|
Activity Number:
|
424
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statisticians in Defense and National Security
|
| Abstract - #308479 |
|
Title:
|
An Optimization Approach for Identity Matching in the Federal Sector
|
|
Author(s):
|
David Lloyd*+ and Keith Miller+
|
|
Companies:
|
MITRE Corporation and MITRE Corporation
|
|
Address:
|
7515 Colshire Dr, McLean, VA, 22102-7508, 7515 Colshire Dr, McLean, VA, 22102-7508,
|
|
Keywords:
|
identity ; matching ; entity resolution ; search algorithms ; search performance metrics ; name matching
|
|
Abstract:
|
The approach incorporates the use of ground-truthed control data sets along with the use of sample sets of adjudicated matching tests carried out by separate matching validation groups as a basis for the computation of separate F scores for each search algorithm used. An overall F score is then calculated from the application of all combinations of search algorithms against a target set of queries used with the target identity database comparable to the database used in the Federal Agency. Although the approach is modified to include the use of other identity attributes in order to optimize large-scale processing, the primary intent of the approach is to reduce the risks associated with the name-matching component of the identity matching process.
|
