This looks fine to me, though I would think we want all variables, not just roles, included in the ontology, no?

@adam3smith probably not. A lot of this stuff is designed to be modular, so they probably already have what they need in terms of properties (what you call variables).

FWIW, I created BIBO as a preferred citation-oriented RDF ontology, but if there's some need for this lighter-weight alternative, I don't have any objection.

A lot of this stuff is designed to be modular, so they probably already have what they need in terms of properties (what you call variables).

My understanding is that this is about mapping between existing entities though, not using CSL to provide non-existent items/properties. But I might be misunderstanding. I have asked for clarification over on discourse.

The list of CSL variables is very specific to CSL so there is unlikely a simple 1-to-1 or n-to-1 mapping from other data sources and formats to CSL. Roles, publication types, and categories are easier to map and they are more likely to be expanded if needed.

Hmm, sure, but how can you create citations without such a mapping? That needs to be somewhere anyway, no?

@adam3smith the mapping from Wikidata data model to CSL data model is hard-coded in citation.js. It is mainly based on Wikidata properties such as author that don't change that often. The mapping from Wikidata document types to CSL document types, however is more fluid because new document types are added regularly. At the moment there are almost 1500 document types in Wikidata so I'd like to put the mapping to CSL document types editable into Wikidata like already started for other ontologies.

@nichtich great, thanks for clarifying. See @zuphilip's question about whether type itself needs to be included and either clarify or accept the suggestion. With that, I think we're good to merge here.

This PR will then redirect the URI https://citationstyles.org/ontology/type/article to https://citationstyles.org/ontology/ but also any other type or even classification or roles. Wouldn't it be nicer when the URI stays unchanged when called? And when maybe at least the title is different for https://citationstyles.org/ontology/type/article and https://citationstyles.org/ontology/type/book?

@zuphilip thanks for review and addition. Sure it would be nice when the URI does not resolve in a redirect and also returns proper RDF. But this is nice to have only. The primary purpose of this PR is to document the existence of official URIs. For human readers common redirect target page is enough.

To what extent should the URIs be permanent? The current URLs don't have the CSL version in them, and beyond expansions we might rename or remove things in future CSL releases.

URIs should be as permanent as CSL and resolve to the most current definition of a concept. New URIs may be added with new CSL versions but I rarely doubt that the general meaning of a category, role, or publication type changes between CSL versions.

(looks fine to merge)

Alright, let's do this then. @nichtich let us know once you have this implemented in WikiCite(?)

Thanks, working on it. You will later get a mapping with schema.org via Wikidata for free.