Intelligent Software Agent Technology USPACOM Success with Virtual Information Processing Agent Research (VIPAR) at the Virtual Information Center

1
Intelligent Software Agent TechnologyUSPACOM
Success with Virtual Information Processing Agent
Research (VIPAR) at the Virtual Information Center

• Thomas E. Potok, Ph.D.
• Applied Software Engineering Research Group
  Leader
• Computational Sciences and Engineering Division
• Oak Ridge National Laboratory
• Research Team
• Mark Elmore, Joel Reed, Jim Treadwell

2
Oak Ridge National Laboratory

• Established in 1943 for the World War II
  Manhattan Project.
• ORNL today pioneers the development of new energy
  sources, technologies, and materials
• The advancement of knowledge in
• Biological, Chemical,
• Computational, Engineering,
• Environmental, Physical, and Social Sciences. 
• Budget 870 million, 80 Department of Energy,
  20 work for others.
• 3800 employees, 1500 scientists and engineers

3
Technology Challenge

Can we effectively process increasing volumes of
data?
Agents
Sensors
Multimedia
Data
Image
Text
Binary
One small step for man
11010010
1970
1980
1990
2000
2010
Time
4
Simple Agent Example
Agent, find me the book War and Peace, and I
need it tomorrow
Amazon
2 Days 18.50
Dedicated Agents
Barnes and Nobel
1 Day 21.75
B. Dalton
1 Day 20.25
Does the agent understand buying books?
Form a plan to buy the book
5
Simple Agent Example
Agent, find me the book War and Peace, and I
need it tomorrow
Amazon
2 Days 18.50
Dedicated Agents
Barnes and Nobel
1 Day 21.75
B. Dalton
1 Day 20.25
Agent Communities
Library
1 Day Free
Does the agent understand buying books?
Form a plan to buy the book
eBay
1 Day 12.50
Execute the plan
On-Line www.online-literature.com
Now Free
6
USPACOM Virtual Information Center
Manually gather Information from 2030 newspapers
Manually Organize Information
Intellectually Analyze And Summarize Information
Inside China
CINC Summary
Russia Today
Jakarta Post
Summary
Analysts
Unorganized Newspaper Articles
7
USPACOM Virtual Information Center
Manually gather Information from 2030 newspapers
Manually Organize Information
Intellectually Analyze And Summarize Information
Inside China
CINC Summary
Russia Today
Jakarta Post
Summary
Analysts
Unorganized Newspaper Articles
What if technology could do this for the analyst
in real time?
8
New Intelligent Agent Approach

• Gathers and organizes 10,000 article per day the
  articles for the analysts
• Can increase the capability of an analyst by 10
  fold
• Leverage the analysts intellectual capacity for
  more precise decision support

Technologically Gather, Organize, and
Summarize Thousands of articles
Intellectually Analyze And Summarize Information
Inside China
Russia Today
Jakarta Post
Summary
Analysts
Article Retrieval Agents
Organize and summarize Articles
9
VIPAR Technology

• Agglomerative hierarchical clustering
• Dynamic vector space model
• Entropy based term frequency measures
• Resource Description Framework (RDF) agent
  ontologies
• K-means knowledge discovery
• Phylips tree rendering

10
Clustering A Simple Example
2. Terms Dictionary
1. Original and Parsed Documents
3. Vector Space Model
4. Document Similarity
5. Euclidean Distance
7. Phylips tree representation
6. Agglomerative Hierarchical Clustering
8
1
2
3
4
6
7
11
VIPAR Example Taliban Search 15 FEB 2002
Taleban

• Accountability in Afghanistan, FRIENDLY FIRE
• U.S. paid off warlords
• U.S. Taliban fighter will remain in custody
• Army Troops Fired Upon in Kandahar
• Lindh Pleads Not Guilty to Terror Aid
• Kabul warns of Taliban, al-Qaida holdouts
• American Taliban pleads not guilty
• World Scene February 11, 2002 Combined
  dispatches and staff reports
• Geneva rules for Taliban
• Bending the Geneva Rules
• Musharraf concedes rivalry renewed with Iran
• 'Axis' speech starts paying off
• Bush ties drug use to terrorist support
• Troops search site for dead al Qaeda
• Rumsfeld assails POW critics

