applying semantic rules for dynamic service-oriented...
TRANSCRIPT
©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Applying Semantic Rulesfor Dynamic
Service-Oriented Architecture
Applying Semantic RulesApplying Semantic Rulesfor Dynamic for Dynamic
ServiceService --Oriented ArchitectureOriented Architecture
Suzette Stoutenburg, Leo Obrst, Suzette Stoutenburg, Leo Obrst,
Deborah NicholsDeborah Nichols, Ken Samuel, and Paul Franklin, Ken Samuel, and Paul Franklin
22 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Rules and Application Development
Research Team MembersResearch Team Members
Suzette StoutenburgE540/Colo Spgs
Leo Obrst, Ph.D.E544/McLean
Principal Investigators
Ontology and Logic Environment
Karen FoxE540/Colo Spgs
Deborah NicholsE544/McLean
Paul FranklinE540/Colo Spgs
Ken Samuel, Ph.D.E544/McLean
33 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
ObjectivesObjectives
Identify sponsorIdentify sponsor--specific requirements for semantic specific requirements for semantic rules and ensure that they are captured in the rules and ensure that they are captured in the emerging World Wide Web Consortium (emerging World Wide Web Consortium (W3CW3C) ) recommendation recommendation �� Rules separated from executable codeRules separated from executable code�� Rules integrated with ontologies to leverage logicRules integrated with ontologies to leverage logic
Demonstrate a Demonstrate a dynamicdynamic ServiceService--Oriented Oriented Architecture using rules that allow adaptive change Architecture using rules that allow adaptive change in standard operating proceduresin standard operating procedures�� Rules supporting situation awareness Rules supporting situation awareness �� Rules adjusting in response to realRules adjusting in response to real--time eventstime events
Optimization of the Logic Programming Optimization of the Logic Programming environment that integrates and reasons over environment that integrates and reasons over rules and ontologiesrules and ontologies
44 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Mission Use Case Key ConceptsMission Use Case Key Concepts
Convoy
Region of interest (ROI) provides“safety zone” around convoy
Unknown GMTI
Hostile GMTI
Intel Summary and ROI
Friendly GMTI
Path of UAVROI around unknown
GMTI
Convoy route
UAV
A track is said to be the “focal object” of an ROI
55 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Scenario Ontology OverviewScenario Ontology Overview
TheaterObject
hasFocalObject
DynamicRegionofInterestMilitaryUnit DynamicRegionofInterestMilitaryUnit
subclassOf
subclassOf
RegionOfInterest
subclassOf
Convoy ConvoyRouteConvoy ConvoyRoute
hasCurrentRoute
ConditionsAndAlerts
conditionAffects
describedBy
ObservationArtifact
GMTIObservation
IntelSummary
VMTIObservation
subclassOf
Primary Ontology
Major classwithin ontology
66 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Applications of RulesApplications of Rules
Support intelligent processing of realSupport intelligent processing of real--time eventstime events�� Synthesize information from multiple sourcesSynthesize information from multiple sources�� Apply data integrity constraintsApply data integrity constraints�� (Future work) Mediate between conflicting sources, and(Future work) Mediate between conflicting sources, and�� Derive new knowledge based on source type and reliabilityDerive new knowledge based on source type and reliability
Construct conceptualization of the battlespace for Construct conceptualization of the battlespace for enhanced situational awarenessenhanced situational awareness�� Transfer attribute values from ObservationArtifacts to Transfer attribute values from ObservationArtifacts to
TheaterObjectsTheaterObjects�� Establish Establish ‘‘safety zonessafety zones’’ and and ‘‘threat areasthreat areas’’ ((ROIsROIs) for objects) for objects�� Detect and report threats to friendly forcesDetect and report threats to friendly forces
Assist decision makers in applying Rules of Assist decision makers in applying Rules of Engagement to developing situationsEngagement to developing situations
77 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Year 2 Prototype ModificationsYear 2 Prototype Modifications
Modifications to the ontologiesModifications to the ontologies�� Added properties to support new rules and represent additional Added properties to support new rules and represent additional
