Lockheed Martin ATL and XMDR Jan 05

ATL & XMDRTechnologies Overview(Developed and Future Pursuits): 

ATL & XMDR Technologies Overview (Developed and Future Pursuits) Benjamin Ashpole bashpole@atl.lmco.com 856-792-9744 Dr. Raj Kant rkant@atl.lmco.com 856-792-9730 http://www.atl.external.lmco.com/projects/ontology/

Introduction: 

Introduction ATL Overview Technology Topics Ontology Alignment prototype Software Agents technology COgnitive Algorithm Composition Handler (COACH) concept Dynamic and Static Application Analysis Service-2-Service matchmaking Explanation generation Service Navigation & Execution Security, Authentication Evaluation I3Con (2004) EON (2004) IC2, STS, MS2 (planned for 2005) ATL and XMDR: Goals

Advanced Technology Laboratories Overview: 

Advanced Technology Laboratories Overview Jim Marsh, Director (856) 792-9820 jmarsh@atl.lmco.com

Advanced Technology Laboratories … converting research into solutions: 

Our mission … Solve world class information technology problems Provide a consistent stream of technology discriminators for military applications Our formula … Advanced technology Innovation in advanced computing and intelligent software Exploitation and hardening of emerging technologies Domain expertise Path to a product Integrated solutions with quantified payoff Proven technology transition … leads to paradigm-changing payoffs Add FCS Picture? Advanced Technology Laboratories … converting research into solutions

Advanced Technology Laboratories: 

Advanced Technology Laboratories Established in 1929 Key location: Cherry Hill, NJ 88K sq. ft. Multiple labs (up to TS/SCI) Core capability: Advanced Information Technology Intelligent systems Information architectures Embedded systems Wireless communications Complex system simulation/analysis Lockheed Martin 26% DARPA 37% Other 4% 2003 Customers Full Time Employees = 148 as of 10/4/04 Non-ATL/Interns/Visiting Researchers = 14 Total = 162 Technologists - 78% Management 7% Other 15% Gov. Labs 25% • PhD 17% • MS 49% • BS 34% Classified 8%

Emerging Challenges Require Compelling Technology: 

Netted, Embedded & Complex Systems Advanced Networking Wireless Communications Adv Signal Proc & Embedded Proc. Complex Systems Intelligent Systems Network Centric Operations Autonomy and Collaboration Situation Understanding Decision Support Adaptive Information Systems Dynamic Info. Integration Information Extraction/Exploitation Cognitive Computing Network Mission Assurance National Missile Defense Time Critical Strike Homeland Security Anti-Terrorism IW NBC Joint/Coalition Operations Rapid Response Info Superiority Force Multipliers Collaborative Auton. Vehicles Human Augment. Emerging Challenges Require Compelling Technology Business Areas

Current DARPA Programs andTransition Targets: 

Expanding CRAD Base Across DARPA & Government Labs LM Current DARPA Programs and Transition Targets Technology Focus Human Machine Interaction Situation Understanding Plan Understanding & Monitoring Autonomy & Teaming Information Protection Agent-Based Systems Composable Simulation Network Centric Enablers IPTO Aug Cog ASSIST COORDI-NATORs RAP Teams CRABS FAST C2AP FM-UAO PCA, ACIP NA3TIVE SAPIENT IXO DTT JAGUAR HURT PCES ARMS ATO FTN DTN XG-Comms Connection-less Nets MNM TTO UCAR UCAR UCAR UCAR UCAR MDC2 Gov. Labs ONR, NWDC, Marines Army AATD, CECOM, ARL ONR, AFRL, AATD Classified NWDC, JFCOM, JL ACTD JTL ACTD Various (AFRL, NWDC, CECOM, etc.) R&D Emphasis Transition, System Emphasis Ongoing Bid LM BU SI Owego (UCAR), MS2 (HAIL) SI Owego (UCAR, HSKT, AMUST-D) SI Owego (UCAR) ADP Ft. Worth (AO FNC, Autonomy FNC ) MS2 (DD(X), DW, LCS Assured RT) SI Owego (UCAR)

Slide8: 

ATL R&D Interests Technology Topics

Slide9: 

Technology Topics

ONtology TRAnslation PROtocol (Ontrapro) – Ontology Mapping & Alignment: 

