- From: Niklas Lindström <lindstream@gmail.com>
- Date: Thu, 14 Feb 2013 23:59:28 +0100
- To: Alf Eaton <eaton.alf@gmail.com>
- Cc: Ed Summers <ehs@pobox.com>, "public-schemabibex@w3.org" <public-schemabibex@w3.org>
Hi Alf, On Thu, Feb 14, 2013 at 1:58 PM, Alf Eaton <eaton.alf@gmail.com> wrote: > Yes, the conclusion seems to me to be that if there was a > "WrittenWork" type, then that might be the most appropriate place for > a "citation" property in its strictest sense. As there isn't, > CreativeWork seems like the right place at this time. I agree. > Who should now forward this proposal to the public-vocabs list and the > WebSchemas wiki? > > For an example of "citation" in action, see > https://peerj.com/articles/19/ and the extracted data at > http://www.google.com/webmasters/tools/richsnippets?url=https%3A%2F%2Fpeerj.com%2Farticles%2F19%2F > (it looks like the itemscope/itemref properties might not be quite > right on the citations still, as the Structured Data Testing Tool > doesn't seem to be showing the properties of the itemref target for > the object where it's referred to). The behaviour of @itemref can be quite surprising. It doesn't actually represent a reference (cf. a link), but uses the referenced "data block" to copy embedded microdata into the current result. That's why the JSON output in the wiki shows five items where I suspect you only want three (the article and it's two citations)? If using RDFa I'd recommend to just link to the cited resource instead (using its URI or a document-local identifier if the URI of the thing is unknown), and then describe it in detail if you want to. The corresponding attribute in microdata is @itemid, but I'm not certain that it works exactly the same, since microdata in itself is only a tree of data, where such linkage doesn't exist. When interpreted as RDF it would though (AFAIK). If I may use Turtle, I expect that what you're after is: @prefix : <http://schema.org/> . [] a :ScholarlyArticle; :citation [ a :Book; :author [ a :Person; :familyName "Langer"; :givenName "MC" ]; :datePublished "2004"; :name "The Dinosauria"; :url <http://books.google.co.uk/books?vid=ISBN9780520242098> ], [ a :ScholarlyArticle; :author [ a :Person; :familyName "Sereno"; :givenName "PD" ]; :datePublished "1991"; :name "Basal archosaurs: phylogenetic relationships and functional implications"; :url <http://dx.doi.org/10.2307/3889336> ] . (It may or may not be relevant to the topic at hand to also discuss whether these unidentified nodes should be identified using URIs instead; to formally represent linked data.) .. As for the rest of the topic, I agree with Ed that an iterative approach is likely to be the most effective. Of course, proceed with caution even so (to avoid gratuitous extensions that might end up deviating from other perspectives, say BibFrame, without good reason). Modeling the information as it appears on the web surface isn't always easy, since there is a lot of repetition going on there which can cloud simple designs (and hence be hard to consume and build services upon). So I'd recommend iteration with diligence. :) But indeed, it is reasonably more effective to submit proposals piece by piece as they crystallize. Cheers, Niklas
Received on Thursday, 14 February 2013 23:00:30 UTC