Nuclear Forensics: A Scientific Search Problem

Nuclear forensics discovery is the tracking of origins of nuclear materials. If a significant amount of smuggled nuclear material is seized, can its origin be traced to both track down the would-be terrorists and to prevent further illicit activities? Fred Gey and Ray Larson have just received a $300K grant from National Science Foundation for the project "Recasting Nuclear Forensics Discovery as a Digital Library Search Problem," as part of the Domestic Nuclear Detection Office Academic Research Initiative (in cooperation with Department of Homeland Security).

This talk will begin by discussing the nature of scientific search. Is scientific search: (1) search of the scientific literature; (2) search for scientific/engineering discoveries (e.g. Patent Search); or (3) a process of 'needle in haystack' search in a technical domain (e.g. fingerprint or DNA matching)? Are there more categories than these three? This project falls into category 3. It approach nuclear forensics as a specialized search problem. The 'signatures' of seized smuggled nuclear samples are to be matched against large libraries of radio-chemically analyzed existing samples (up to 100,000 records) to discover the geolocation and origin of the seized sample. The project will develop new models of discovery: (1) graph matching on nuclear decay graphs, (2) machine-learning of classification to find geo-locations of sample origins, and (3) rule-based matching capturing the processes by which human forensics experts would identify matches. Educational materials will be developed to bridge between search professionals and nuclear chemists and make each research community aware of the challenges and benefits of interdisciplinary collaboration.

See the award announcement