ontology

OntologyModelingUsingUML

DepartmentofComputerScience,UniversityofRegina,

Regina,Saskatchewan,CanadaS4S0A2

wangx@,chan@

Abstract

Ontologyisacomprehensiveknowledgemodelwhichenablesthe

developertopracticea“higher”levelofreuofknowledge.

Typically,differentmodelinglanguagesareemployedindifferent

ordertoachieveknowledgeinsteadofsoftwarereu,wepropo

forgingaclormappingbetweentheknowledgeandsoftware

paper,wefirstprent

anontologydevelopedandtheninvestigateUMLasanontology

modelingtooltofacilitatethemappingfromknowledgemodelto

stratetheintegratedmodelingapproach

usingUML,weappliedittodevelopanontologyforthedomainof

lectingaremediationtechniqueforpetroleumcontaminatedsites.

Keywords

Ontology,KnowledgeReu,UnifiedModelingLanguage(UML)

uction

Theobjectiveofourstudyistoinvestigatestrongercouplingbetweenthe

knowledgeengineeringandsoftwareengineeringphasofaknowledge-bad

othphamphasizeamodel-badapproach,themodelsare

oftennotdirectlytranslatablebetweenthetwophastoenableaamless

perprentsourinvestigationof

UMLasanontologymodelingtooltofacilitatethemappingfromknowledgeto

hemodel-badapproachinthetwophasofknowledge

engineeringandsoftwareengineeringarediscusd.

Thenotionof‘knowledgelevel’[1]hasprovidedanimportantperspectiveto

ArtificialIntelligence(AI)

emphasisonknowledgelevelhangenderedtheso-calledmodelingapproaches

forknowledgesystems,inwhichdevelopingaknowledge-badsystemisviewed

astheconstructionofariesofmodelsrelatedtosomeproblem-solvingbehavior.

Withinthiscontext,ontology,asakindofknowledgemodel,hasbecome

importantasavehicletoenhanceknowledgesharingandreu,andobject-

orienteddatabadesign.

Insoftwareengineering,

enabledeveloperstovisualizeasystemandtospecifythestructureorbehaviorofa

r,eventhoughbothsoftwareengineersandknowledgeengineers

umodel-badmethodologies,theyoftenemploydifferentmodelinglanguages

,fied

ModelingLanguage(UML)isastandardlanguageforwritingsoftwareblueprints,

whichmaybeudtovisualize,specify,constructanddocumentanobject-

gethegapbetweenontologicalmodelsandsoftware

engineeringmodels,weinvestigatereprentingadomainontologyusingUML.

Toillustratemodeldevelopment,weapplyourmodelingapproachtothesample

domainoflectionofpetroleumremediationtechnologyforcontaminatedsites.

Thepaperisorganizedasfollows:Section2describestheproblemdomainof

petroleumremediationlectionandprentstheknowledgemodelsthatwere

developedusingahierarchicaltreeofclassandentity-relationship(ER)

n3brieflyexplainsourrationaleforintegratingtheknowledge

ion4,theapproachofadoptingUMLforthe

nclusionremarksareinSection5.

gyDesigninthePetroleumRemediation

Domain

2.1OntologyModeling

Twomajorcomponentsofaknowledge-badsystemaredomainknowledgeand

ntologiesinfluenceproblem-solving

knowledge,theymainlyplayaroleinanalysing,modelingandreprenting

pmentofaknowledgebatypicallyassumes

logy,ontheother

hand,isanexplicitspecificationofaconceptualization,whichcanrveasa

giesareoftenequatedwith

taxonomichierarchiesofclass[2],withclassdefinitions,andthesubsumption

nbeudasthebasisofknowledgeacquisitiontoolsforgathering

logy

m

discussingtheprocessofontologyconstruction,wefirstdescribetheapplication

problemdomain.

2.2DomainofSelectionofPetroleumRemediationTechnologiesfor

ContaminatedSites

Pollutionfromthepetroleumindustryiscurrentlyamajorenvironmentalconcern

uatelydealwitheachpollutionsituation,anappropriate

ironmentalengineermustmake

adecisionwhethertocontrolorreducethecontaminantinthesoiland

r,contaminatedsiteshavedifferentcharacteristicsdepending

onthepollutant’sproperties,hydrologicalconditions,andavarietyofphysical

characteristicssuchasmasstransferbetweendifferentphas,chemical,and

ore,remediationtechnologiesapplicablefordifferent

lectionprocessisdifficultandpos

animportantchallengeforenvironmentalengineerswhoneedsupporttoolsinthis

,implementingashareableknowledgebainthe

domainofremediationlectionprocessisasteptowardsfulfillingthisneed.

