

Principal Investigator: David Bader, Georgia Institute of Technology