Lars Willighagen and me want to create a mapping between publication types at Wikidata and CSL publication types. This requires an URI for each publication type. I’ve created a pull request to officially define these URIs. This CSL Ontology could later be extended to a more sophisticated data model in RDF but the core part, enough to start with, is definition or URIs. Are there any ideas to further extend/use CSL in RDF?
I’m generally sympathetic, but it’d be helpful to clarify why exactly the mapping is needed. We’ve generally been somewhat reluctant to view CSL as a widely used metadata format rather than a very narrowly targeted one, both because we purposefully want to keep it quite narrow in scope, focused on citation rather than description more broadly conceived, and because we just don’t feel we have the people-resources to provide good stewardship for a metadata format in addition to the CSL language.
None of this is to imply a “no” on this, I’d just like to understand the motivation behind it a bit better before committing.
I agree with @Sebastian_Karcher
On the “stewardship” point he raises, that has actually been a problem with BIBO, so much so that we decided to work with the DCMI to move hosting to them.
I don’t have time to think about this really, much less to do anything on the long-term maintenance front.
My use case is a mapping from Wikidata publication types to CSL publication types. Wikidata has more then 1.000 publication types connected in a multi-hierarchy. To export bibliographic items from Wikidata, this types must be mapped to CSL types. The mapping is currently limited to a hard-coded set of publication types in citation-js (see this discussion). The complete mapping can better be managed by the Wikidata community as part of Wikidata. This requires CSL publication types to be referable in form of URIs. We also need to map contributor types but the list of roles is much smaller and more stable so it can be hard-coded. So the use case is no full ontology, just a list of URIs.
Have a look at this query to get a current list of all Wikidata publication types and their mappings to other ontologies, if available.
Ah OK, so this is literally to be able to cite Wikidata items using citations.js and not to use CSL item types as descriptive metadata. Yeah, that sounds like a perfectly valid use case to me, thanks for explaining.
edit: let’s keep the question of variable on the gh issue.
1 Like