RDA FTW or WTF
description
Transcript of RDA FTW or WTF
RDA FTW or WTF
LITA LOD SIG ALA AC 2014
ALA Annual 2014 LITA/ALCTS LOD SIG 2
RDA
For The Winor
What The Fheck
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 3
Perspective
(Mine)...and everything I say is arguable
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 4
A very brief history of bibliographic metadata standards
• CHAOS• …(passage of time)…• AACR3• CHAOS (LOD)• RDA• Bibex (schema.org)• BibFrame
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 5
RDF vs. XML (yet again)
• XMLish (bound) data is– Concrete, finite, closed, constrained by Schema• XML Schema, JSON Schema, RelaxNG, SQL, et al)
– Good for creation, storage• RDF (unbound) data is– Flexible, infinite, open, constrained by Logic– Good for distribution, aggregation– VERY likely to be globally invalid (illogical)
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 6
A brief word or 3 about RDF
• Data model assumes an ‘Open World’• Any ‘body’ can ‘say’ any Thing about any Thing• Knowledge has no boundary– There is no ‘record’
• Every Thing that is ‘said’ is ‘true’– Until it’s ‘inconsistent’
…because it’s all about POV
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 7
Anglo-American vs. (everyone else)
• Software Developers are used to English everything
• “That ship has sailed” -- Richard Wallis (English speaker)
• Semantics aren’t consistent, even across English-speaking cultures– Semantics (meanings) matter in bib metadata
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 8
Anglo-American vs. (everyone else)
• Cataloging by/for French (or Chinese) speakers may require different semantics
• Global metadata, especially LOD, requires cross-cultural Linking– This requires cross-cultural mapping (not
crosswalks) of often dissimilar semantics
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 9
MARC 21 vs. (everything else)
• Both semantics and syntax– Semantics has centuries of accumulated value– Syntax is too limited for LOD– Syntax uses opaque identifiers
• Doesn't play well with FRBR– Not even a little bit – doesn’t deal with
abstractions
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 10
RDA Data Model
• RDA instructions are for people• Explicitly multilingual, multicultural– Intense, cross-cultural semantic commitment– Semantics and labels continuously adjusted for
cross-cultural uniformity • Explicitly based on FRBR and DCAM– Wait... DCAM?
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 11
RDA RDF
• 'Opaque' identifiers– http://rdaa:P50029– No ontological commitment– Multicultural– Supports RDA's commitment to cross-cultural
semantic alignment
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 12
RDA RDF
• Lexical Aliases– http://rdaa:founder.en– Language-specific identifiers (URIs)– 'Readable' by humans in each language– Convey minimal semantics (Bad but
whadareyagonnado?)– Always 'resolve', for the machines, to the
canonical opaque identifier• ...and resolve historical aliases
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 13
RDA RDF
• Unconstrained Properties– http://rdau:P60694– Implies no membership in a set (class) of the Thing
Being Described– Required for cross-domain mapping • see bf:Work, schema:CreativeWork, isbd:Resource
– Supersets of the constrained properties
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 14
A brief word aboutSemantic Mapping vs. Crosswalks
• Semantic mapping– Preserves original data context– Assigns meaning to the mapping relationship– This is what the Unconstrained facilitate
• Crosswalk– 1:1 relationship – always ‘same as’– Discards original data context
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 15
RDA & GIT
• Uses git for versioning• Uses GitHub for– Distribution
• https://github.com/RDARegistry/RDA-Vocabularies– Documentation (GitHub pages)
• http://www.rdaregistry.info/– Issue tracking
• https://github.com/RDARegistry/RDA-Vocabularies/issues– Release tracking
• https://github.com/RDARegistry/RDA-Vocabularies/releases
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 16
rdaregistry.info & git
• Nginx server• Hosts the data in ‘resolvable’ form• Content negotiation for all RDF ‘flavors’• Gets updated from GitHub• ‘code’ pulled from GitHub Master branch• Always updated to match current release• Eventually able to request specific release
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 17
rdaregistry.info & the OMR
• Open Metadata Registry– http://metadataregistry.org/schema/list.html?filte
rs%5Bagent_id%5D=177&filter=filter
• Editorial interface• Generates RDF– Uses API from http://rdf-translator.appspot.com/
• Maintains a local git repository• Pushes results to GitHub2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 18
Soo…
What’s the answer?(does FRBR matter?)
At this point in our program Jon becomes a typical talking head…
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 19
BibFrame isn’t the ‘answer’
• Discards MARC 21 semantics• Redefines existing properties• Redefines frbr:work• ‘Borrows’ semantics from RDA without
reference • Discards FRBR semantics• Unique approach to frbr:work• Proprietary approach to MARC 21 mapping2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 20
Schema.org bib extensions (Bibex) isn’t the ‘answer’
• Oriented toward global search engines• Redefines frbr:work• Hard (not impossible) to extend• Constrained by limitations of HTML-based
container
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 21
Schema.org bib extensions (Bibex) isn't the 'answer’
• Hard to map• Hard to translate• What are the instructions?• ‘Unique’ approach to frbr:work• Proprietary approach to MARC 21 mapping
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 22
RDA isn’t the ‘answer’
• Lots of cruft from AACR2 and MARC 21• Minimal community involvement in
development of the data model• Strong commitment to FRBR
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 23
MARC21 isn’t the ‘answer’
• Semantics tied to syntax• Field:indicator:subfield ‘means’ some Thing– See http://marc21rdf.info– 11,078 Things can be said in current 0xx-7xx
• Hard to extend• Can’t be globally extended without local pain
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 24
POV
For decades we’ve shared a single POV:MARC
LOD forces the consideration of multiple POV
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 25
So what’s the answer?
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 26
Harvest globally
• MARC 21+RDA+BibFrame+Bibex+...• The global (open) web of data is full of:• Known knowns• Known unknowns• Unknown knowns• Unknown unknowns
– We can exploit that… be flexible– The Robustness principle
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 27
Process Locally
• Aggregate– To integrate and surface inconsistencies
• ‘Validate’ – according to local knowledge
• Map– To apply your knowledge to the unknown
• Cherry-pick– To create variable ‘boundaries’
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 28
Publish Globally
• MARC 21+RDA+BibFrame+Bibex+…• With consistency and precision• Make your data ‘knowable’ • The Robustness principle
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 29
Controversy
“…Where there is conflict, let me sow harmony,Where there is doubt, let me sow faith,Where there is despair, let me sow hope,Where there is darkness, let me sow light,…”
2014-06-29
ALA Annual 2014 LITA/ALCTS LOD SIG 30
Cheers!
Aka Thanks!
http://managemetadata.com
2014-06-29