ONtology TRAnslation PROtocol (Ontrapro) – Ontology Mapping & Alignment XMDR Use Cases: Research challenges Semantic Normalization, Disambiguation, and Harmonization Multiple entries within the same taxonomy Similar but significant differences in the entries of two ontologies Mapping and Interrelationships Structure alignment of two different ontologies ATL’s ONTRAPRO prototype addresses these challenges Data Description (Schema or Ontology) Data Data Ontrapro

Example: Wine OntologiesTerm Dissimilarities(largely similar but not exactly same): 

Example: Wine Ontologies Term Dissimilarities (largely similar but not exactly same) Technology Topics

Slide12: 

Example: Wine Ontologies start with…Edit Distance Mapping (and other syntactical comparisons) Technology Topics

Example: Wine Ontologiesand then….multiple graphical structure mappings: 

Example: Wine Ontologies and then….multiple graphical structure mappings Technology Topics

Slide14: 

Example: Wine Ontologies and finish with multiple filters to resolve remaining discrepancies Technology Topics

Ontrapro – ongoing research activities: 

We are integrating learning techniques with the alignment heuristics to create situationally optimal ontology aligners. Ontrapro – ongoing research activities

Coordination Overhead Comparison: 

Coordination Overhead Comparison Domain Standard Ontrapro O(n2) O(n) O(1) (no prior) Manual Specialized Benefits Brittle-to-change No domain notion Manual Tradeoff for Generic Change-w/-Standard Single domain scope Automatic Specialized Benefits Change-at-Will Multi-domain ready Protocol XML OWL Effort, considerations, and time to wait -- the number of interfaces developed by some party other than the service creator to learn before one may connect each service to each other, by agents, SOAP, etc.-- are least for ONTRAPRO automatic alignment approach.

Slide17: 

Technology Topics

Software Agents – EMAA: 

Software Agents – EMAA ATL has an agent architecture used on dozens of DARPA and DoD Service Lab contracts: The Extensible Mobile Agent Architecture (EMAA) is a suite of mature software modules. XMDR Use Cases: Support for Development of a “Universal” Grid Unspecific (generic) domain Resources discovered, navigated, composed together via some intermediating semantic metamodel. Data Aggregation Information retrieval by linking together resources into aggregate data reports. Collating resources already registered within the XMDR registry. ONTRAPRO+Agents approach compliments the XMDR’s intermediating semantic metamodel. Our approach uses Agents as work-flow mechanism.

Software Agents – EMAA (cont.): 

Software Agents – EMAA (cont.) ATL’s EMAA agents: Comprise of a series of net-centric application resources described in OWL-S as web services Are composed of building blocks, “tasks” that interconnect multiple web-service resources Composition has been done dynamically through metadata planning, MPAC (Meta Planning for Agent Composition). EMAA agents usually extract information reports by aggregating outputs of some services and feeding these reports as inputs to others EMAA agents can also “enact” upon data by executing a pre-built processing instruction ATL builds “Semantic Web Agents”, described in terms found from a well-connected ontology Additional agent research at ATL also available for use: agent learning, adaptivity, and collaboration

Dynamic Agent Composition (example: charting Ship route in a channel): 

Dynamic Agent Composition (example: charting Ship route in a channel) Y-axis has available sensors as web-services X-axis has route-planners that use available sensor inputs

Dynamic Agent Composition(example: charting Ship route in a channel): 

Meta Planning of Agent Composition enables dynamic route planning Dynamic Agent Composition (example: charting Ship route in a channel)

Slide22: 

Technology Topics

Dynamic and Static Application Analysis: 

Dynamic and Static Application Analysis XMDR Use Case: Support a Data Grid Ontrapro reduces the need for standards for interoperability ATL has research interests in Dynamic and Static analysis applications to extract application’s inputs, outputs, and design intent API & Static Analysis: Examine software source-code documentation. Design documents in UML Whitepapers, conference papers, journal entries through NLP Dynamic Code Analysis At run-time analysis of object creation, method invocation

Slide24: 

Technology Topics

Service 2 Service Matchmaking: 

Service 2 Service Matchmaking XMDR Use Case: Discovery, Location and Retrieval Retrieve part or all of a terminology/concept structure Retrieval based on related items: “data element, property, concept, class, domain, context, classification scheme, ontology” Retrieve identity of registrar responsible for it Ontrapro creates a comparison of ontologies as a result of attempting to align them This allows us to find similar services semantically We have additional algorithms to match service descriptions

Slide26: 

Technology Topics

Explanation Generation: 