2.3OntologyDesignfortheDomainofPetroleumRemediationSelection

tegorizationscan

soncategorizationthatpertaintoontology

designintheAIfieldincludeSowa’sontology[3],Dahlgren’sontology[4],and

Gensim[5].Sincethedomainontologyofaknowledge-badsystemisanexplicit

specificationoftheobjects,concepts,andotherentitiesthatarepresumedtoexist

insomeareaofinterestaswellastherelationshipsthatareheldamongthem[2],it

definesthetoftermsandrelationsofadomainindependentofanyproblem-

ly,suchmethod-specificformulationofdomain

ore,toparatethe

potentiallyreusabledomainknowledgefromthemethod-specificknowledgeisa

considerationthatguidedourstructureofthedomainontology.

Constructionofanontologyforaparticulardomainrequiresaprofound

analysis,whichrevealstherelevantconcepts,attributes,relations,constraints,

softhisprocesshavebeendiscusdin

[6]ignoftheontologystructure

fo

ontologystructureisconstructedwithanumberofassumptionswhichare

discusdasfollows.

threesub-categoriesareclass,process,andrelation.A“Class”canbeaTangible

retwomajorcategoriesunderTangibleThing:

lly,theclassontology

includesalltangibleorabstractconceptsorsubstancesthatarerelevantinthe

petroleumremediationprocess,suchaschemicals,sitemedia,standards,and

experiments.

A“Process”canbeasimpleprocess,complexprocess,orcombinationprocess.

Ifataskcanbeaccomplishedintwostepsusingobjectswithinasingleclass

hierarchysuchasmixandadd,

taskisaccomplishedinmorethantwostepsusingobjectswithinasingleclass

hierarchy,binationprocessapplies

r

words,ataskisaccomplishedinmorethantwostepsusingobjectsfromdifferent

classhierarchies.

A“Relation”coverspropertiesofclassincludingtheirinternalstructureand

ioncanbeoneofthreetypes:binaryrelation,

multiplerelation,yrelationisarelationbetweentwo

class;multiplerelationisarelationinvolvingmorethantwoclass;instance

relati

instancerelationisonlytrueforaspecificclassorinstance.

Things

Tangible_thing

Abstraction

ProcessRelation

Decomposable_object

Non_decomposable

_object

Site_media

Chemical

Standard

Experiment

Remediation

Soil

WaterGas

Groundwater

Chemical

_contaminant

Organic_chemical

_contaminant

Non_organic_chemical

_contaminant

Petroleum_

contamination

_standard

Air_pollution

_standard

Water_pollution

_standard

USEPA_

standard

Saskatchewan

_standard

Alberta_

standard

Sampling_

experiment

Soil_sampling_

_experiment

Water_sampling_

_experiment

Air_sampling_

_experiment

Petroleum_waste

_remediation

Water_pollution

_remediation

Air_pollution

_remediation

In_situ_

remediation

Ex_situ_

remediation

Place

Time

Efficiency

Simple_

process

Complex_

process

Combination_

process

Class

Binary_

relation

Multiple_

relation

Instance_

relation

gyDesign:ClassificationHierarchyofthePetroleumRemediation

Domain(Takenfrom[6])

2.4EntityRelationshipModeling

Anotheraspectoftheontologyinvolvesrelationshipsamongtheclassinthe

-Relation(ER)diagramshavebeenudto

provideaconvenientframeworkfordatabadesign,developmentand

prentrelationshipsamongentitiesinvolvedinan

informationsystemandtheyareudheretoreprentrelationshipsamongclass

intheontology.

