Customizable RO-Crate Metadata Exporter for Dataverse

A customizable external metadata exporter for Dataverse for exporting dataset metadata as an RO-Crate JsonLD.

RO-Crate is a "community effort to establish a lightweight approach to packaging research data with their metadata [...] based on schema.org annotations in JSON-LD" (Research Object Crate).

A problem in creating an RO-Crate metadata exporter for Dataverse is that the default Dataverse metadata blocks do not exactly map to the musts and shoulds of the RO-Crate metadata specifications. Moreover, Dataverse enables installations to create their custom metadata blocks. While this gives Dataverse installations great flexibility, it adds another level of complication to RO-Crate metadata export, as an exporter with hardcoded mappings will be insufficient for some installations.

This exporter provides a customizable solution to address these issues. It reads the mappings between dataset metadata from a CSV file, which can be edited with a spreadsheet editor to customize the exporter's output.

Customization

The export is customized by editing the file named dataverse2ro-crate.csv. The file needs to be in the same folder as the exporter's .jar file.

The CSV is organized into entity types to facilitate referencing entities from other entities. Each entity starts with a row that containing the entityName and fields source and sourceField containing path from where the properties of the entity will be taken from. The following fields contain mappings between the property that appears in the ro-crate-metadata.json (targetPropertyName) and the field or the value which will be used to populate that field value.

For example:

entityName	targetPropertyName	source	sourceField	value
Author		datasetVersion/metadataBlocks/citation	author
	@id			authorIdentifier
	@type			"Person"
	name			authorName
	affiliation			refersTo:AuthorAffiliation
...

The first row indicates that the properties of Author entity will be taken from the author citation metadata block. The following properties will be mapped for each author.

The second and fourth rows map @id of the author entity in ro-crate-metadata.json to authorIdentifier of each author, and name to authorName. The third row, @type will be Person for each author without need for mapping, indicated by putting the string in quotation marks ("Person").

The last row for this entity, affiliation, is a separate contextual entity for the organization whose @id needs to appear for that author's affiliation property. The reference to AuthorAffiliation entity is indicated with a refersTo: prefix (refersTo:AuthorAffiliation). Then the AuthorAffiliation can be defined separately:

entityName	targetPropertyName	source	sourceField	value
...	...	...	...	...
AuthorAffiliation		datasetVersion/metadataBlocks/citation	author
	@id			authorAffiliation
	@type			"Organization"
...	...	...	...	...

Some entities appear only once but need to take values from separate metadata fields. For example, the dataset needs persistentUrl from dataset for identifier, publicationDate from datasetVersion... For such entities, leave the source and sourceField empty in the first column, and fill those fields for each property separately:

entityName	targetPropertyName	source	sourceField	value
Root
	@id			"./"
	@type			"Dataset"
	identifier			persistentUrl
	name	datasetVersion/metadataBlocks/citation		title
	description	datasetVersion/metadataBlocks/citation	dsDescription	dsDescriptionValue
	license			refersTo:License
	author			refersTo:Author
...	...	...	...	...

As seen in this example, each dataset need to take values from diverse sources such as the dataset itself (indicated as an empty source), dataset version (datasetVersion), citation metadata block (datasetVersion/metadataBlocks/citation)...

The default CSV provided with the exporter can be used out-of-the-box. For reference about the default metadata blocks of Dataverse and what properties are contained in other fields, please refer to Dataverse documentation and the JSON metadata export in Dataverse.

Data entities

Mapping of data entities is currently hardcoded, thus cannot be included in the customization csv.