DOI Questions & Answers

Online services

Coming up

DOI Governance Issues

Questions From : Natasha Simons, Griffith University, Presentation at eResearch Australasia, Nov 2011.

NB: a large number of FAQs about DOIs can be found at: http://www.doi.org./faq.html

Also - this page is a work in progress. If you have other questions about DOIs that are not answered here or on other ANDS pages (Cite My Data, Data Citation Resources) please email: contacts@ands.org.au

Questions:

4: What should our DOI minting policy be?

5: What are the costs?

5.1 What is the cost of minting a DOI?
5.2 Is there an ongoing maintenance cost?

6: What are the workflows applicable to a DOI?

6.1 What should our internal workflows for obtaining and storing a DOI be?
6.2 Can this process be completely automated, or would human mediation be advisable?
6.3 If human, should we have something like a ‘request a DOI' check box for new records of digital objects?
6.4 Does the internal workflow apply where the digital objects are other than our RDA records?

7: In a research collaboration, who mints the DOI?

7.1 How can we be sure only one DOI is minted?

8: Versioning. What happens when the dataset is updated or expanded?

8.1 Where a dataset which has already got a DOI, is updated or expanded (or changed in any way), should a new DOI be issued for the updated/expanded dataset?
8.2 If a new DOI is minted for an updated / changed dataset, how do you refer to the "old" dataset in the metadata record?

1: What material should have a DOI?

1.1 Q: Would we want DOIs for all collection records we send to RDA?

A: A DOI assumes:

This data will be exposed
This data forms part of the scholarly record
This data can be kept persistent
The minimum metadata required by DataCite can be supplied.http://ands.org.au/cite-data/resources.html#Using_Identifiers_to_Cite_Data

The Identifier Decision Tree at: http://ands.org.au/cite-data/id_decision_tree.pdf could help in deciding which datasets might benefit from DOIs or which ones simply need a Handle.
Back to Top

1.2 Q: What about DOIs for each digital object within a collection (i.e. item level records)?

A: Firstly consider the data users: would they expect to use (and thus cite) the data at collection or item level? An institution might consider citing at multiple levels:

Parent collection DOI
Data item DOIs

Collections are intrinsically more citable than items, and collections can impose indexing on items to retrieve them.

1.3 Q: What about DOIs for digital objects that are not in RDA but are in our repository (e.g. theses, reports)?

A: There is a well established infrastructure for assigning DOI's to publications like theses and reports (e.g. Crossref). For other data objects you should be able to use Cite My Data to mint DOI's provided the citeable item meets the requirements described above.
Back to Top

1.4 Q: Is there a way of finding of how many DOIs we have (at some point when we have some) and getting a list of these?

A: Data owners can ask services@ands.org.au and ANDS will also discuss requirements for adding more query functions to the Cite My Data service. Data owners can also query the DataCite Metadata Store (MDS) web services directly.
Back to Top

1.5 Q: How many DOIs can we mint? How many should we mint?

A: DOIs are free to mint and are effectively infinite. The number to mint will depend on the ability of the institution to keep all datasets with a DOI persistently available and whether the data meets the criteria for DOIs - see http://ands.org.au/cite-data/resources.html#Using_Identifiers_to_Cite_Data .
Back to Top

1.6 Q: What about digital objects we hold that already have a publisher-issued DOI? Should we mint one as well because it will point to our landing page and not the publisher's landing page?

A: DOI.org states that "It is the intention that wherever practical only one DOI name should be assigned to a specified entity." http://www.doi.org/faq.html#46 The FAQ gives examples and warnings of where multiple DOIs may be issued for the one object.

Do not mint DOIs for the sake of getting a different landing page. This violates the point of persistent identifiers, which are meant to be at FRBR (Functional Requirements for Bibliographic Records) Manifestation rather than FRBR Item level. Instead, look closely at Open URL [http://en.wikipedia.org/wiki/OpenURL], which is how the publishing industry already addresses resolution of DOIs to multiple locations. Data owners could also use the optional metadata to build references to other DOIs forming part of a collection or dataset changing in time.
Back to Top

2: Should we store the DOI?

2.1 Q: The DOIs will display in our collection records in Research Data Australia, but should we also store the DOI in our systems?

A: Storing locally is a decision data owners will need to make based on local factors but the DOI needs to be integrated into your data management, otherwise you will keep using fragile URIs to refer to it instead. The DOI once minted is also stored in RDA at ANDS and also in the global DataCite and DOI infrastructure.

