- From: Antoine Isaac <aisaac@few.vu.nl>
- Date: Fri, 9 Jan 2015 16:28:03 +0100
- To: Public DWBP WG <public-dwbp-wg@w3.org>, Ig Ibert Bittencourt <ig.ibert@gmail.com>, Ghislain Atemezing <Auguste.Atemezing@eurecom.fr>, "Mark Harrison" <mark.harrison@cantab.net>, <eric.kauz@gs1.org>
Dear vocabulary section co-editors, all Reading the current section on vocabulary best practices [1] a first issue is about ontologies vs "controlled vocabularies" (aka looser "Knowledge Information systems" as in the page Mark and I had worked on [2]) The two are really different beasts, from a technical perspective, even though the terminologies for them (and their components) are similar, and some best practices (e.g. versioning) apply to both. I think the current wording is not perfect [3], and intend to make suggestions about it. But it still goes a long way explaining the problem. What I miss is an explicit best practice that would call on data publishers to make a clear choice about the semantics of the vocabularies in their data. I.e., roughly, define their vocabularies as RDFS/OWL ontologies or SKOS-level vocabularies and make sure they're not 'overformalizing' (using OWL for something that can/should be created using SKOS). Then the section could elaborate a bit on using SKOS, as Mark and I had done at [2]. Does this sound reasonable? Thanks for the feedback, Antoine [1] http://w3c.github.io/dwbp/bp.html#dataVocabularies [2] https://www.w3.org/2013/dwbp/wiki/Making_controlled_vocabularies_accessible_as_URI_sets#Intro [3] E.g. 'lighweight' gathers RDFS and SKOS, while the two are meant for very different things, technically, even though we can call an RDFS schema a 'lightweight ontology'.
Received on Friday, 9 January 2015 15:28:32 UTC