Creating Citable Data IdentifiersRyan ScherleMark Diggory
Mimosa house 807 South Virginia Dare Trail Kill Devil Hills, NC USA 27948
1903-12-17 36.019705 N, 75.668769 W
79330-S84-A41 WP0ZZZ99ZTS392124
Loxosceles reclusa
Citing identifiers Mimosa house 807 South Virginia Dare Trail 1903-12-17 27948 Loxosceles reclusa 36.019705 N, 75.668769 W 79330-S84-A41 WP0ZZZ99ZTS392124
Identifiers matter Some identifiers are machine-friendly,
some are human-friendly For citations, you need to strike a
balance Good identifiers are a critical selling
point for an repository
http://purl.dlib.indiana.edu/iudl/lilly/slocum/LL-SLO-009276
Principles ofcitable identifiers
1. Use DOIs http://dx.doi.org/10.5061/dryad.123ab Scientists are familiar with DOIs
1. Use DOIs http://dx.doi.org/10.5061/dryad.123ab Scientists are familiar with DOIs DOIs are supported by many tools and
services
1. Use DOIs http://dx.doi.org/10.5061/dryad.123ab Scientists are familiar with DOIs DOIs are supported by many tools and
services
Current support:Eprints Dspace FedoraNo No With work
2. Keep identifiers simple http://dx.doi.org/10.5061/dryad.123ab Complex identifiers are fine for machines, but
they’re bad for humans. Despite best intentions, humans sometimes
need to work with identifiers manually.
http://dx.doi.org/10.1179/1743131X11Y.0000000009
http://dx.doi.org/10.1016/B978-0-12-220851-5.00003-4
2. Keep identifiers simple http://dx.doi.org/10.5061/dryad.123ab Complex identifiers are fine for machines, but
they’re bad for humans. Despite best intentions, humans sometimes
need to work with identifiers manually.
Current support:Eprints Dspace FedoraYes Yes Yes
3. Use syntax to illustrate relationships http://dx.doi.org/10.5061/dryad.123ab/3 Adding a tiny bit of semantics to an
identifier is incredibly usefulhttp://files.eprints.org/691/http://files.eprints.org/447/http://files.eprints.org/556/
Useful for various human “hacks” Useful for statistics
3. Use syntax to illustrate relationships http://dx.doi.org/10.5061/dryad.123ab/3 Adding a tiny bit of semantics to an
identifier is incredibly useful
Current support:Eprints Dspace FedoraNo No With work
4. When “meaning-bearing” content changes, create a versioned identifier Scientists want data to be invariant to
enable reuse by machines Even a single bit makes a difference Watch out for implicit abstractions…
http://dx.doi.org/10.5061/dryad.123ab/thumbnail
What about DOI conventions?
5. When “meaningless” content changes, retain the current identifier Descriptive metadata must be editable
without creating a new identifier. Humans rarely care about metadata
changes, especially for citation purposes!
Caveat: machine-oriented systems may consider the “metadata” to be data, which requires identifier changes
Current versioning supportEPrints Support for flexible versioning/relationships,
but no support for expressing these relationships in identifiers.
DSpace None.
Fedora Implicit versioning of all data and metadata. This is highly useful, but it is too granular for citation purposes.
Principles of citable identifiers1. Use DOIs2. Keep identifiers simple3. Use syntax to illustrate relationships 4. When “meaning-bearing” content changes, create a versioned identifier5. When “meaningless” content changes, retain the current identifier
Hacking DSpace to support…
DOI identifier registrationSemantics in identifiersCitation publicationVersioning
DSpace identifier services Handle system independence
More future identifier systems will come. Granular control
Separate reservation from registration Citation
Registration of metadata with external services
DSpace identifier services
DataCite content service
Promoting accurate citationsAdded suggested citation formats up front
Versioning Versioning is item “editioning” Creation of new versions is a “user
mediated” process (submitter or reviewer)
Versioning does not alter the original item
Version relationships are maintained independent of the item’s metadata
Submission-based revisions
Result: Citable data versionsdoi:10.5061/dryad.bb7m4
Future technical directions Add metadata versioning under the
hood -- may need to rethink some of the current system
Integrate our changes to core DSpace Moving these features into the core
requires further discussion with the Dspace user community
How are we doing?For 186 articles associated with Dryad deposits:
77% had “good” citations to the data 2% had “bad” citations to the data 21% had no data citations
Standards for data citation are still evolving. Journals have yet to agree on where to place data citations, and authors are just starting to become familiar with the concept.
What should you do now? Analyze how data is used and cited
outside the repository Determine whether use is more
machine-oriented or more human-oriented
Design identifiers and identifier management to facilitate the observed uses