- From: Sebastian Heath <sebastian.heath@gmail.com>
- Date: Sun, 20 Nov 2011 09:07:25 -0500
- To: RDFa WG <public-rdfa-wg@w3.org>
Thanks to Toby and Ivan for their responses. Yes, it's a feature request, no doubt about that. I'll take two approaches to further advocating the feature andresponding to the comments so far. 1) I think it's important that rdf/xml supports a similar mechanismwith its rdf:ID mechanism [1][2]. The full discussion is in thatdocument so I quote only this from [1]: "rdf:ID provides an additionalcheck since the same name can only appear once in the scope of anxml:base value (or document, if none is given), so is useful fordefining a set of distinct, related terms relative to the same RDF URIreference." That would apply to the @id/@typeof combo in RDFa. And it's relevant that rdf:ID is widely deployed in the wild. Inparticular, it's used in the skos and owl communities. 2) I reference skos and owl, in part, to move our consideration of usecases beyond foaf based examples. I mean it as a serious comment, andnot a bit of snark, when I say that the semantic web is more than justa rolodex. The semantic web instantiates lots of resources - usingthat term in its rdf sense - that are concepts, not physical thingslike people. It's here that the @id/@typeof combdo will beparticularly useful. A potential use case: list of geographic co-ordinates in latitude/longitude. A geographic co-ordinate system is not a physical thing. And norlocations specified in a co-ordinate system. Sure, you can use thosepositions to get to physical place but there's a difference. So to markup one entry in a list using @id/@typeof is simple based onthe w3 geo vocab and [3] some other well-known vocabs: <div id="pos1" typeof="geo:Point"> <span property="dc:title">stop 1</span> <span property="geo:lat">55.701</span> <span property="geo:long">12.552</span> <span property="rdfs:label">A nice place to visit</span> <img rel="foaf:depiction" src="#image1"/> </div> I think it's trivial to see which triples this would produce if the@id/@type syntax is supported. It should be clear that markup such as the above is the originalinstantiation of the concept being both identified and described. Andis done in such a way that it is visible to both dom-aware andrdfa-processor aware. I phrase it that way because for my needs, thatis an answer to Toby's concern about "what is being identified" and toboth Ivan's and Toby's concern about how this overlaps withTAG/Semantic Web issues. Again, it is OK for RDFa to be the originalinstantiation of a concept and for RDFa markup to support thatinstantiation as a full node on the Semantic Web/Cloud of Linked Data(in my case with a preference for Linked Open Data). And I think that the @id/@typeof syntax would be a simplificationover the current syntax and so meets Ivan's concerns about richness. Perhaps personalizing too much, I have lots of use cases for thisconstruct. I'm not just pushing it in abstract. Another one is thepresence of bibliographic citations in born-digital scholarship. Theassociation between an article and a source is usefully implemented asan addressable node on the semantic web. This can be done by enforcingequality between @id and @about but that's fragile in comparison tothe robust and clear semantics of the proposed @id/@typeof construct. -Sebastian
Received on Sunday, 20 November 2011 14:08:02 UTC