information from messagesinformation from messages�� Added subclasses of weapons and vehicles Added subclasses of weapons and vehicles �� Classes and relations for other domains (missile, maritime, airsClasses and relations for other domains (missile, maritime, airspace)pace)
Rules made more robustRules made more robust�� Detect when new TheaterObjects should be createdDetect when new TheaterObjects should be created�� Some geospatial rules to detect enemy approaching, overtaking, eSome geospatial rules to detect enemy approaching, overtaking, etc. tc. �� Added Added ““extensionsextensions”” to SWRL, by creating predicates to:to SWRL, by creating predicates to:
Create new instancesCreate new instancesApply message data to the set of instancesApply message data to the set of instancesSupport nonSupport non--monotonic reasoningmonotonic reasoning
Dynamic rule changes supported in a Dynamic rule changes supported in a SOASOA with with communications across Enterprise Service Bus (ESB)communications across Enterprise Service Bus (ESB)
88 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Rules for Dynamic ServicesRules for Dynamic Services
Rule Set 1
Situation Awareness
Service
Rule Set 2
Goal: Dynamic service behavior based on realGoal: Dynamic service behavior based on real --time events time events
Create a generic service, e.g., Create a generic service, e.g., ““ What are the What are the implications of information updates for implications of information updates for
convoy security?convoy security? ””
Abstract from service behavior and Abstract from service behavior and express in the form of rulesexpress in the form of rules
If good visibility, send If good visibility, send out out UAVUAV to take closer to take closer
look at unknown look at unknown movermover
If poor visibility, If poor visibility, avoid encounters avoid encounters
with unknown moverswith unknown movers
Applicable Rule Set produces Applicable Rule Set produces appropriate response:appropriate response:
In normal daytime conditions, Rule Set 1 In normal daytime conditions, Rule Set 1 is in effectis in effect
RealReal--time weather event causes Rule time weather event causes Rule Set 2 to take effectSet 2 to take effect
99 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Sample Rules (Day)Sample Rules (Day)
an unknown mover is within the convoy’s ‘safety zone’alert the commander that there is a nearby unknown mover
IFTHEN
an unknown mover is within the convoy’s ‘safety zone’recommend that the convoy deploy UAV reconnaissance
IFTHEN
1010 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Sample Rules (Night)Sample Rules (Night)
an unknown mover is within the convoy’s ‘safety zone’alert the commander that there is a nearby unknown mover
IFTHEN
[above] & the convoy has a viable alternate route recommend that the convoy change to the route
IFTHEN
1111 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Logic Programming EnvironmentLogic Programming Environment
Developed Developed SWORIERSWORIER (Semantic Web Ontologies & Rules for (Semantic Web Ontologies & Rules for Interoperability with Efficient Reasoning) Interoperability with Efficient Reasoning) �� Reasons over ontologies and rules and answers queries with LogicReasons over ontologies and rules and answers queries with Logic
ProgrammingProgramming
Designed general Prolog predicates to implement the semantics Designed general Prolog predicates to implement the semantics of OWL (e.g., of OWL (e.g., hasPropertyWithhasPropertyWith, , disjointClassesdisjointClasses, , logicNotlogicNot))
Addressed major challengesAddressed major challenges�� Prolog and OWL define negation differentlyProlog and OWL define negation differently
�� Prolog assumes closed world, while OWL assumes open worldProlog assumes closed world, while OWL assumes open world
�� Disjoint and complementary classesDisjoint and complementary classesPrevious work claimed this was not possible [Previous work claimed this was not possible [VolzVolz et al., 2003]et al., 2003]
�� Simpler approach for equivalenceSimpler approach for equivalence
Initial performance of Initial performance of SWORIERSWORIER slow; improvements gained slow; improvements gained with with extensionalizationextensionalization, caching, and code minimization, caching, and code minimization
