Role of NIST in Realizing OKN · distribution-function Enhance Branch Generic-control Combine Identify Connect Increase Couple Indicate Distribute Inhibit Divide-flow Limit. .

RoleofNISTinRealizingOKN

RamD.Sriram,Ph.D.Chief,SoftwareandSystemsDivisionInformationTechnologyLaboratory

[email protected] 1

Outline

• NISTBackground• OKNRelatedProjects• Expectations

2

NIST - Bird’s eye view

CourtesyHDRArchitecture,Inc./SteveHall©Hedrich Blessing

TheNationalInstituteofStandardsandTechnology(NIST)iswhereNobelPrize-winningsciencemeetsreal-worldengineering.

Withanextremelybroadresearchportfolio,world-classfacilities,nationalnetworks,andaninternationalreach,NISTworkstosupportindustryinnovation– ourcentralmission.

NIST’sMission

•TopromoteU.S.

innovationandindustrial

competitivenessby

advancingmeasurement

science,standards,

andtechnologyinwaysthat

enhanceeconomicsecurity

andimproveourqualityoflife.

U.S.EconomyDepends onNISTMeasurements

• Majorassets– ~3,000employees– ~2,800associatesandfacilitiesusers– ~1,300fieldstaffinpartnerorganizations– Twomainlocations:Gaithersburg,Md.,andBoulder,Colo.

– Fourexternalcollaborativeinstitutes:basicphysics,biotech,quantum,andmarinescience

– NobelPrizeWinners:1997,2001,2005,2007

NIST:BasicStatsandFacts

FY2016Appropriations$971.3M

©R.Rathe

MEP($155M)

Construction($119M)

NISTLabs($690M)

Others($7.3M)

RamD.Sriram

NationalInstituteofStandardsandTechnology(NIST)

RESEARCHCOMPONENTS

InteractingwithNIST

• GuestResearchers/FacultyAssociates• GrantsandContracts• IPA• SummerStudents• NRCPostDoctoralProgram• CollaborativeProposals

Outline


9

OKNRelatedProjects

• CoreProductModel&Beyond(EL&ITL)• Ontologies&CategoryTheory(ITL&EL)• ExtractingDomainSpecificTermsFromDocuments(ITL&MML)

• TREC(ITL)

10

GenericInformationModelingRequirements

• Modelconstruction• Representationacrossscales• Broadaccommodationformultipleformalisms• Separationofdomain-specificconcerns• Integrationandaggregationacrossmodels• Modelevolution• Flexibilityandmodularity• Scalability

11

CoreProductModel(1994-2010)• Objective:base-levelproductmodelthatis:

– generic– extensible– independentofanyoneproductdevelopment

process– capableofcapturingfullengineeringcontext

• Keyfeature:explicitrepresentationofFunction– Form- Behavior

(incontrasttoSTEPAP203thatessentiallyrepresentsonlyform)

KnowledgeRepresentation

Form Function

Behavior

Relationships Information

Geometry

Material

A product is represented by a hierarchy of entities of the class Artifact, which is an aggregation of Function, Form and Behavior. Function represents what the artifact is supposed to do; Form represents the proposed design solution for the design problem specified by the Function; and Behavior represents the evaluation of how the artifact implements its function.

FormRuptureDisc

DigitalPicture DetailedDesign3DModel

Function

SimplifiedFunctionFlow

DetailedFunctionFlow

FunctionLinks

LinksbetweenFunctionandArtifact

DesignObjectsFunctiontypes FunctionTaxonomy

Function Conveyance-function Usage-function Advance Sink Channel Absorb Conduct Consume Convey Destroy Direct Dissipate Divert Eliminate Guide Empty Generic-move Export Rotate Remove Transfer Source Translate Add Transmit Create Transport Emit Signal/Control- Extract function Generate Actuate Import Adjust Supply Close Storage Decrease Accumulate Delay Collect Detect Store Display Combination/ Equalize distribution-function Enhance Branch Generic-control Combine Identify Connect Increase Couple Indicate Distribute Inhibit Divide-flow Limit . . . . . .

