Upload
haquynh
View
221
Download
0
Embed Size (px)
Citation preview
MotivatingSKOSinKnowledgeGraphsandintroducingSKOSShuttle,aweb
availableSKOSservice.
September7th,2018ExpertGroupMeeting:Software&Tools
14:00-17:00at Scigility,Zü[email protected] willkommen!
Content• IntroductionandMotivation• DefinitionKnowledge-Graph• Navigatingthe“giantgraph”• RDF(short)• SKOS• WhousesSKOSThesauri?• 5UseSKOScases• Thesis• SKOSShuttle@semweb
[Taxonomygovernance&blockchain discussion]
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide2
qDothewellknownrelationalmodelsinevitablyleadtobetterresults?qCanschemamodificationsbeeasilyintegratedintorunningsystems?qDoestherelationaltechnologyeasilysupportinterrelationwithotherexternalsources?
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide5
MotivationKnowledgeGraphs
qThe„value“ofdataisdirectlyproportionaltoitsinterrelationwithotherdata...Inthewebaswellasinintheenterprise.
qPartiallyformalized,stronglyinterrelateddatafacewellformalizedbutisolateddata.
Source/1/
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide6
MotivationKnowledgeGraphs
q KnowledgeGraphsdepict multi-levelstructurespredestined to expressand linkmeaningful dataas entities.
q Themore industry groups willbuild (and publish)their Knowledge Graphs,the higher the granularityof the resultingmain (giant)KnowledgeGraphandthe efficiency of their interrelation.
Source/1/
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide7
MotivationKnowledgeGraphs
2012
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide9
MotivationKnowledgeGraphsWhatisagraph
http://lod-cloud.net/2018
In mathematics,andmorespecificallyin graphtheory,a graph isastructureamountingtoasetofobjectsinwhichsomepairsoftheobjectsareinsomesense"related".Theobjectscorrespondtomathematicalabstractionscalled vertices (alsocalled nodes or points)andeachoftherelatedpairsofverticesiscalledan edge (alsocalledan arc or line)– Wikipedia.
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide10
MotivationKnowledgeGraphsWhatisagraph
(Human)Knowledge isafamiliarity,awareness,orunderstandingofsomeoneorsomething,suchasfacts,information,descriptions,orskills,acquiredthroughexperienceoreducationbyperceiving,discovering,orlearning.
(Human)Knowledge canrefertoa theoretical or practical understandingofasubject.Itcanbeimplicit(aswithpracticalskillorexpertise)orexplicit (aswiththetheoreticalunderstandingofasubject);itcanbemoreorlessformalorsystematic.(Wikipedia)
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide11
KnowledgeGraphs(explicit)Knowledge
Linked-Data?In computing, linkeddata isamethodofpublishingstructureddatasothatitcanbeinterlinkedandbecomemoreusefulthrough semanticqueries (e.g.SPARQL).ItbuildsuponstandardWeb technologiessuchas HTTP, RDF and URIs,butratherthanusingthemtoservewebpagesforhumanreaders,itextendsthemtoshareinformationinawaythatcanbereadautomaticallybycomputers. (Wikipedia)
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide12
KnowledgeGraphsKnowledgeGraphvs.LinkedData
Ahighquality KnowledgeGraphisbuiltusingcontrolledvocabulariestodefineitscomponents(entities).
Controlledvocabulary:artifact(structureddocument)formallyanduniquelydefiningentities,meaningandrelationshipsamongentities.
Ontology:mostcompleteformalizingdocumentwhereentitiesarenamed,formallydefinedandinterrelated.(e.g.OBO,OWL).
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide13
KnowledgeGraphsQualityinKnowledgeGraphs?
Controlledvocabularies provideawaytoorganizeknowledgeforsubsequentretrieval.Theyareusedin subjectindexing schemes, subjectheadings, thesauri,taxonomies andotherformsof knowledgeorganizationsystems (e.g.SKOS).
Controlledvocabularyschemesmandatetheuseofpredefined,authorizedtermsthathavebeenpreselectedbythedesignersoftheschemes,incontrasttonaturallanguagevocabularies,whichhavenosuchrestriction.(Wikipedia)
Thesauri arecontrolledvocabularies,whereentitiesareinterrelatedinastandardwaywitheachotherdefiningaheterachy ofconcepts(ideallyasaDAG– directacyclicgraphs).
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide14
KnowledgeGraphsControlledvocabularies
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide15
Examples of Knowledge Graphs(1/6)
Google’sGraph
Enhancessearchwithrelevant(associated)additional(relevant?)results.