1212 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Logic Programming Logic Programming ChallengesChallenges
Logical NegationLogical NegationOpen/Closed WorldOpen/Closed WorldComplementary/Disjoint Complementary/Disjoint Classes*Classes*Disjunctive Disjunctive Conclusions*Conclusions*Enumerated Classes*Enumerated Classes*Equivalent Individuals*Equivalent Individuals*
* Solutions to be implemented* Solutions to be implemented
Error MessagesError Messages
Existential Existential Quantification*Quantification*
Cardinality Restrictions*Cardinality Restrictions*
Cyclic HierarchiesCyclic Hierarchies
Anonymous ClassesAnonymous Classes
Duplicate FactsDuplicate Facts
1313 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Knowledge Base DesignKnowledge Base Design
Translation(XSLT)
Extensionalization
User
Que
ries
An
swers
Changes
ExtensionalProgram(Prolog)
Que
ries
Answ
ers
ExtensionalProgram(Prolog)
Dynam
ic
GeneralRules
(Prolog)
IntensionalProgram(Prolog)
IntensionalProgram(Prolog)
Developer
CardinalityRules
(Prolog)
RuleML rulesSWRL
rulesOWLontologies
Translation(XSLT)
Extensionalization
User
Que
ries
An
swers
Changes
ExtensionalProgram(Prolog)
Que
ries
Answ
ers
ExtensionalProgram(Prolog)
Dynam
ic
GeneralRules
(Prolog)
IntensionalProgram(Prolog)
IntensionalProgram(Prolog)
Developer
CardinalityRules
(Prolog)
RuleML rulesSWRL
rulesOWLontologies
Translation(XSLT)
Translation(XSLT)
Extensionalization
UserUser
Que
ries
An
swers
Changes
ExtensionalProgram(Prolog)
ExtensionalProgram(Prolog)
Que
ries
Answ
ers
ExtensionalProgram(Prolog)
Dynam
ic
GeneralRules
(Prolog)
GeneralRules
(Prolog)
IntensionalProgram(Prolog)
IntensionalProgram(Prolog)
IntensionalProgram(Prolog)
IntensionalProgram(Prolog)
IntensionalProgram(Prolog)
DeveloperDeveloperDeveloper
CardinalityRules
(Prolog)
CardinalityRules
(Prolog)
RuleML rulesSWRL
rulesSWRLrulesOWLontologies
Rules not expressible in SWRL were represented in Pr olog directly; now in the process of specifying extensio ns on SWRL
1414 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Design DecisionsDesign Decisions
Enterprise Service Bus (ESB) technology selected Enterprise Service Bus (ESB) technology selected for underlying communication infrastructurefor underlying communication infrastructure�� Supports integration with other applications and prototypes Supports integration with other applications and prototypes
Google Earth used as clientGoogle Earth used as client�� Provides realistic maps and zoomProvides realistic maps and zoom--in capabilitiesin capabilities�� Supports integration with other applicationsSupports integration with other applications
Ground facts for instances stored in the Knowledge Ground facts for instances stored in the Knowledge Base (instead of database) for simplicityBase (instead of database) for simplicitySeparate simulation for Convoy used, making Separate simulation for Convoy used, making Convoy behavior independent of message injectionConvoy behavior independent of message injectionIssue: How should rules be swapped?Issue: How should rules be swapped?�� Since instances are stored in the Knowledge Base, how are they Since instances are stored in the Knowledge Base, how are they
persisted after the rule swap?persisted after the rule swap?�� Decision: Instantiate multiple Knowledge Bases, apply Decision: Instantiate multiple Knowledge Bases, apply
updated instances to eachupdated instances to each
1515 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Architecture Options for RulesArchitecture Options for Rules
KnowledgeBase 1
(Ontology, Rules, Instances)
KnowledgeBase 2
(Ontologies, Rules, Instances
Logic Server 1 Logic Server 2
KnowledgeBase1
(Ontology, Rules)
KnowledgeBase2
(Ontologies, Rules)
Logic Server 1 Logic Server 2
Database( Instances)
Option 1: Instances in Knowledge BaseInstances must be persisted across multiple KBs
Option 2: Instances in Database layerTools must support links from KB to database
Option 3: PreferredMeta-rules control application to KBinstances in database
KnowledgeBase 1
(Ontology, Rules)
KnowledgeBase2
(Ontologies, Rules)
Logic Server 1 Logic Server 2
MetaRules
Database( Instances)
1616 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Prototype ArchitecturePrototype Architecture
Events(GMTI, INTSUM)
SituationalAwareness
Service