CPM:FourCategoriesofClasses1. Classesthatprovidesupportinginformationfor

theobjects(abstractclasses)forstoringcommoninformation– CoreProductModel,CommonCoreObject,

CommonCoreRelationship– CoreEntity,CoreProperty

2. Classesofphysicalorconceptualobjects– Artifact,Feature,Port,Specification,Requirement– Function,TransferFunction,Flow,Behavior– From,Geometry,Material

3. Classesthatdescriberelationshipsamongobjects,theyarederivedfromCommonCoreRelationship– Constraint,Usage,Trace,EntityAssociation

4. Classesthatarecommonlyusedbyotherclasses.– Information,ProcessInformation,Rational

CPM:ThreeKindsofAssociationsAllobjectclasseshavetheirownseparate,independentdecompositionhierarchiesbyattributessuchassubArtifacts/subArtifactOffortheArtifactclass.

Thereareassociationsbetween:– aSpecificationandtheArtifactthatresultsfromit– aFlowanditssourceanddestinationArtifactsanditsinputand

outputFunctions– anArtifactanditsFeatures.

FouraggregationsarefundamentaltotheCPM:– Function,FormandBehavioraggregateintoArtifact– FunctionandFormaggregateintoFeature– GeometryandMaterialaggregateintoForm– RequirementaggregatesintoSpecification.

OpenAssemblyModel• Objectives:

– standardrepresentationforassemblyandsystem-leveltoleranceinformation

– extensible• Keyfeatures:

– emphasizesinformationrequirementsforpartfeaturesandassemblyrelationships

– assemblyasaconceptandassemblyasadatastructure

CPM&OPMleads:Fenves &Sudarsan

ExtensiontoCPM:OpenAssemblyModel

ProductRepresentation:Summary

PropertySpecificationBehavior

Form Geometry Material

Object

Function

Strength view Kinematics viewShape view

Idealization

Mapping

Master Model

Functional Model

DirectUndirect

SetRel. ConstraintAssemblyRel Reference

Relationship

Requirement

Family Designation

PFEM_Artifact

ConfigurationProduct series

Component series

Evaluation

Design Evaluation

Family Derivation

Series Derivation

Version Derivation

Family

Product Family

Component Family

Series

Product Version

Component Version

Version

Entity

FeatureArtifact

Part

Artifact Association

Ass’y Association

Composite Feature

Ass’y Feature

Assembly

Ass’y feature association

Ass’y feature association rep.

Position orientation

Relative motion

Connection

Intermittent

Movable

Fix

OAMFeature

Tolerance

Kinematic Pair Kinematic Path Parametric Ass’y Constraint

Evolution Rationale

Rationale

Design Rationale

Design Justification Evolution.

Family Derivation Rat.

Develop. Spec. Evolution

Develop. Spec.

Regulation

Technology

Design Justification

Optimal Suf.

PrincipleCase Rule Catalog

Authority Trade Off

Design Evolution Rat.

1

2

3

4

5

6

ImpactsofNIST’s ProductModelingWork• MajorportionsofCPM/OAMwasincludedinSTEPPart109• CPM/OAMisbeingusedbyseveralorganizations• PortionsofCPM/OAMwasusedinCADvendorsoftware(EU)• OntoSTEP releasedasopensourcesoftware• ExtensionstoCPM/OAM

– GeorgiaInst.ofTech:Design/Analysisintegration§ Createscomplexanalysismodelsfaster§ BeingintegratedintoworkwithLockheedMartin

– UniversityofWisconsinMadison:Heterogeneousmaterialmodeling§ Supportsanewapproachtomesh-freeanalysis

– ArizonaStateUniversity:Extendedfeatures§ BeingusedinaprojectforUSArmy

– SyracuseUniversity:Tolerance,kinematics,andassemblyanalysis§ Providesgreatersupportforintegratingproductspecificationswithmanufacturingand

inspection.– UniversityofMaryland,CollegePark:Assembly-basedsearch

§ Allowsorganizationstouseexistingdesigns§ Currentlyinnegotiationswithanindustrialpartner