Queriable viahttp://google.com
KnowledgeGraphs
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide16
DBPedia
PublishesWikipedia’sDefinitionsandInterrelationsals“LinkedData”
Queriableviahttps://dbpedia.org/sparql
KnowledgeGraphsExamples of Knowledge Graphs(2/6)
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide17
Wikidata
Publishescommonaccessiblecollectionsofentitiesas“LinkedData”
Queriableviahttps://query.wikidata.org/
KnowledgeGraphsExamples of Knowledge Graphs(3/6)
KnowledgeGraphs
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide18
Europeana
Interrelatedcollectionsofart,fashion,music,photographyandhistoryas“Linked-Data”
Queriable viahttp://sparql.europeana.eu/
Examples of Knowledge Graphs(4/6)
KnowledgeGraphs
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide19
UMLSmany controlledvocabularies inthe biomedical sciences(created1986).
1mioConcepts,5mioNames
Queriableviahttp://sparql.bioontology.org/
Examples of Knowledge Graphs(5/6)
KnowledgeGraphs
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide20
Lindas (SwissFederalArchives)https://lindas-data.ch/indexEN.html)
InterrelatedcollectionsofeGOV Dataas“Linked-Data”
Queriable viahttps://lindas-data.ch/sparql-ui/
Examples of Knowledge Graphs(6/6)
KnowledgeGraphs
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide21
Discussion:
WhatwouldbeaKnowledgeGraphallaboutinyourenterprise?WhendataarenotinterrelatedistagainaKnowledgeGraph?Isthisthesituationwe(still)(partially)havetoday?Wherearethe“edges”inyourcase ?
HowcanaKnowledgeGraphbebuilttoday?
UsingRDF– “Edges”(Statements)
RDF:RessourceRescriptionFramework
OneRDF-StatementisanSPOtriple
Everythingbrokendowntobinaryrelationshipsatseverallevels.
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide22
BuildingKnowledgeGraphsKnowledgeGraphs
Howto(efficiently)query1’205 datasetswith 16’012linksanddozensofbillionsofunique triples?
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide23
NavigatingKnowledgeGraphsNavigatingthe“giantgraph“
Example of federated SPARQLQueryfired from http://factforge.net/sparql :
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide24
NavigatingKnowledgeGraphsNavigatingthe“giantgraph“
PREFIX bif: <bif:> PREFIX edm: <http://www.europeana.eu/schemas/edm/>PREFIX ore: <http://www.openarchives.org/ore/terms/>PREFIX dc: <http://purl.org/dc/elements/1.1/> PREFIX dct: <http://purl.org/dc/terms/>PREFIX dbc: <http://dbpedia.org/resource/Category:>PREFIX foaf:<http://xmlns.com/foaf/0.1/>
SELECT ?surname ?date ?creator ?title { SERVICE <http://sparql.europeana.eu/> {
?cho ore:proxyIn ?proxy; dc:title ?title; dc:date ?date; dc:creator ?creator . FILTER(bif:contains(?creator, str(?surname))).
} SERVICE <http://dbpedia.org/sparql> {
SELECT ?surname { ?musiker dct:subject dbc:Italian_Baroque ;foaf:surname ?surname . } LIMIT 5
}}
WorksofknowncomposersofItalianbaroque
(Limitedto5composers)
1. Download, indexandquerydatasetsfromtheLODCloudlocally,(impractical forsimplequeryresolution).
2. Runafederatedqueryagainstallknownsources,whichisasgoodasthequeryendpointsthatitrelieson.
3. Browseonlinesourcesina“follow-your-nose”way4. UseLOD-a-lot/2/doesnotpreventyoufromknowingthe
knowledgeschema.5. Eachgraphis“shaped”(modelled)differentlyWillyou(effitiently)getyoutheinformationyouaresearchingfor?
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide25
NavigatingKnowledgeGraphsNavigatingthe“giantgraph“
AnentityinRDFisasetofRDF-Statements
dbr:Immanuel_Kant rdf:type yago:SocialScientist110619642dbr:Immanuel_Kant a yago:WikicatGermanAgnosticsdbr:Immanuel_Kant a yago:WikicatPhilosophersOfArtdbr:Immanuel_Kant rdfs:label "ImmanuelKant"@dedbr:Immanuel_Kant rdfs:label " تناكليوناميإ "@ardbr:Immanuel_Kant rdfs:label "इमानु्यएल काण्ट"@sadbr:Immanuel_Kant owl:sameAs dbpedia-wikidata:Q9312Seee.g.http://dbpedia.org/page/Immanuel_Kant tocheckoutallrelationships,resp.attributesoftheentity“ImmanuelKant”
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide26
KnowledgeGraphsRDF-Entity“Kant”
AnentityinRDFisasetofRDF-Statements
dbr:Immanuel_Kant rdf:type yago:SocialScientist110619642dbr:Immanuel_Kant a yago:WikicatGermanAgnosticsdbr:Immanuel_Kant a yago:WikicatPhilosophersOfArtdbr:Immanuel_Kant rdfs:label "ImmanuelKant"@dedbr:Immanuel_Kant rdfs:label "ImmanuelKant"@endbr:Immanuel_Kant owl:sameAs dbpedia-wikidata:Q9312...Siehe http://dbpedia.org/page/Immanuel_Kantumalle Relationen,bzw.Attributen zur Entität “ImmanuelKant” einzusehen.
