2. Let’s start from here
FRBRer FRBRoo
ISBD
BibO
MARC 21
UNIMARC
RDA
DC
BIBFRAME?
Schema.org/bibex?
Bibliographic RDF element sets
Local
3. Similar things, different povs
• It’s the same bibliographic universe
• With common concepts found in most bibliographic
schema/element sets
• Author, title, subject, format, etc.
• Plus specialized concepts for non-global use
• Musical key, parallel title, etc.
• Allowing semantic maps between particular schema
elements/properties (ontologies)
5. Environment
Many element sets and vocabularies
Don’t need complete ‘schema-to-schema’ maps
Concept-focused maps/ontologies are the
consensus, not the schema boundary
Common concept maps are in process - more can be
created, and viewed as part of a ‘contract’
What’s the common minimal data that you need to
provide to be part of a global service? What else is
necessary for the description?
6. Design strategies
Bottom up, not top down: the evidence of global
consensus lies in the commonality of multiple local
environments
Top down requires agreement prior to evidence of
usage
Some approved elements never get used; MARC
21 has several examples
The consensus may not lie at “the top”, i.e. the
“dumbest” element
7. From local to global (data)
• “Contract” specifies set of properties that data must
interoperate with
• Local data can interoperate via direct mapping, or via
connection to any part of a concept-focused map
• Local data remains in original format for local
applications
• Automatically dumbed-down for global services
using maps
• “Think global, act local” = add mappings from local
properties to global graphs
8. Role of Standards Organization
Build on library community strengths in collaboration
and trust
Maintain “contract” for accepting data in global
service(s)
Consensus identification of component elements
New candidate elements identified by local usage
“Endorsement” mechanism brings new elements
into contract
9. Local to global (development)
• Local development proceeds at own pace
• No need to wait for consensus approval
• Global endorsement necessarily and usefully lags
behind local developments
• E.g. W3C/HTML5; schema.org
• “Tell us what to do”
• Do your own thing!
10. Beware of Zombie Issues
Assumption of ‘records’ as units of management
Records can be inputs or outputs
Round tripping
It’s not about data ‘residence’ in one schema or
another—more of a ‘view’
De-duplication—no more ‘master records’
Data at the statement level is available for many
kinds of aggregation
11. Provenance and Filtering
‘Who says?’ is an essential question when evaluating
statements
Not all data statements are created equal, but
trustworthiness is hard to determine without
provenance
Provenance info is the basis for data filtering
No other technique works quite as well to determine
quality
12. What’s Needed?
Infinite namespaces, without encodings, sequences,
hierarchies
Support for innovation at every level
Commitment to move forward (not back), and to learn
the right lessons from experience
Leadership from institutions and individuals