Structured Probabilistic Attributes, Relationships, and Knowledge (SPARK)

Terrorist organizations and other covert groups operate in independent units dispersed throughout the battlespace. Their structure and continuously adapting tactics present a complex and subtle profile that can be difficult to detect and track. In operations against such enemies, detailed depictions of friendly and enemy social networks become as important as maps of the physical terrain in conventional warfare. DAC proposes a system called Structured Probabilistic Attributes, Relationships, and Knowledge (SPARK). The SPARK approach to social network modeling provides four major advantages to current methods: 1) It automates the construction of detailed multi-mode social networks, eliminating the need for time-consuming manual network construction, 2) it models entities as complex systems of attributes and probabilistic rules that, unlike traditional SNA methods, can accommodate uncertainty and incompleteness in the source data, and 3) it provides automated support for reasoning about complex social structures, and 4) it includes an approach to visualizing multi-mode social networks that highlights high-level trends, preserves low-level details, and reduces visual clutter.