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide27
KnowledgeGraphsRDF-Entity“Kant”
@prefixdbr: <http://dbpedia.org/resource/>
@prefixdbp: <http://dbpedia.org/property/>
@prefixrdfs: <http://www.w3.org/2000/01/rdf-schema#>
@prefixowl: <http://www.w3.org/2002/07/owl#>
@prefixdbpedia-wikidata:http://dbpedia.org/property/
@prefixyago: <http://dbpedia.org/class/yago/>
prefixes=schemas(namespacesdefinedby
vocabularies)
Namespaces(ControlledVocabolaryorOntology)
An„well organized“Graphcan be traversed /queried quicker
SKOS(SimpleKnowledge Organisation System)is aW3Crecommendation since 2009.
ItisanOWLontology(hencesemanticallyandformerlydefined)https://www.w3.org/2009/08/skos-reference/skos.rdf
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide28
NavigatingKnowledgeGraphsNavigatingthe“giantgraph“
https://www.w3.org/2009/08/skos-reference/skos.rdf
scopeNoteConceptScheme
ConceptsemanticRelation
hasTopConceptinScheme
Collection
OrderedCollection
rdf:List
memberList
Documentation
historyNoteexample
editorialNotedefinitionchangeNote
member
note
member
Labeling
hiddenLabelaltLabelprefLabel
Notationnotation
Semantic
related
narrower
broader
broaderTransitivenarrowerTransitive
Mapping
relatedMatch
narrowMatch
broadMatch
closeMatchexactMatch
Example
Creme X0815
Beauty Products
Creme Y1345
Liquid Products
broader
closeMatch
broadMatch
BeautyCremeX(alternativeLabel)
Source:SeanBechhofer „SKOS:Past,Present andFuture.ESWC2010.
NavigatingKnowledgeGraphsSKOSModel(partial)
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide29
SKOS“1x1”
Eachconcept:•Onepreferredlabel•Manyalternativelabels•Oneormoredefinitions•Oneormorescopenotes
•Interrelatedconceptsbroader/narrower/related
•Conceptsinterrelationbuildapolyhyerarchy
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide30
NavigatingKnowledgeGraphsSKOSExamplefor„Bird“
qDaimlerChrysler(automotive,EU)http://www.daimler.com/en/
qFLUOR(engineering,GLOBAL) http://www.fluor.com/pages/
qROCHE(health,GLOBAL)http://www.roche.com/
qBiotechQuality(biotech,Belgium)http://www.thebiotechqualitygroup.com/en/
qSwissConfederation(admin,Switzerland)https://www.bbl.admin.ch/bbl
qIAEA(atomic,Austria)https://www.iaea.org/qCTCN(climate,GLOBAL)https://www.ctc-n.org/qHealthDirect(health,AUSTRALIA)https://www.healthdirect.gov.au/q…
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide31
NavigatingKnowledgeGraphsWhois using SKOS(Thesauri)since years ?
qAGROVOC:MultilingualagriculturalthesaurusqEARTh:anEnvironmentalApplicationReferenceqEUROVOC:TheEU’smultilingualthesaurusqGEMET:GEneralMultilingualEnvironmentalThesaurusqGETTYAAT:Art&ArchitectureThesaurusqMeSH:MedicalSubjectHeadingsqREEGLE:RenewableEnergyandEnergyEfficiencyPartnershipqSTW:ThesaurusforEconomicsqJURIVOC:ThreelingualthesaurusoftheSwissFederalTribunalqTheBaselRegisterofThesauri,Ontologies&Classifications(BARTOC)listsmorethan2’800Vocabularies.