TheNextSteps

• EngineeringLaboratoryiscurrentlyinvolvedintheIndustryOntologiesFoundry

24

OKNRelatedProjects


• TREC(ITL)

25

26

OntologySpectrum

weak semantics

strong semantics

IsDisjointSubclassofwithtransitivityproperty

ModalLogic

LogicalTheory

Thesaurus HasNarrowerMeaningThan

Taxonomy IsSub-Classificationof

Conceptual Model IsSubclassof

DBSchemas,XMLSchema

UML

FirstOrderLogic

XML

ER

ExtendedER

DescriptionLogicDAML+OIL,OWL

RDF/SXTM

SyntacticInteroperability

StructuralInteroperability

SemanticInteroperability

Courtesy:LeoObrst,MITRE

problem normal normal

Image segmentation

Feature Vectors Computation

Mapping calculated featurevectors into disease ontology

Highlighting regions withdeviation from normal conditions

Input Image

Output:Suggestion of the potential diagnosis Sugessted diagnosis: Ulcer...

Methodology for Image to Diagnosis Through Disease Ontology

27

Fragment of Disease Ontology in Protege’-Ontoviz

28

UMLRepresentationofInflammatoryBowelDisease

+endoscopic feature1 : <unspecified> = ulcers+location : string(idl) =distal small interstine / proximal colon/ distal colon/ rectum

Idiopathic inflammatory bowel disease

-endoscopic feature 2 : <unspecified> = strictures-endoscopic 3 : <unspecified> = cobblestonning

Crohns disease-endoscopic feature 2 : <unspecified> = pseudopolyps

Ulcerative colitis

29

FormoreaboutOntologies:SeeOntologySummitWebPages

• TheOntologySummitisanannualseriesofeventsthatstartedin2006withthejointsponsorshipofOntolog andNIST

• Thesummitislargelyaself-organizing,bottom-up,volunteerdriveneffort,thatsolicitscontributionsfromparticipantsaroundtheworldinbothindustryandacademia

• Eachyear'sSummit(differentthemeeveryyear)consistsofaserieseventsandcontinueddiscoursespanningthreemonths,culminatinginafree,two-dayface-to-faceworkshopandsymposium

• URL:http://ontologforum.org/index.php/OntologySummit

30

• Mostsemanticmodels(e.g.,OWL)arebasedonformallogicandsettheory.Thisisusuallysufficientforenablinginformationinteroperabilitybetweensystems.

• Logicalmodelsareusuallylessrelevantinsystemanalysisandbehavior.

• Otherrelevantclassesofsemanticsincludestochastic,dynamicalsystems,geometricandcomputationalmodels.

• Weareexploringtheuseofcategorytheory(CT)asaflexiblefoundationforrepresenting,analyzingandinterlinkingthisbroaderclassofsemanticmodels.

TowardCategoryTheory

Inmechanicalengineering,Newton’scalculusprovides:

● Alanguageforrepresentingmechanicalstates,processesandbehavior.

● Acollectionofstandardtechniquesandalgorithmsforanalyzingmechanicalsystemsexpressedinthislanguage.

Thepredicatecalculusplayssimilarrolesforlogic,andtheλ-calculusforcomputation.

CTisacalculusofabstractprocesseswhichgeneralizesallthreeofthesecontexts.

FormalrepresentationinCTprovides“hygiene”,helpingtoguideourthinkingandtoavoiderrorsandmisrepresentation.

ACalculus ofInformationInmechanicalengineering,Newton’scalculusprovides:

• Alanguageforrepresentingmechanicalstates,processesandbehavior.

• Acollectionofstandardtechniquesandalgorithmsforanalyzingmechanicalsystemsexpressedinthislanguage.

Thepredicatecalculusplayssimilarrolesforlogic,andtheλ-calculusforcomputation.

CTisacalculusofabstractprocesseswhichgeneralizesallthreeofthesecontexts.