SomeoftheclassdepictedintheclassificationhierarchyshowninFigure1

aredescribedingreaterdetailinFigure2whichshowsdetailedinformationon

mplerelationsinthedomainaredescribedin

[6].

ationofKnowledgeandSoftwareModels

Theknowledgeinthepetroleumremediationdomainhasbeenanalydand

reprentedusingthetoolsofclassificationhierarchy,classhierarchyandentity-

licitreprentationcanbeconvertedintoan

implementedontologyusingontologyeditorssuchasProtégé[7].Theconstruction

Site_media

Soil

Groundwater

Soil_groundwater

Slot:

SuperClass:

Class:

isalink

specieslink

Soil_type

Soil_hydraulic_permeability

Soil_heterogeneity

Soil_isotropy

Horizontal_hydraulic_conductivity

Vertical_hydraulic_conductivity

Groundwater_type

Site_size

Site_hydraulic_conductivity

Site_volume

Site_area

Depth_of_site

class/object/entity

attribute/slot/species

Water

Gas

Soil_type

Soil_hydraulic_permeability

Soil_heterogeneity

Soil_isotropy

Horizontal_hydraulic_conductivity

Vertical_hydraulic_conductivity

pH_value

Groundwater_type

pH_value

eClassHierarchy(Takenfrom[6])

oftheimplementedontologymodelhasbeendescribedin[6].However,adomain

rotégécan

constructanontologymodel,theimplementedreprentationoftheontologyis

,abridgeisneededto

crossthegapinmodelingbetweentheknowledgemodelandaknowledgebad

system.

Ideally,thesoftwaredevelopmentprocessshouldconsistoftwocompletely

paratestages:Ur-centeredStage,whichisrelatedtotheursandtheirneeds,

andSystem-centeredStage,whichistofindthecomputersolutionthatsatisfies

isperspective,theknowledgemodelismainlyrelatedtothe

firststagewhilethesoftwaremodelstridesthetwostagesbutemphasizesthe

dgemodelingcannotproducegoodsoftwarebecauofitslackof

,abridgeisneededtocrossthegapinmodeling

betweentheknowledgemodelandaknowledgebadsystem.

Inthispaper,wepropodthattheUML(trademarkoftheObjectManagement

Group,Inc.)togetherwithObjectConstraintLanguage(OCL)canbeudasthe

reprentationlanguagetobridgethisgap.

ationoftheIntegratedModelingApproach

4.1UMLasanontologymodelinglanguage

Theontologyprentedinction2canbereprentedusingUnifiedModeling

Language(UML).UMLisastandardlanguageforwritingsoftwareblueprints.

UMLcanbeudtovisualize,specify,construct,anddocumenttheartifactsofa

propriateformodelingsystemsrangingfrom

enterpriinformationsystemstodistributedWeb-badapplicationsandrealtime

nexpressivelanguage,capableofreprenting

r,despite

itxpressiveness,aUMLgraphicalmodel,suchasaclassmodel,cannotbyitlf

y,additionalconstraintsneed

aintsspecifiedinnatural

,formalandoftenmathematically

derivedlanguagesareneeded,whichareoftendifficultfortheaveragebusinessor

ftheUnifiedModelingLanguagecanuOCLtospecify

formal

tionaladvantageofemployingthe

UMLwithOCLformalismisthateachsymbolintheUMLnotationiswell-

modelreprentedinUMLwith

,

somesampleUMLreprentationsoftheontologyinthepetroleumremediation

domainarediscusd.

4.2OntologyofthePetroleumRemediationDomain:aUML

Reprentation

Theontologyofthepetroleum

remediationdomainmodelcanbe

mappedintotwomodeling

componentsasshowninFigure3:

theClassificationandRelationship

ssificationmodule

includestheclassofconceptsin

thepetroleumremediationdomain

whiletheRelationshipmodule

describesrelationshipsamongtho

modulescanbe

regardedasthehighlevel

encapsulationoftheanalysisand

designmodelforthesystemtobe

ingonthetarget

system,modificationandrefinements

portionsareexplainedinthe

followingsubctions.

Ontologies

4.2.1Classification

Figure4showsthetop-levelclassificationdiagramofontologydesigninthe

ndardnotationsinUMLareudto

reprenttheclassandthegeneralizationassociation.

