Investigate strategies for automating DOI minting

[mjordan]

In order to mint a DataCite DOI, an object must have the following metadata properties:

• title
• creators
• publication year
• publisher
• resource type

Most objects would have the first four properties, and provided values for those elements are present, we could mint a DOI without any user intervention. But, 'resource type' must be one of the following values:

• Audiovisual
• Collection
• Dataset
• Event
• Image
• InteractiveResource
• Model
• PhysicalObject
• Service
• Software
• Sound
• Text
• Workflow
• Other

Currently, the DataCite DOI module injects a required field into the node add/edit form that forces a user to choose one of these values. That works fine when we manually mint DOIs manually on a single object. There is also an Action that allows the minting of batches of DOIs using Views Bulk Operations, but it requires that each object in the batch has the same resource type.

So, our challenge is, how do we assign one of these resource type values for each Islandora object in automated or large-scale batch workflows?

• Always use "Other"?
• Require each repository to add a field to their Islandora objects' content types and make that field required?
• Provide some way of allowing admins to map from whatever "type" or "genre" vocabularies they are using to this list?

Having the user choose one of these in the UI works for mediated/manual DOI minting workflows, but if we want to mint a DOI on object creation, or mint DOIs for a large batch of objects, we need to avoid the manual assignment of resource type values on each object.

[mlhale7]

Don Richards brought this to my attention and I wanted to share my two cents. I think the type is likely to be either "Text" or "Dataset" in most cases (for published journal articles, textual data or numeric datasets), but it would be good to give admins some control. In MODS these materials would likely have "text" or "software, multimedia" (for numeric data) as a typeOfResource value. I'd say that "form" could be more useful than "genre", but that's considering the data we have at UTK if you wanted to look beyond typeOfResource. "Other" isn't technically wrong, but it's something a metadata person would eventually want to take time to update. If it were me, I would like to have some control over the value assigned. Some institutions may not add any datasets at all so they might prefer "Text" as the default if they are primarily ingesting articles. I'm not that familiar with the larger context of this issue, so let me know if anything is not relevant. I've only worked with CrossRef, not DataCite, and I'm unsure if this issue relates more broadly to published articles / IR content.

[mjordan]

@mlhale7 thanks for the response. The challenge is that DataCite requires the use of their list of resource types, which will not likely be the same as the genre or form terms that are used in the local Islandora. I like the idea of allowing local admins to establish a default. But, the problem really boils down to what value from DataCite's list is assigned when the form that provides the list is not used, e.g. in automated or batch DOI assignment. A default value would be useful for many cases but if used in automated or batch assignments it would also result in some cases in an inaccurate term being chosen.