Explanation Generation XMDR Use Case: Help Support “A client application pulls metadata from the MDR in order to provide online help for an application end user.” Provided from the registered application directly OWL-S  Natural Language Descriptions of a semantic web service converted to a human-understandable paragraph on what it does Description of process What we intend to have happen What actually happened Automatic and dynamic: extraction of web-service spec/intent; comparison with actual result; generation of explanation; NLP output

Slide28: 

Technology Topics

Service Navigation and Execution Tools: 

Service Navigation and Execution Tools XMDR Use Case: Navigation Applications uses MDR to support navigation of registered data elements and concepts between data elements Execution Tools Users could navigate components found in an MDR registry Users could directly execute components if desired, from within the COACH framework Parameter study and optimization tools built in COACH

COgnitive Algorithm Composition Handler (COACH) concept: 

COgnitive Algorithm Composition Handler (COACH) concept XMDR Use Case: Navigation Applications use MDR to support navigation of registered data elements and concepts between data elements ATL is developing the COACH framework concept Users could navigate components found in an MDR registry Users could directly execute components if desired, from within the COACH framework Parameter study and optimization tools built in COACH

Information Interpretation and Integration Conference (I3CON): 

Information Interpretation and Integration Conference (I3CON) Experiment Participants Jerome Pierson (INRIA) John Li (Teknowledge) Lewis Hart (AT&T) Marc Ehrig (University of Karlsruhe) Todd Hughes (LM ATL) Guest Speakers Ben Ashpole (LM ATL) Bill Andersen (Ontology Works) Mike Pool (Information Extraction and Transport) Yun Peng (University of Maryland Baltimore County) Mike Gruningner (University of Maryland)

Information Interpretation and Integration Conference (I3CON): 

Information Interpretation and Integration Conference (I3CON) August 24-26, 2004 in Gaithersburg, MD ATL organized Published paper Positive Review in AFRL/IF Directorate Monthly Web Newsletter Experiment Participants Jerome Pierson (INRIA) John Li (Teknowledge) Lewis Hart (AT&T) Marc Ehrig (University of Karlsruhe) Todd Hughes (LM ATL) Guest Speakers Ben Ashpole (LM ATL) Bill Andersen (Ontology Works) Mike Pool (Information Extraction and Transport) Yun Peng (University of Maryland Baltimore County) Mike Gruningner (University of Maryland)

I3CON: Experiment Results (1): 

I3CON: Experiment Results (1)

I3CON: Experiment Results (2): 

I3CON: Experiment Results (2)

I3CON: Experiment Results (3): 

I3CON: Experiment Results (3)

Evaluation of Ontology Tools Workshop: 

Evaluation of Ontology Tools Workshop Participants: Customers: Dan Adams (NGA) Sam Chance (NRL) Kevin Keck (BNL) U.S: Mark Mayberry (Mitre) Chris Priest (HP) International: Willa Wei (MDA) Toru Ishida (KU) Shigoeki Hirai (AIST) Marc Ehrig (KU) Jerome Euzenat (INRIA) Alex Smirnov (RAS) Marco Neuman (DIT) Ian Horrocks (UM) Jeremy Carrol (HP) Contest results:

ATL and XMDR: Goals: 

ATL and XMDR: Goals ATL has prototypes and concepts that can help solve some of the key XMDR use cases (as shown in previous slides): ONTRAPRO, COACH, EMAA ATL can build Semantic Web Services out of each of these technologies and enable XMDR to use them as parts of its architecture and prototype. Support initial generation of ontology translators between web-services and service model Support initial generation of ontology content for the prototype Support easy migration of web-services from one version of service model to the next revision. ATL technology compatible to the XMDR framework: Usage of OWL, OWL-S, RDF and RDFS Usage of SWRL, UML, and RDQL. Most of our software written in Java. ATL is actively seeking solutions to the other XMDR use cases.

Soothing Images for Question Time: 

Soothing Images for Question Time

Slide39: 

Backup Vgs

Ontrapro accomplishments to Date: 

Ontrapro accomplishments to Date New, integrated alignment algorithms Syntax aligners Lexical aligners Structural Aligners Preprocessors Filters Enhanced display Experimentation and evaluation of alignment performance

Cognitive Algorithm Composition Handler (COACH): 

Cognitive Algorithm Composition Handler (COACH) COACH (Cognitive Algorithm Composition Helper) Stable architecture Composable experiment management Enhanced GUI Increased linkage to meta data Serialization support New runners and optimizers Fine grained control over search spaces Cluster extension Enables massive parameter studies, optimizations, and simulation evaluations Enables experience sharing between learners

Ontrapro Technology Status: 

Ontrapro Technology Status