Data Sciences Technology for Homeland Security Information Management and Knowledge Discovery

Kolda, T ; Brown, D ; Corones, J ; Critchlow, T ; Eliassi-Rad, T ; Getoor, L ; Hendrickson, B ; Kumar, V ; Lambert, D ; Matarazzo, C ; McCurley, K ; Merrill, M ; Samatova, N ; Speck, D ; Srikant, R ; Thomas, J ; Wertheimer, M ; Wong, P C

Lawrence Livermore National Laboratory

关键词: Decision Making;    Management;    99 General And Miscellaneous//Mathematics, Computing, And Information Science;    99 General And Miscellaneous//Mathematics, Computing, And Information Science;    Security;   

DOI  :  10.2172/917886
RP-ID  :  UCRL-TR-208926
RP-ID  :  W-7405-ENG-48
RP-ID  :  917886

美国|英语

来源: UNT Digital Library

【 摘 要 】

The Department of Homeland Security (DHS) has vast amounts of data available, but its ultimate value cannot be realized without powerful technologies for knowledge discovery to enable better decision making by analysts. Past evidence has shown that terrorist activities leave detectable footprints, but these footprints generally have not been discovered until the opportunity for maximum benefit has passed. The challenge faced by the DHS is to discover the money transfers, border crossings, and other activities in advance of an attack and use that information to identify potential threats and vulnerabilities. The data to be analyzed by DHS comes from many sources ranging from news feeds, to raw sensors, to intelligence reports, and more. The amount of data is staggering; some estimates place the number of entities to be processed at 1015. The uses for the data are varied as well, including entity tracking over space and time, identifying complex and evolving relationships between entities, and identifying organization structure, to name a few. Because they are ideal for representing relationship and linkage information, semantic graphs have emerged as a key technology for fusing and organizing DHS data. A semantic graph organizes relational data by using nodes to represent entities and edges to connect related entities. Hidden relationships in the data are then uncovered by examining the structure and properties of the semantic graph.