If ‘yes', where to store? In our repository or our metadata store (Research Hub)? Or both?

A: Both - for different reasons. The metadata Sore becomes system of record for referencing data from the outside world; repository is where the fragile URI is mediated by the persistent identifier.

If ‘yes' should we display as well as store?

A: Yes as this encourages citation using the DOI rather than the URI.
Back to Top

3: Where should the DOI point?

3.1 Q: Should the DOI point to a landing page or to an actual digital object?

A: A DOI should point to a landing page (catalogue page) for the dataset, rather than to the data itself. A landing page generally contextualises objects and provides at least the following information:

Preferred citation
Descriptive metadata
License information for reuse of the data
A link to an accompanying paper
How to access the data

eg The DOI 10.4225/02/4E9F5DE549B8D resolves to this landing page: http://www.genomics.csse.unimelb.edu.au/GSS/input-genesets.html
Back to Top

4: What should our DOI minting policy be?

A: A policy might be:

Where storage of and access to the material will be long-term and managed by the institution.
Where the author or primary contributor of the material is a staff member or student.
Where the material does not already have a DOI.

Some policies of interest:

DOI.org Policy on DOIs: http://www.doi.org/handbook_2000/policies.html
ETH Zurich: policy for the registration of Digital Object Identifiers (DOIs): http://www.doi.ethz.ch/EN_DOI_Policy.pdf

These policies are a good starting point, although the final choice for each data owner will also depend on local factors.
Back to Top

5: What are the costs?

5.1 Q. What is the cost of minting a DOI?

A: There is no cost to mint DOIs through the ANDS Cite My Data machine-to-machine service for publicly funded Australian research organisations or government agencies that make their data available to the Australian Research Data Commons.
Back to Top

5.2 Q: Is there an ongoing maintenance cost?

A: There are no maintenance costs in terms of the DOI. However, there are other associated costs for persistence as part of routine data management. Any change to the URI must be notified to the DOI provider and this is only partly automated. See also: http://www.doi.org/faq.html#30
Back to Top

6: What are the workflows applicable to a DOI?

A: ANDS M-2-M service provides an easy, simple workflow for obtaining a DOI: see: http://ands.org.au/services/cite-my-data.html

6.1 Q: What should our internal workflows for obtaining and storing a DOI be?

A. Minting workflows will vary widely from data owner to data owner. larger institutions will probably prefer a fully automated system that integrates well with other systems and provides management reporting and auditing.
Back to Top

6.2 Q: Can this process be completely automated, or would human mediation be advisable?

Software supporting these workflows will need to be able to access a M-2-M web service (Cite My Data) and possibly the DataCite MDS system. As dedicated data portals become more widely available (e.g. PANGAEA) local systems may need to inter-operate with those too.
Back to Top

6.3 Q: If human, should we have something like a ‘request a DOI' check box for new records of digital objects?

A: Manual workflows may be suitable for small institutions or data holdings, but an automated approach will likely be needed for larger institutions.

6.4 Q: Does the internal workflow apply where the digital objects are other than our RDA records?

A: This will depend on the data owner's use of RDA and RIF-CS, especially RIF-CS V1.3.
Back to Top

7: In a research collaboration, who mints the DOI?

A: There are no hard and fast rules for these kind of arrangements--it may be easiest to have a prior agreement on who mints the DOI just as the institutions might have for who is lead author on a publication or submits a publication."
Back to Top

7.1 Q: How can we be sure only one DOI is minted?

A: You can't. To ensure that only one is minted it is best to have agreement between the collaborating institutions as to which one will be responsible for minting the DOI.
Back to Top

8: Versioning. What happens when the dataset is updated or expanded?

8.1 Q: Where a dataset which has already got a DOI, is updated or expanded (or changed in any way), should a new DOI be issued for the updated/expanded dataset?

A: Yes -- many datasets will change in version, scope and content over time so it's important to have a strategy for dealing with change. In most cases the data owner will want to issue a new DOI for the changed dataset.
Back to Top

8.2: If a new DOI is minted for an updated / changed dataset, how do you refer to the "old" dataset in the metadata record?

A: If a new DOI is minted for a changed dataset (recommended) it is suggested that the metadata ‘RelatedIdentifier‘ (and other metadata as needed) be used to refer to the previous versions of the dataset.
Back to Top