New Format: RO-CRATE Metadata

[ross-spencer]

Format name

RO-CRATE Metadata

Version number

Multiple versions are available including early drafts, but I don't know the best approach to recording these in PRONOM, whether to record each one individually, or separately, or which versions specifically to record.

Four primary versions are listed in side-menu of the RO-CRATE spec which would cover the last six years:

• https://www.researchobject.org/ro-crate/specification/1.1/index.html

Extensions

• jsonld
• json

MIME/Media Type

• appllication/json
• application/ld+json

Description

RO-CRATE metadata (ro-crate-metadata.json or ro-crate-metadata.jsonld) is the primary metadata object of a RO-CRATE package ("crate"). RO-CRATE metadata provides metadata about a dataset, its authors, and objects, using linked open data principles encoding as JSON-LD to provide a graph of this information. The existence of a RO-CRATE metadata object in a folder signals that the folder and its contents may be a RO-CRATE crate.

Format type

• Text (Structured)

Vendor

The researchobject.org community.

File format identification signatures

This is still a work in progress, but based on the Twitter JSON identifier, the pattern should probably be:

<opening JSON bracket><wildcard><@context"><wildcard><RO-CRATE schema uri>

E.g.:

{*"@context":*"https://w3id.org/ro/crate/

7B*2240636F6E7465787422*2268747470733A2F2F773369642E6F72672F726F2F63726174652F

This will identify RO-CRATE generically, but the version number could be included in the schema URI to identify specific versions.

The addition of a wildcard between "@context" and "https" denotes that the context can be encoded in an array as well as a string, e.g.:

"@context": [ "https://..."

This is likely to have a variable amount of white-space depending on formatting and linting, e.g. JSONLint.com will format this form with ~13 whitespace characters in-between the left bracket and the https schema.

Relevant links, documentation, extra information

I have captured much of my work on this here previously including two sample snippets that can be saved into sample files: http://fileformats.archiveteam.org/wiki/RO-CRATE

Credit

Ross Spencer