A new signal primitive for monitoring complex information networks

Complex information networks provide globally ubiquitous real-time, and unstructured information with the result that humans are overloaded with large volumes of rapidly streaming dynamic data. Complex information networks are characterized by many interacting entities with well- or semi-defined protocols. Intelligent data reduction methods are needed to understand and exploit dimensional variances and dynamics of information networks, thereby improving search, understanding, and situational awareness. We present a new signal primitive called conversation exchange dynamics (CED) that accentuates novel or anomalous patterns in a complex information network, reduces network data, and provides an aggregated primitive upon which to build monitoring and detection systems. Two complex information networks applications that present significant challenges are computer network intrusion detection and social media network analysis. We conclude with a description of ongoing efforts to apply CED to social media network analysis.