Scalable Identifiers for Natural History Collections
Scalable Identifiers for Natural
History Collections
12 August 2012
University of California Curation Center
California Digital Library
California Digital Library
Serving the University of California
• 10 campuses
• 360K students, faculty, and staff
• 100’s of museums, art CDL supports the research
galleries, observatories, marine lifecycle
centers, botanical gardens • Collections
• 5 medical centers • Digital Special Collections
• 5 law schools • Discovery & Delivery
• 3 National Labs • Publishing Group
• UC Curation Center (UC3)
What EZID data citation offers
• Precise identification of a dataset (DOI, ARK)
• Credit to data producers and data publishers
• A link from traditional literature to the data
• Exposure and research metrics for datasets
(Web of Knowledge, Google)
EZID: Long term identifiers made easy
Take control of the
management and distribution
of your research, share and get
credit for it, and build your
reputation through its collection
and documentation
EZID: Long term identifiers made easy
Take control of the
management and distribution
of your research, share and get
credit for it, and build your
reputation through its collection
and documentation
DataCite
German National Library of Economics (ZBW)
German National Library of Science and Technology (TIB)
Canada Institute for Scientific and Technical Info. (CISTI)
German National Library of Medicine (ZB MED)
Technical Information Center of Denmark
GESIS - Leibniz Institute for the Social Sciences, Germany
Institute for Scientific & Technical Information (INIST-
Australian National Data Service (ANDS)
CNRS), France
ETH Zurich, Switzerland
TU Delft Library, The Netherlands
The Swedish National Data Service (SNDS)
The British Library , UK
California Digital Library (CDL), USA
Office of Scientific & Technical Information (OSTI), USA
Purdue University Library
EZID Clients
A current, partial list
UC Berkeley Library (on behalf of the UC Berkeley The Digital Archaeological Record (tDAR)
campus) Sponsored accounts:
Open Context Dryad Digital Repository
CRCNS.org
UC San Diego Library (on behalf of the UC San Diego Fred Hutchinson Cancer Research Center
campus)
American Astronomical Society(AAS) LabArchives
Centre national de documentation National Center for Atmospheric Research
(NCAR)
pédagogique(CNDP)
Cornell Institute for Social & Economic USGS/Earth Sciences Data Clearinghouse
Research (formerly National Biological Info. Infrastructure)
New features in development
• Suffix pass-thru: do NT and get N/ST/S for free
• Service replicas: manager and resolver
• Content negotiation and inflections: ? ?? / .
• URN (Uniform Resource Name) support (urn:uuid:)
• ARK community and governance, eg, registries
Some identifier dimensions
• registration (storing and updating ids for
resolution)
• non-registration (id awareness via rules)
• persistence flavors
• resolution
• clusters (closely coupled ids)
• other relations (part, whole, related)
Identifier generation
• inspiration ("I think I'll call it MyKitty/Photos")
• systematic inspiration (title/author/vol/issue)
• counter (421, 422, 423, ...)
• timestamp
• hash computed over content (MD5, SHA256)
• hash of randomized timestamp plus registry
(uuidgen, noid)
• randomized counter plus registry (EZID/noid)
Identifier registration
• use filesystem tree as resolver (any old
website)
• use web server config file
• use web server backing database
• use a service (bit.ly, EZID, DataCite, local
Handle service)
Identifier non-registration
Identifiers “exposed” but not registered, eg,
awareness via rules
• extension (abc/def is "part of" abc)
• parameter (abc_N_M works for N or M less
than 100,000)
• general query (arbitrary data cells)
Identifier persistence flavors
• persistent id to very dynamic content
(eg, home page)
• persistent id to stable but correctable content
(eg, landing page)
• persistent id to never-changing content
(eg, spreadsheet)
– persistent ids to non-recommended content
• persistent id to stable but growing content
(serial pub)
Identifier resolution
• DNS (domain names)
• DNS + HTTP (any website)
• DNS + HTTP + redirects (eg, URL
shorteners, N2T/EZID system)
• DNS + HTTP + redirects + Handle resolver
(DOIs and Handles)
GUID Definitions
• GUID -- Definition 1 (wikipedia)
– A 128-bit id generated per RFC 4122, eg,
– uuidgen -> EEF45689-BBE5-4FB6-9E80-
41B78F6578E2
• GUID -- definition 2 (earth sciences?)
– any globally unique identifier
Service replicas
• EZID is an id manager that populates N2T
– It tolerates down time
– Other id manager services might one day populate N2T
• N2T (Name-to-Thing) is an id resolver that ...
– It is very intolerant of down time, since it services all
access requests for locations and metadata
– N2T replicas underway
URN support
• N2T and EZID are agnostic about kinds of
things, names, and metadata
– Digital, physical, abstract, living, fictional, groups, etc.
– Any metadata & known profiles (DataCite, Dublin Kernel)
– ARK, DOI, URN, Handle, IVOA, LSID, PMID, etc., requiring
namespace “write” permission, eg, via DataCite
• In test: Uniform Resource Names (URNs)
– urn:uuid namespace
Under the hood keysmithing terms:
bows, shoulders, blades, tips, covers
Suffix pass-thru: NT gets N/ST/S for free
Idea: if name N points to target T, then requests for N
extended by any suffix N/S can take you to T/S
• For dataset doi:10.5072/Big4 with 10,000
nameable components,
– Register and manage 10,001 names or 1 name?
– Eg, http://x.y.z/foo/Big4/db/table/cell/45-8.txt could be
reached with doi:1.5072/Big4/table/cell/45-8.txt
• In test with ARKs. Conflict with other resolvers?
Tombstone and other surrogate pages
Tombstone, incubation, and other surrogate pages
(probation?) auto-generated from metadata, eg,
http://n2t.net/ezid/tombstone/id/ark:/20775/bb3243444z
Reserved identifiers and multiple targets
• Some ids must be created and managed (reserved)
before going public, eg, for manuscript preparation
• In test: infrastructure for multiple targets and
multiple instances of any metadata element
• What should user experience be for multiple targets?
– Present a menu of targets (burden of choice)?
– One target chosen for them (burden of inflexibility)?
Identifier (ARK) inflections: ? ?? / .
• Inflect: change endings w.o. creating new words
– Terminal ? means “I want metadata”, which is similar to
linked data content negotiation (also in EZID test)
– Terminal ?? means “I also want support metadata”
– Drawing board: / could mean “I want a landing page”
and . could mean “I want the usual computable thing”
• Allow inflections beyond ARKs to DOIs/URNs?
Example: http://n2t.net/ark:/13030/qt0349g1rh?
Renninger, Heidi; Phillips, Nathan; Hodel, Donald. “Comparative hydraulic and
anatomic properties in palm trees (Washingtoniarobusta) of varying
heights”. 2009-04-29. ark:/13030/qt0349g1rh
HTML content with
embedded comments in
ANVL/ERC and RDF
erc:
who: Renninger, Heidi,; Phillips,
Nathan,; Hodel, Donald,
what: Comparative hydraulic and
anatomic properties in palm
trees (Washingtoniarobusta)
of varying heights
when: 2009-04-29
where: ark:/13030/qt0349g1rh
ARK community and governance
• ARK mailing list: arks-forum@googlegroups.com
• Topics: governance, community, standardization
• Registry maintenance: shoulders and NAANs
• N2T consortium with alternative EZID-like services