2012
DHSS 2012 Proceeding
Evidence Marshalling with Inference Networks: An Application to Homeland Security
Authors: Ken McNaught, Peter Sutovsky
Abstract
When trying to reason about some adversary?s likely intentions, an intelligence analyst frequently needs to combine multiple pieces of evidence observed at different times and having different degrees of relevance, coming from sources with varying degrees of credibility. The evidence marshalling process concerns the structuring of evidence to help analysts and investigators organize their thinking and make better sense of a situation. Here we show how the qualitative structure of a Bayesian network offers a useful approach to evidence marshalling. We propose a framework consisting of four types of nodes, arranged in layers ? hypothesis nodes, ground truth nodes, evidence nodes and credibility nodes. An example is presented in the context of homeland security.
I3M Scientific Sponsors
I3M Industrial Sponsors
I3M Media Sponsors
