資源描述:
《多領(lǐng)域本體的支持如何重用這些本》由會(huì)員上傳分享,免費(fèi)在線(xiàn)閱讀,更多相關(guān)內(nèi)容在教育資源-天天文庫(kù)。
1、AutomaticallyOntologyConstructedFromOnlineOntology2008-12-25數(shù)據(jù)庫(kù)的語(yǔ)義檢索需要多領(lǐng)域本體的支持。如何重用這些本體?WEB上越來(lái)越多的InterLinkingRDFdata,粗略估計(jì),RDF三元組的數(shù)量為:2billionInterlinkedbyaround3millionRDFlinks。當(dāng)前基于keyword的語(yǔ)義搜索引擎(swoogle,SWSE,Falcons等)僅返回一個(gè)本體列表,并沒(méi)有進(jìn)一步的提供本體重用支持。Web上越來(lái)越多的本體為用戶(hù)提供API,如YAGO,DBPedia,Gename,D
2、BLP等BackgroundandMotivation本體開(kāi)發(fā)的方式Fromscratch-很困難,需要領(lǐng)域?qū)<业膮⑴cOntologyreuse-一種好的選擇本體的構(gòu)建步驟確定應(yīng)用領(lǐng)域術(shù)語(yǔ)明確術(shù)語(yǔ)間的聯(lián)系、屬性及約束本體的“模塊化”(Modularization)技術(shù)允許用戶(hù)協(xié)同開(kāi)發(fā),導(dǎo)入本體模塊(module),但并沒(méi)有提供抽?。╡xtraction)或合并(merge)用戶(hù)感興趣的本體部分的支持.如何自動(dòng)構(gòu)建面向特定應(yīng)用的領(lǐng)域本體?BackgroundandMotivationArchitectureOntoExtractorOntoRankerExtracto
3、rMap&MergeOnto1Onto2Onton……OWLOntoKBSearchURIsTermsReasonerSPARQLProcessorOWLOntologyTranslatorTranslatingMethodsManytomany-AtableTonlyusedtorelatetwoothertableT1,T2(manytomany)T1->c1,T2->c2,op1.domain=c1,op1.range=c2,op2->domain=c2,op2->range==c1.Referentialintegrityconstraints(i.e.if
4、TisrelatedtoT1byaforeignkeywhichisprimarykeyatthesametime,thenTismappedtoasubclassoftheclasscorrespondingtoT1.)T,T1->C,C1,CsubClassOfC1TranslationfromRDBtoRDF(S)/OWLReferentialconstraints(i.e.ifatableThasareferentialconstraintwithT1,andifc,c1aretheclassescorrespondingtoT,T1respectively
5、,thenweassigntocanobjectpropertyopwhoserangeisc1,andweassigntoc1anobjectpropertyop'whoserangeisc.TopreservetheoriginaldirectionofthereferentialconstraintfromTtoT1,wesettheobjectpropertyopasfunctional)t,t1->c,c1,op.domain=c,op.range=c1,op’.domain=c1,op’.range=c,op.hasFunctionalProperty.
6、Inothercase,tablenamecanbetranslatedintoclass;columnnamesthatarenotforeignkeycanbetranslatedintoDatatypeProperty.TranslationfromRDBtoRDF(S)/OWLDrawbacksWithoutconsideringmoreconstraintinformationandrecordsinformation,examplecolumnsvalueconstraint.Translatingallcolumnnamesthatarenotfore
7、ignkeyintodatatypepropertymaybenotappropriateforsomespecificapplication.TranslationfromRDBtoRDF(S)/OWLSwoogleOntoRankDifferentiallytreatthefollowing4relations.Eachofrelationswereassigneddifferentlyweight.Imports(A,B)User-terms(A,B)Extends(A,B)Asserts(A,B)OntologiesrankingOntokhojCons