Title: Trying to Find Potential Matches for a Garbled or Misspelled Name? Get Help from NYMROD

Release Date: 2014-06-18

Document Date: 2008-06-11

Description: This internal NSA posting from the Project Director from the Center for Content Extraction explains how the agency’s automated name recognition tool works: see the Der Spiegel article ‘A’ for Angela: GCHQ and NSA Targeted Private German Companies and Merkel, 29 March 2013.

Document: (S//SI//REL) Trying to Find Potential Matches for a Garbled or Misspelled

Name? Get Help from NYMROD


Project Director, Center for Content Extraction (T1221)

Run Date: 06/11/2008

(S//SI//REL) Have you ever had trouble finding out more about a SIGINT target
with a garbled or misspelled name? The NYMROD system was invented to help

1 of 3

with just that problem. The NYMROD name-matching system, developed by the
Center for Content Extraction (CCE). can accept queries consisting of personal
names, perform a "fuzzy" match of the input name to one or more sets of stored
names, and return a list of potential matches for presentation to the user, NYMROD
finds potential name matches without wildcards and can match across scripts (for
example, between Arabic and Roman scripts),

(S//SI//REL) NYM ROD'S first sets of names stored for matching have all been taken
from intelligence reports from NSA, CIA, and DoD databases, with CREST
(transcript database) names coming on-line soon, Since its initial release in January
2008, analysts have been using NYMROD to find information relating to targets
that would otherwise be tough to track down. The user interface is very simple: you
just type in the name you are looking for, set a matching threshold from 0,1 (very
tolerant of differences) to 1,0 (exact match), select the datase(s) you want to search
within, and submit the query. The results presentation will allow you to browse the
snippets of text that contained the matched name. You can try out this capability
for yourself if you "go NYMROD" in your web browser,

(S//SI//REL) "But I don't want yet another tool!", many of you have said, and the
CCE has taken this to heart, NYMROD also offers a web service that other systems
can use to create their own datasets and perform matching queries. We are
working with partners such as the Target Knowledge Base, the Unified Targeting
Tool, and FASTS COPE to begin integrating the service into those systems so that its
operation is available to their users without anyone having to leave their normal
work environment. Developers can find more information about NYM ROD'S web
service, including an Interface Control Document, on the CCE's documentation web


(U//FOUO) NYMROD's performance is monitored using a quality-assurance
process. We work closely with our commercial and R6 technology suppliers to
continuously improve the underlying matching software so that we can offer
performance enhancements with each quarterly release of NYMROD, Our April
2008 release made large strides in performance on Arabic names, while the
upcoming release in July will offer greatly improved Chinese name matching. The
CCE welcomes your feedback and inquiries. Please contact us at any time using the
CCE HELP e-mail alias.

2 of 3

3 of 3


Click to send permalink to address bar, or right-click to copy permalink.

Un-highlight all Un-highlight selectionu Highlight selectionh