Enterprise Service Bus
2. Message injection starts
AMZI LogicServer 1
AMZI LogicServer 2
1. Logic Servers instantiated with Rule Sets
3. Adapter performs schema-based transformation of messages
4. New input applied to Knowledge Bases – positions displayed on Client
Google EarthClient
Adapter
6. Sit’n Service detects events on ESB, queries for ale rts
7. Rules fire to produce “Single Integrated Picture ”
8. Alerts and Recommendations based on battlefield movement and Intel are displayed on client
MessageInjector
5. Rules fire to relate messages to Theater Object s
SimulatedPosition of
Friendly Convoy
OptimizedKnowledge
Base 1
OptimizedKnowledge
Base 2
1717 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Client DisplayClient Display
RECOMMENDATION: Take alternate routeALERT: Intelligence report of unknown force in vicinity
1818 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Summary of FindingsSummary of Findings
OWL is expressive and meets the majority of OWL is expressive and meets the majority of identified DoD requirementsidentified DoD requirementsSWRLSWRL has robust integration with ontologieshas robust integration with ontologiesSWRLSWRL does not fully support the DoD does not fully support the DoD requirements identified in this use caserequirements identified in this use caseTranslation of OWL to executable (Prolog) Translation of OWL to executable (Prolog) environment difficultenvironment difficultTranslation of rule languages fairly Translation of rule languages fairly straightforwardstraightforwardIntegrated framework of tools and capabilities Integrated framework of tools and capabilities needed to support semantic web needed to support semantic web developmentdevelopment
1919 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Future WorkFuture Work
Develop Develop metameta--rules to rules to select rule setselect rule setCompletion of Completion of SWORIERSWORIERStorage of Storage of instances in a instances in a databasedatabaseExtend to new Extend to new domainsdomains
2020 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Backup SlidesBackup Slides
2121 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Year 2 Scenario: Year 2 Scenario: Handling Dynamic EventsHandling Dynamic Events
Convoy moving throughenemy territory
Convoy nears sniperposition, changes route
Unknown GMTI reportedalong route
UAV retasked toinvestigate GMTI
UAV confirms enemyGMTI
Enemy GMTI in proximityConvoy alerted
Convoy moving throughenemy territory
Convoy nears sniperposition, changes route
Unknown GMTI reportedalong route
No need for UAV to investigate; route mustchange regardless, per
new rules of engagement
Convoy changesroute
Convoy CDR reportsunexpected sandstorm
New rules ofengagement applied
2222 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Two Dynamic Changes (Online)Apply the location and speed of a convoy
Apply the location and speed of an enemy motorized infantry unit
Desired Time:Baseline:
With Code Minimization:
a few seconds25.2 minutes10 milliseconds
Two Queries (Online)What are the locations and speeds of all known units?
What are all of the alerts and recommendations?
Desired Time:Baseline:
With extensionalization:& Dynamic Changes:& Code Minimization:
a few seconds33 minutes110 milliseconds58 minutes130 milliseconds
Extensionalization (Offline)
Desired Time:Baseline:
With Avoiding Reanalysis:& Code Minimization:
a few hoursCRASH13 hours6.5 hours
SWORIERSWORIER: Efficient Reasoning, : Efficient Reasoning, Dynamic UpdatesDynamic Updates
2323 ©© 2006 The MITRE Corporation. All rights reserved2006 The MITRE Corporation. All rights reserved10 November 200610 November 2006
Project ProductsProject Products
Stoutenburg, Suzette; Leo Obrst; Deborah Nichols; Ken Samuel; KeStoutenburg, Suzette; Leo Obrst; Deborah Nichols; Ken Samuel; Ken Laskey; Adrian Johnson; Jason Peterson; n Laskey; Adrian Johnson; Jason Peterson; Karen Fox. 2006. Karen Fox. 2006. Dynamic Service Oriented Architectures through Semantic TechnoloDynamic Service Oriented Architectures through Semantic Technologygy. Submitted to 4th . Submitted to 4th International Conference on Service Oriented Computing (International Conference on Service Oriented Computing (ICSOCICSOC 2006).2006).Samuel, Ken; Leo Obrst; Suzette Stoutenburg; Deborah Nichols; AdSamuel, Ken; Leo Obrst; Suzette