RSD CodeMeta Crosswalk

Some Generic Information

• The metadata “schema” of RSD is basically their relational database/API. There is no schema in the sense of XML or JSON.

  So there is no simple one-to-one mapping between key/value pairs. It could be a simple column on a table, it could be a foreign-key relationship, or something else.

• The API of RSD is documented here: https://research-software-directory.org/swagger/

• CodeMeta Terms

• FairMQ Example entries:
  RSD: https://helmholtz.software/software/fairmq
  CodeMeta: https://github.com/FairRootGroup/FairMQ/blob/master/codemeta.json

Problematic

• The “description” on RSD is markdown (good). the description field in codemeta isn’t specified (to my knowledge).

  • We could decide that “OSSR flavoured CodeMeta” uses markdown.
  • How does this interact with zenodo? Do we need to change our converter?
    • [Thomas] Zenodo description is html. for the eOSSR, we convert our markdown description to html before uploading to Zenodo.
    • can make that part of the official converter (as an option)

• The URL on RSD itself has a part (I think, those thingies are called slug): For example the fairmq in https://helmholtz.software/software/fairmq

  • If an RSD entry would be generated “from scratch” from an OSSR entry, we would need to come up with a slug.
  • If the RSD entry already exists, one could put the URL as a relatedLink on codemeta. Maybe the url or samsAs.

• Related publications / etc. (scroll down to “Mentions” on the RSD page).

  • codemeta allows relatedLink

• Can’t we extract a list of terms from the swagger API (e.g. as json)?

Proposals Ready

“Participating organisations” on RSD

RSD has a concept of “Participating organisations”. This could be represented as “Organization” entries in the maintainer field of codemeta.

• See this Pull Request on FairMQ where Christian added this for FairMQ.

• Our codemeta-to-zenodo-converter has been fixed to ignore anything but "Person"s in contact fields here.

• see comment on maintainer above, a Participating organisation is most probably not a maintainer

• [Thomas] is maintainer really the right / most common place for Organizations? - Codemeta currently supports ony Person (a physical person is really maintaining software more than an org, no?) - Orgs can be author, contributor, copyrightHolder, creator, funder, publisher, sponsor

  • codemeta 4.0 should hopefully allow Orgs in the maintainer field: See Pull Request on Codemeta

“Projects” on RSD

We could probably use the project URL (at RSD) as a isPartOf entry?
We would match the URLs in isPartOf against an rsd-project-URL and only convert those.

Meeting discussion

• we would need a documentation / description of the terms used in the database (otherwise we have to guess the meaning of every term) --> ask RSD
• in RSD, what version appears? is description automatically updated for example? can it be? If we use URL in the description, is it dynamically rendered?
• slug --> how to generate it? --> ask RSD
• mentions --> ask RSD

Metadata list

Draft here: https://gitlab.com/escape-ossr/eossr/-/blob/rsd/eossr/metadata/schema/escape_codemeta_crosswalk.csv
(see RSD column)

• version --> ?
• author
  • only one in RSD
• maintainer
  • this is the RSD maintainer (able to edit the RSD entry) != software maintainer
• contributor
  • in RSD, has a role --> need to map each type of person in CodeMeta to that contributor role
  • role can be anything (text entry)
• datePublished
  • updated_at ?
  • is this automatically retrieved or can be edited ?
  • does this appear somewhere?
  • is this internal = date of the RSD entry, not of software publication
• license
  • what RSD uses to find licenses? SPDX?
  • codemeta is either CreativeWork or URL
  • can be lists in both cases (CodeMeta)
• isPartOf
  • if points to RSD project URL, can be converted
• participation organisation
  • if org.id in CodeMeta is a ROR URL, can be converted (orgs must be fetched from any of author, contributor, copyrightHolder, creator, funder, publisher, sponsor)