-levelDiagramofOntologyDesigninPetroleumRemediation

Domain

LevelOntologyDesigninPetroleumRemediationDomain

(SamplePortion)

mple,themiddle-level

classificationdiagramoftheclassofDecomposableObjectisshowninFigure5

andthelower-levelclassificationdiagramoftheclassofMediaisshowninFigure

6.

evelOntologyforMediaClass

Inadditiontothetaxonomyofclass,ontologiesalsoincludeallrelevant

constraintsbetweenclass,attributevalues,instancesandrelations(oraxioms).

mple,in

Figure6,theOCLexpressionfortheclassofSoilspecifiesallthepossibletypesof

theclassofMedia,twonotesareattachedindicatedbythedotted

cifiesthepossiblevaluesoftheattributeofHydraulicConductivity

andtheotherprovidestheformulaforcalculatingthevolumeofthesitemedia.

4.2.2Relationship

RelationshipsamongtheclasscanbereprentedintheClassdiagraminUML.

Figure7isasampleportionoftheRelationshipdiagramdepictingfiveclassin

thepetroleumremediationdomainontology.

4.2.3Process

possiblewaytoexplainthestepsoftheprocessistouanaturallanguage

,theactivitydiagramisudtoreprenttheprocessfromthe

mple,fromtheactivityview,the

processofDeterminetheMediaSizeinvolvestwosteps:(1)tomeasuretheareaof

thecontaminatedsite,(2)tocalculatethevolumeofthecontaminatedsite

accordingtotheformulalistedintheclassificationhierarchy(eoneofthenotes

attachedtotheclassofMediashowninFigure6).TheUMLactivitydiagramfor

eprocesscanalsobedescribedfromthe

ninFigure9,whentheareaofthecontaminatedsiteis

lessthan1600m2anditsvolumeislessthan25000m3,thesizeofthesiteissmall.

Iftheareaofthesiteisgreaterthan1600m2butlessthan2000m2,thevolumeof

siteisgreaterthan25000m3butlessthan30000m3,thesizeofthesiteismedium.

Otherwi,expressionsareudfortheguard

expressions,whicharetheconditionsformakingthedecisions.

onshipAmongSomeClassinthePetroleumRemediationDomain

inethemediaSize

(Activityview)(OperationView)

sion

Animportantbenefitofdevelopinganontologyforasoftwaresystemisthatit

supportsa“higher”levelofreuthanisusuallythecainsoftwareengineering,

thatis,er,anontologyenables

thedevelopertoreuandshareapplicationdomainknowledgeusingacommon

vocabularyacrossheterogeneoussoftwareplatformsandprogramminglanguages.

Italsoenablesthedeveloperstoconcentrateonthestructureofthedomainortask

rerealizationof

thebenefits,wepropothatintegratingtheontologyintothesoftware

tegrationcanberealizedbyusingUML

applicationillustratesthatUMLandOCLshowpromiforreprentingconcepts,

attributeconstraints,relationshipsandprocessknowledgeufulforboth

knowledgeengineersandsoftwareengineers.

Inourapproach,theuofexplicitdomainontologiescanguidesubquent

phasofthesoftwaredevelopmentprocessbecautheontologyisdirectly

suresthesoftware

developedisgroundedsolidlyintheknowledgemodelderiveddirectlyfromthe

logythatcapturesandreprentsthebasic

conceptsandrelationshipsamongconceptsinagivenapplicationareacanguide

thegenerationofalargenumberofsoftwarecomponentsthatcontributetoafinal

danceprovidedbyanontologyisimportantalsoforlongterm

withOCLcanfunctionasthe

reprentationtoolfordevelopingtheintegratedknowledgeandsoftwaremodel.

FutureworkincludesinvestigatingusingUMLwithOCLforreprentingproblem

solvingmethodsandtaskknowledge.

Acknowledgement

WearegratefulforthegeneroussupportofastrategicgrantfromNaturalSciences

andEngineeringRearchCouncilofCanada.

Reference:

cialIntelligence1982;18:87-127

lationapproachtoportableontologyspecifications.

KnowledgeAcquisition1993;5:199-220

ationalJournalofHuman-

ComputerStudies1995;43(5/6):669-686StuderR,BenjaminsV.R,FenlD.

KnowledgeEngineering:&Knowledge

Engineering1998;25:161-197

Academic,Boston,MA,1988

tativebiochemistryanditsapplicationtothe

:ArtificialIntelligenceandMolecular

ess/TheMITPress,1993,pp.289-325

c

KnowledgeAcquisitionWorkshop(PKAW2000),11-13December,2000,

Sydney,Australia

:///projects/protege