Embed Size (px)
DESCRIPTION
Short talk for the session and panel discussion: "DATA ENRICHMENT AND TRANSFORMATION IN THE LOD CONTEXT: POOR AND POPULAR VS. RICH AND LONELY—CAN'T WE ACHIEVE BOTH?" at DCMI Conference Lisbon 2013
Citation preview
VOCABULARY USAGE IN
DATOS.BNE.ESDaniel Vila-Suero
REUSE AS MUCH AS POSSIBLEMaximize the coverage of mappings from MARC 21 to RDF
6 CLASSES
14 OBJECT PROPERTIES
>200 DATATYPE PROPERTIES
FROM MORE THAN 10 DIFFERENT VOCABULARIES
String
MAPPINGS ARE PUBLICLY AVAILABLEhttp://bne.linkeddata.es/mapping-marc21
BE PRAGMATIC
DCMI ElementsDCMI TermsIFLA FRAD
IFLA FRBRerIFLA FRSAD
IFLA ISBD ElementsMADS/RDF
RDA Group 2 ElementsRDA Relationships for WEMI
SKOS...
WHAT IS THE CORE DATA MODEL ?IFLA FRBR
Ok, and how does the data look?
frbr :Person frbr :Work
frbr :Expressionfrbr :CorporateBody
frbr :Manifestation
skos:Concept
Group 2 Group 1 Group 3
Don Quijote de la ManchaFrench manifestations
(213)
Novelas EjemplaresSpanish manifestations
(303)
Don Quijote de la ManchaSpanish manifestations
(840)
Don Quijote de la ManchaEnglish manifestations
(247)
Don Quijote de la Manchafrbr:Work
Miguel de Cervantes
Don Quijote de la ManchaGerman manifestations
(49)
EntremesesSpanish manifestations
(86)
frbr:Work frbr:isEmbodiedIn frbr:Expression
frbr:Expression frbr:IsManifestedBy frbr:Manifestation
frbr:Person frbr:isCreatorOf frbr:Work
( ) Number of resources
WHY THIS?
frbr :Person
frbr :Work
frbr :Expression
frbr :Manifestation
AND NOT THIS?
Person
Bib. resource
1. Nº OF AUTHORITY RECORDS
0
500.000
1.000.000
1.500.000
2.000.000
frbr :Work frbr :Person frbr :Expression
Nº of records
frbr :Work frbr :Person frbr :Expression
2. CLUSTERING OF RESOURCES IS EXPLICIT IN THE DATA (AND THE MODEL)
Manifestations
Person
Works
Expressions
++ NAVIGATIONhttp://bne.linkeddata.es
3. LINKS TO OTHER DATASETSAlso at Work and Expression levels
(VIAF, idRef, etc.)
LINKS TO EXTERNAL DATASETSAlso at the Work and Expression level
(VIAF, idRef, etc.)
bne:XX3383563
viaf:184295284
SOME CLICKS AFTERThe user can access the “curated and reliable” cluster of editions
of Don Quijote de la Mancha in Italian
SOME ISSUES
• Pure FRBR representation not always possibledct:subject, dct:language at Manifestation level instead of Work level
• Around 30% of bibliographic records still not connected to their Expression
• Manifestations are mostly described with strings, not many links
• Does the modelling seem to complex to people outside the library community?
NEXT STEP
Data portalAPIs
MARC 21, XML, metadata schemas
BNE Knowledge Graph(RDF with rich data model)
For humansand machines (schema.org)
For developersJSON(LD)
LDP, LD API
SPARQLHTTP
RICH
POOR?
MORE INFO AT
• http://datos.bne.es
• “datos.bne.es: A library linked dataset”. Vila-Suero et al., Semantic Web Journal 2013
• “datos.bne.es and MARiMbA: An insight into Library Linked Data”. Vila-Suero and Gómez-Pérez. Library Hi-tech, to appear
2013
THANK YOU VERY [email protected], @dvilasuero
