Augmenting Homeland Intelligence Through Emergent Semantics
Ira Monarch, David Fisher, Pranab Nag · Infotech@Aerospace · 2005
Effective intelligence work must uncover possible scenarios, however improbable, that need to be tracked until they can be eliminated from consideration. What is most important is identifying and maintaining contact with improbable hypotheses or ones that are inconsistent with one's own or ones that appear to have high probability. Improbable patterns must also be maintained as information accumulates, and not be drowned out by mounting evidence for other patterns. All patterns, both probable and improbable, need to be tracked and evaluated as they evolve. The paper will focus on methods, techniques and technologies for capturing the emerging semantics of accumulating information emphasizing empirically-based bottom-up text processing that suggests semantic relations based on term associations. It will also discuss combining text-processing with disciplined top-down semantic formalisms (such as ontologies) and maintenance and use of outlier data that can be combined with subsequent data leading to potential new discoveries. We will also note that the complexity and sheer volume of information requires that computational tools are only one part of the infrastructure, processes and interactions needed. The paper and presentation will describe both completed and proposed work.