1 / 10

Fifth Semantic Inter-Operability for E-Government Conference

Fifth Semantic Inter-Operability for E-Government Conference. Practical Solutions for Knowledge Discovery Challenges: Knowledge Representation and Complexity Control Alianna J. Maren, Ph.D. Chief Scientist; EagleForce Associates, Inc. ajmaren@theeagleforce.com 703.564.7436.

dulcea
Download Presentation

Fifth Semantic Inter-Operability for E-Government Conference

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Fifth Semantic Inter-Operability for E-Government Conference Practical Solutions for Knowledge Discovery Challenges: Knowledge Representation and Complexity Control Alianna J. Maren, Ph.D. Chief Scientist; EagleForce Associates, Inc. ajmaren@theeagleforce.com 703.564.7436 2010 Corporate Ridge, Suite 550, McLean VA www.theeagleforce.com

  2. GIG Enterprise Services Concept EagleForce Associates

  3. Knowledge Discovery Challenge I N S I G H T Concept Relation- ships EagleForce Associates

  4. Comparative Volumes of Information Processing EagleForce Associates

  5. Representation Levels Solve Biological “Computationally Hard” Problems Multiple Representation Levels Needed for Biological Vision Processing EagleForce Associates

  6. Comparative Volumes of Information Processing Digital Data Processing Biological Vision Processing Gaussian Difference of Gaussians Can use Shannon Information Theory to reduce noise Uses “difference” of larger Gaussian from more narrowly-focused central one to reduce noise *D. Young, “Gaussian masks, scale space, and edge detection,” http://www.cogs.susx.ac.uk/users/davidy/teachvision/vision3.html Similar methods suitable for first-stage processing in both biological and digital knowledge discovery EagleForce Associates

  7. EagleForce Intelligence Development:The 7 Levels of Knowledge Discovery EagleForce Associates

  8. The First Three Representation Layers Concept Association Relationship / Verb Identification “C4: Ali al-Sistani” 1 2 3 4 N 1 “C2: Muqtada al-Sadr” Meet 2 Concept Identification 3 4 Negotiate e.g.: Concept Category 2 N “Muqtada al-Sadr” Agreed e.g., “Muqtada al-Sadr” Insurgent Leader Radical Shi’ite Cleric Iraq, Najaf Religious Leader Devout Following “Muqtada al-Sadr” “Ali al-Sistani” e.g., “Ali al-Sistani” Concept Classes Concept-to-Concept Matches Concept -Relationship- Concept Level 2: Level 1: Level 3: Concept Extraction Concept Association Syntactic Discovery EagleForce Associates

  9. TERRORIST TAXONOMY Tactic: Bombing Target: Hotel Group 3 Group 2 School Government-Related Facility Group 1 Perpetrator Suspected perpetrator member of minority ethnic group with ties to UBL Target Terrorist Bombing Hotel Location Infrastructure Taxonomy Country Taxonomy Communication Taxonomy Weapons and Tactics Taxonomy Other Taxonomical Layers Geospatial Taxonomy Example: Multiple Interacting Taxonomies EagleForce Associates

  10. Summary • Effective knowledge discovery requires an architecture • Multiple representation levels • Increase abstraction with increasing computational complexity • Defer complex processing until truly required • Strong neuromorphic basis; use aspects of brain architecture to inspire computational solution EagleForce Associates

More Related