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide32
NavigatingKnowledgeGraphsSomemajor public RDFSKOSThesauri
BIO-Zen
BIRNLex
SubjectHeadings
HealthDirect
Europeana
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide33
NavigatingKnowledgeGraphs5usecases
https://www.w3.org/2006/07/SWD/wiki/EucBiozenDetailed
BIO-Zenallowsthedescriptionofbiologicalsystemsandtherepresentationofscientificdiscourseonthewebinahighlydistributedmanner.Itisintendedtobeusedbyresearchersanddevelopersinthelifesciences.SKOSisusedinbio-zen fortherepresentationofmanyexistinglifesciencesvocabularies,taxonomiesandontologiescomingfromthe"OpenBiomedicalOntologies"(OBO)collection.Thesizeofallconvertedtaxonomiestakentogetherisontheorderofmillionsofconcepts.
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide34
NavigatingKnowledgeGraphsUsecase1
https://www.w3.org/2006/07/SWD/wiki/EucBirnLexDetailed
BIRNLex:alexiconforneurosciencesIntegratedontologyandlexiconusedforvariouspurposes— someend-user/interactive,othersback-end/infrastructure— withintheBIRNProjecttosupportsemantically-formaldataannotation,semanticdataintegration,andsemantically-driven,federatedqueryresolution.
UsesboththeDublinCoremetadataelementsandSKOSwhereeverpossible
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide35
NavigatingKnowledgeGraphsUse case 2
SubjectHeadings(libraries)services
http://rameau.bnf.frlcsh.info
Webservicesusing SKOS toallowforsimplebrowsingofthesubjectheadings.
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide36
NavigatingKnowledgeGraphsUse case 3
healthdirect AustraliaAdvisingPortal
Theplatformbundlescontentfromover140specializedinformationproviders.Usersbenefitfromvaluableandquality-assuredmedicalinformation,whichismadeaccessiblethroughsmartsemanticsearchapplications.Semanticcapabilitiessuchasfacetedsearchandautocompletionenrichtheportalwithintelligencethatenablenon-expertstodospecializedresearch.
https://www.healthdirect.gov.au
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide37
NavigatingKnowledgeGraphsUse case 4
https://pro.europeana.eu/page/linked-open-data
EUROPEANA
Morethan20millionentitieslinkedinanowndatasetintheEDM(Europeana DataModel)… usingDublinCore,OAI-ORE,SKOS andCIDOC-CRMmodels
Providingaccesstoculturalheritagedomainssuchasart,musicandfashion.
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide38
NavigatingKnowledgeGraphsUse case 5
Thesis:
HavingSKOS*organizingthe“giant”graphwouldstandardizeaccess
QualityControlE.g.viaGraphValidatorsSKOS*:somesuitableSKOSextension
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide39
NavigatingKnowledgeGraphsNavigatingthe“giantgraph“(thesis)
SKOSasaStandardq Thesaurusmanagersenablepractitioners todevelop
thesauriwithlimitedtraining.
q Domainexperts canmodeltheirfieldsofexpertisewithoutITsupport.
q SKOSShuttle– thesaurusmanagementsoftwareaaS iscompliantwithSKOSandensuresinteroperabilityofthesauriwithothertaxonomies,knowledgegraphsandAPIs.
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide40
IntroductiontoSKOSShuttle(3+4slides)https://skosshuttle.ch
SKOSShuttleScenariosq Taxonomygovernanceq Creationand/orMaintenanceofbigThesauriq SearchsupportviaRESTAPIq Blockchainsecuredthesauriq Suitableforeverydomain
q Historyq Artq Musicq Manufacturingq Physicsq …
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide41
SKOSShuttle– webthesauri(2)q Customrelationsq Customattributes “SKOS++”q Unitsofmeasureq OWLextension - “SKOSOWL”(tobefurtherdevelopped)qWorkswith7majorRDFstores:AllegroGraph,Blazegraph,
GraphDB,Jena-Fuseki,MarkLogic,Stardog,Virtuoso
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide42
SKOSShuttle– InformationGovernance
DeliverRDFchangestoconnectedthesauri(prodenvironments)viaRDFPushnotifications
18.Mai2017 Seite43
SKOSShuttle– ArchivingtoBlockchain
SecurelyarchiveRDFchunks(orwholeSKOSThesauri)Toaprivateethereumj instanceoperatedatSemweb
RDFcodeiscompressedanduploaded/downloadedto/fromtheblockchain
Nextstepanddiscussion:
Blockchainissafeand(normallyhighly)redundant.ShoulditbealsohostingaSPARQLengine?Whowillstartdoingit?
18.Mai2017 Seite44
Bibliography/1/:PanJ.Z.,2017ExploitingLinkedDataandKnowledgeGraphsinLargeOrganisations/2/:https://aic.ai.wu.ac.at/qadlod/lodalot/iswc2017/
MotivatingSKOSinKnowledgeGraphs September7th,2018Slide50
Thankyou!