Stoutenburg; Deborah Nichols; Adrian Johnson; Ken Laskey; Jason Peterson; rian Johnson; Ken Laskey; Jason Peterson; Karen Fox. 2006. Karen Fox. 2006. Transforming OWL and Semantic Web Rules to Prolog: Towards DescrTransforming OWL and Semantic Web Rules to Prolog: Towards Description Logic Programsiption Logic Programs. . Federated Logic Conference, August 2006, Seattle, Washington.Federated Logic Conference, August 2006, Seattle, Washington.Stoutenburg, Suzette; Leo Obrst; Deborah Nichols; Ken Samuel; KeStoutenburg, Suzette; Leo Obrst; Deborah Nichols; Ken Samuel; Ken Laskey; Adrian Johnson; Jason Peterson; n Laskey; Adrian Johnson; Jason Peterson; Karen Fox. 2006. Karen Fox. 2006. Toward a Standard Rule Language for Semantic Enterprise IntegratToward a Standard Rule Language for Semantic Enterprise Integration; Year 1 Results: ion; Year 1 Results: Evaluating Proposed Approaches for a Standard Semantic Web Rule Evaluating Proposed Approaches for a Standard Semantic Web Rule LanguageLanguage. MITRE Technical Report, March . MITRE Technical Report, March 2006.2006.Stoutenburg, Suzette; Deborah Nichols. 2005. Stoutenburg, Suzette; Deborah Nichols. 2005. Toward a Standard Rule Language for Semantic Enterprise Toward a Standard Rule Language for Semantic Enterprise IntegrationIntegration. 2005 International Conference on Rules and Rule Markup Langua. 2005 International Conference on Rules and Rule Markup Languages for the Semantic Web, poster, ges for the Semantic Web, poster, RuleML 2005, November 2005, Galway, Ireland.RuleML 2005, November 2005, Galway, Ireland.Stoutenburg, Suzette. Stoutenburg, Suzette. Logics for the Semantic Web (for the nonLogics for the Semantic Web (for the non--logician).logician). MITRE Working Note, January 2006.MITRE Working Note, January 2006.Obrst, Leo; Todd Hughes; Steve Ray. 2006. Obrst, Leo; Todd Hughes; Steve Ray. 2006. Prospects and Possibilities for Ontology Evaluation: The View frProspects and Possibilities for Ontology Evaluation: The View from om NCORNCOR.. Workshop on Evaluation of Ontologies for the Web (Workshop on Evaluation of Ontologies for the Web (EON2006EON2006), Edinburgh, UK, May 22, 2006.), Edinburgh, UK, May 22, 2006.Obrst, Leo; Werner Obrst, Leo; Werner CeustersCeusters; Inderjeet Mani; Steve Ray; Barry Smith. 2006. ; Inderjeet Mani; Steve Ray; Barry Smith. 2006. The Evaluation of Ontologies: The Evaluation of Ontologies: Toward Improved Semantic Interoperability.Toward Improved Semantic Interoperability. Chapter in: Chapter in: Semantic Web: Revolutionizing Knowledge Discovery in Semantic Web: Revolutionizing Knowledge Discovery in the Life Sciences,the Life Sciences, Christopher J. O. Baker and KeiChristopher J. O. Baker and Kei--Hoi Cheung, Eds., Springer, forthcoming 2006.Hoi Cheung, Eds., Springer, forthcoming 2006.Obrst, Leo; Patrick Cassidy; Steve Ray; Barry Smith; Dagobert SoObrst, Leo; Patrick Cassidy; Steve Ray; Barry Smith; Dagobert Soergel; Matthew West; Peter Yim. 2006. The ergel; Matthew West; Peter Yim. 2006. The 2006 Upper Ontology Summit Joint Communiqu2006 Upper Ontology Summit Joint Communiquéé. Journal of Applied Formal Ontology. Forthcoming.. Journal of Applied Formal Ontology. Forthcoming.PoliPoli, Roberto; Leo Obrst. 2006. , Roberto; Leo Obrst. 2006. The Interplay Between Ontology as Categorical Analysis and OntolThe Interplay Between Ontology as Categorical Analysis and Ontology as ogy as TechnologyTechnology.. Chapter 9 in Part One: Ontology as Technology in the book: TAO Chapter 9 in Part One: Ontology as Technology in the book: TAO –– Theory and Applications of Theory and Applications of Ontology, Volume 2: The InformationOntology, Volume 2: The Information--science Stance, Michael Healy, Achilles science Stance, Michael Healy, Achilles KameasKameas, Roberto , Roberto PoliPoli, eds. , eds. Forthcoming.Forthcoming.Obrst, Leo. 2006. Obrst, Leo. 2006. Ontological ArchitecturesOntological Architectures. Chapter 2 in Part One: Ontology as Technology in the book: TAO. Chapter 2 in Part One: Ontology as Technology in the book: TAO ––Theory and Applications of Ontology, Volume 2: The InformationTheory and Applications of Ontology, Volume 2: The Information--science Stance, Michael Healy, Achilles science Stance, Michael Healy, Achilles KameasKameas, Roberto , Roberto PoliPoli, eds. Forthcoming., eds. Forthcoming.