FormalrepresentationinCTprovides“hygiene”,helpingtoguideourthinkingandtoavoiderrorsandmisrepresentation.

33

OntologySpectrum

weak semantics

strong semantics

IsDisjointSubclassofwithtransitivityproperty

ModalLogic

LogicalTheory

Thesaurus HasNarrowerMeaningThan

Taxonomy IsSub-Classificationof

Conceptual Model IsSubclassof

DBSchemas,XMLSchema

UML

FirstOrderLogic

RelationalModel,XML

ER

ExtendedER

DescriptionLogicDAML+OIL,OWL

RDF/SXTM

SyntacticInteroperability

StructuralInteroperability

SemanticInteroperability

Courtesy:LeoObrst,MITRE

IntegratingSupplyChainDatabases:CTApproach

Wisnesky R,Breiner S,JonesA,Spivak DI,Subrahmanian E.UsingCategoryTheorytoFacilitateMultipleManufacturingServiceDatabaseIntegration.ASME.J.Comput.Inf.Sci.Eng.2017;17(2):021011-021011-11.doi:10.1115/1.4034268.

OKNRelatedProjects


• TREC(ITL)

35

DisciplinaryInformationSystems

SearchwindowDocumentWindow.…….…..

.…….…..

.…….…..

.…….…...…….…...…….…...…….…...…….…...…….…...…….…..

OCR

.…….…...…….…...…….…...…….…...…….…...…….…...…….…..

ProcessedText

Parsers

Root&RulebasedCo-wordAnalysis

Word2Vec

Compoundnouns

Linguisticmapandtaxonomy

Tools

SearchTermMap

RootandRulebasedmethodallowsfor- Autocompletebasedusageinthe

domainforsearch- Facilitatescreationoftaxonomy- Couldbeconnectedtoontologytools- ExperimentswithIUCR,APS,others- Itenhancesperformanceofothertools

suchasWord2vecfortopicmodeling

OKNRelatedProjects


• TREC(ITL)

37

TREC

• Aworkshopseriesthatprovidestheinfrastructureforlarge-scaleevaluationof(text)retrievaltechnology– realistictestcollections– uniform,appropriatescoringprocedures– aforumfortheexchangeofresearchideasandforthediscussionofresearchmethodology

TREC is an annual benchmarking exercise that has become a de facto standard in Information Retrieval evaluation.

Stephen RobertsonMicrosoft

SIGIR 2007

Establishes the research methodology

TREC has proven to be a valuable forum in which IBM Research has contributed to an improved understanding of search, while at the same time the insights obtained by participating in TREC have helped to improve IBM’s products and services.

Alan Marwick, et al.IBM chapter of the TREC book

2005

Facilitates technology

transfer

In other words, for every $1 NIST and its partners invested in TREC, at least $3.35 to $5.07 in benefits accrued to IR researchers…These responses suggest that the benefits of TREC to both private and academic organizations go well beyond those quantified by this study’s economic benefits.

RTI InternationalEconomic Impact Assessment

of NIST’s TREC ProgramDecember 2010

Amortizes the costs of

infrastructure

This project [the TREC Legal track] can be expected to identify both cost effective and reliable search and information retrieval methodologies and best practice recommendations, which, if adhered to, certainly would support an argument that the party employing them performed a reasonable ESI search, whether for privilege review or other purposes.

Magistrate Judge Paul Grimm Victor Stanley v. Creative Pipe

Forms/solidifies a research community

The TREC data revitalized research on information retrieval. Having a standard, widely available, and carefully constructed set of data laid the groundwork for further innovation in the field. The yearly TREC conference fostered collaboration, innovation, and a measured dose of competition (and bragging rights) that led to better information retrieval.

Hal VarianGoogle Chief Economist

March 4, 2008

Improves the state of the

art

Outline


40

Expectations

• Provideagenerictemplateforstructuringknowledge

• Developevaluationandtestmethodologies• ExploreTRECtrackforevaluatingOKNinvariousdomains

• Leveragefunding

41