- From: Jeni Tennison <jeni@jenitennison.com>
- Date: Wed, 26 Oct 2011 13:15:13 +0100
- To: HTML Data Task Force WG <public-html-data-tf@w3.org>
- Cc: Toby Inkster <tai@g5n.co.uk>, Gregg Kellogg <gregg@kellogg-assoc.com>
On 26 Oct 2011, at 12:53, Toby Inkster wrote: > A few comments on > <https://dvcs.w3.org/hg/htmldata/raw-file/24af1cde0da1/microdata-rdf/index.html>... > > 1. Creating RDF collections for repeated properties is an awful idea. > Consider the example in appendix A. The object of the frbr:realization > is a blank node of type rdf:List. According to the FRBR Core vocab, the > range of frbr:realization is frbr:Expression. If we assume that > frbr:Expression and rdf:List are disjoint classes (which seems to be a > reasonable assumption), you've created a logical contradiction. So the > graph generated by parsing the microdata does not match up to the > expectations of the vocabulary, and probably does not match up to the > expectations of the page author. > > I can understand the desire to preserve document order to cover > certain use cases, but it's possible to do that outside of the RDF > model. (e.g. Rather than parse the page as a whole and operate on the > graph returned, you could supply a callback function to the parser to > be called as each triple is extracted from the page. The callback > function would then receive triples in document order.) > > TLDR: generating collections breaks ranges. Just thinking around this problem, might one way to square this be to generate separate graphs: one containing the separate values in multiple triples and another generating the values as lists? Applications could then look into the "ordered" graph only when/if they needed that information. Jeni -- Jeni Tennison http://www.jenitennison.com
Received on Wednesday, 26 October 2011 12:15:52 UTC