12
VIPAR Knowledge Management The analysts Edge
Same result, easier to see

• Prisoners
• POWs
• Rumsfeld assails POW critics
• Geneva rules for Taliban
• Bending the Geneva Rules
• American Taliban
• U.S. Taliban fighter will remain in custody
• Lindh Pleads Not Guilty to Terror Aid
• American Taliban pleads not guilty
• Afghan Region
• Military Action
• Accountability in Afghanistan, FRIENDLY FIRE
• Army Troops Fired Upon in Kandahar
• Kabul warns of Taliban, al-Qaida holdouts
• Troops search site for dead al Qaeda
• Afghanistan
• World Scene February 11, 2002 Combined
  dispatches and staff reports
• U.S. paid off warlords
• Bush ties drug use to terrorist support

Prisoners
Military Action
POWs
Afghanistan
Other Middle Eastern Countries
American Taliban
Afghan Region
13
Endorsements

• US Pacific Commands CINC Admiral Blair
• A tremendously successful project
• Software agents lead to substantially improved
  analytical products.
• Science and Technology Advisor
• a grand slam home run!
• first time we've seen information discovery and
  knowledge management software working at HQ
  USCINCPAC operationally.

14
Successful Projects

• We have extensive expertise in agent development

Scientific Data Management
VIPAR Knowledge Discovery
Supply Chain Management Agent System
Manufacturing Emulation Agent System
Collaborative Management Environment

• Over 10 successful projects within the last 5
  years

1985
1990
1995
2000
15
Cyber Counterintelligence Example

• 58,238 records of errors when accessing ORNLs
  network on one day in September.
• If printed, over 5000 pages worth of data
• A Critical Decision
• Is someone trying to hack into the system?
• Who are the malicious users?
• What can I ignore?
• How can I analyze this with current tools?
• What elements of this data are crucial?

16
Agent Technology Detecting Usage Patterns
Pattern of person or persons intentionally
hiding their identity
Summary of 58,000 outside hits of the ORNL website
Anonymous users potentially attacking the
system
Normal web user errors Page not found by
user
Potential attacks of internal e-mail systems
Potential pattern of attack searching for
network weaknesses
17
Information Challenge in Modern Command and
Control Environments
Troop assessment Damage assessments Casualty
reports Logistics report Enemy intelligence Networ
k Compromise Anthrax detection
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
DECISION!
Signals and Data
Decision
18
Agents 2007
Troop assessment Damage assessments Casualty
reports Logistics report Enemy intelligence Networ
k Compromise Anthrax detection
Intelligent Agents
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Intelligent Agent
Low Ammo
Video Feeds GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Video Feeds GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Intelligent Agent
Anthrax detected (Medical Records) Enemy Near
(see Video) Support ammo low
Intelligent Agents
Off Course
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Intelligent Agent
Enemy Position
DECISION
Information
Decision
19
Agents 2012
Decision Analysis What if
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Intelligent Agent
Low Ammo
Video Feeds GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Video Feeds GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Intelligent Agent
Anthrax detected (Medical Records) Enemy Near
(see Video) Support ammo low
Intelligent Agents

• Anthrax attack is real?
• 13 of troops will die
• 10,000 civilian will die
• We engage the enemy?
• 70 chance of victory
• With air support 90

Off Course
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Intelligent Agent
Enemy Position
Information
Decision
Decision Modeling and Support Agents
Manpower Agent
Intelligence Agent
Operations Agent
Logistics Agent
Strategic Plans Agent
C4 Systems Agent
Operational Plans Agent
Force Structure Agent
20
Agents Beyond 2012

• What if
• Anthrax attack is real?
• 13 of troops will die
• 10,000 civilian will die
• We engage the enemy?
• 70 chance of victory
• With air support 90

Decision Analysis What if?
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Intelligent Agent
Low Ammo
Video Feeds GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Video Feeds GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Intelligent Agent
How do we prevent future Anthrax attacks? What
other attacks are likely?
Anthrax detected (Medical Records) Enemy Near
(see Video) Support ammo low
Intelligent Agents
Off Course
Video Feed GPS Ammo level sensor Gun
sensor Clothing sensors Battlefield sensors PDA
Orders Maps Personnel info Enemy info Battle
info Customs
Intelligent Agent
Enemy Position
21
Summary

• Intelligent software agents are a significant
  breakthrough technology
• We have a progression of significantly
  successful, deployed, agent systems and research
  to our credit
• Our systems represent a unique discrimination
  from state-of-the-practice, common, data mining
• Enhancements would focus on military-specific
  needs, e.g. languages, types of sources/data,
  decision aides, manpower savings

22
Contact Information
Thomas E. Potok, Ph.D. Potokte_at_ornl.gov 865-574-08
34