Support identifying the type of items within a collection

[paulmillar]

What is the problem that your suggestion solves?

The resourceType metadata property contains a single (mandatory) attribute resourceTypeGeneral, which accepts a single value from a controlled vocabulary. The list of acceptable values includes Collection.

In v4.6, Collection is described as:

An aggregation of resources, which may encompass collections of one resourceType as well as those of mixed types. A collection is described as a group; its parts may also be separately described.

It's currently unclear how to describe the type of the items within the collection, particularly if the collection contains a "large" number of items.

What solution might meet your needs?

Adding an additional attribute to ResourceType metadata property that describes the resourceTypeGeneral of the items within the collection, provided these items have a uniform resource type.

This attribute would only be specified if the resource could also be described by including (potentially many) RelatedIdentifier metadata properties, each with type relationType attribute hasPart and the identified resourceTypeGeneral property.

The most immediate use would be for resources of type Collection, but it might be useful in other contexts.

Your name

Paul Millar

Your organization

DESY

What alternatives have you tried or considered?

In v4.6, Collection is described as:

An aggregation of resources, which may encompass collections of one resourceType as well as those of mixed types. A collection is described as a group; its parts may also be separately described.

This description mentions that parts (or items) of the collection may be described separately without specifying how this is achieved.

From my understanding of the DataCite schema, I believe the intention here is to describe items within the collection using a RelatedIdentifier metadata property for each item within the collection. Such relationships would support a corresponding resourceTypeGeneral attribute, indicating the type of this item within the collection.

I have several concerns:

1. The process through which the type of items within a Collection may be described is not clearly described. The above process (including a RelatedIdentifier for each item) is my interpretation, but (for interoperability) the documentation should be updated to make this clear.
2. I believe the RelatedIdentifier is not really meant to be used in this way. It is documented as providing "identifiers of related resources" (my emphasis) and not their types. Although describing the resourceTypeGeneral of the related resource is supported, this information is optional. The consequence is that globally unique identifiers need to be supplied for all items within the collection, which is unnecessary overhead when simply trying to describe the type of the collection's items.
3. The process doesn't scale well. A collection may contain a "large" number of items that are all of the same type. Requiring a RelatedIdentifier metadata property for each item in the collection could greatly inflate the size of the metadata description. If the goal is only to know what are the types (i.e., the resourceTypeGeneral) of collection's items, then requiring these global identifiers provides no useful information.

One possible way to resolve this last concern (doesn't scale well) might be to describe only a single (representative) item within the collection, using a RelatedIdentifier with relationship hasPart. This would still incur overhead, but it would be limited. My concern here is about an expectation that any list of hasPart relationships is complete. Although the hasPart relationship documentation makes no claim (for or against) completeness, such a list of items could easily carry this expectation, particularly if presented to a human (via some UI).

Is there anything else you would like to share?

From a semantic point-of-view, this would be an example of providing additional information based on the type of the resource. I believe that only the resource type Collection is sufficiently flexible to have this problem. Other resource types that could support hasPart relationships either identify a diverse range of types or (as the parts are of a single type) that type is constrained by the collection-like type. For example, a Book resource could be described with hasPart relationship with several BookChapter resources, but the type of these items (BookChapter) is constrained by the resource having the type Book.

My impression is that type-specific metadata is something that DataCite metadata schema currently doesn't support; for example, I believe there aren't any resource-type-specific DataCite metadata schema profiles. This approach would go against adding an additional attribute (to ResourceType) that would describe the type of a collection's items (assuming these items have a uniform type), but perhaps this concept could be sufficiently broad to be used when describing resources of other types (i.e., not just Collection).

What group(s) would benefit from your suggestion?

• DataCite members registering DOIs and metadata and administering accounts
• Developers building integrations with DataCite services
• Harvesters who index DataCite DOIs and metadata
• Individuals or institutions analyzing DataCite DOIs and metadata
• DataCite staff
• Other

If other group(s), please describe.

No response