Re: Question/feedback re: oa:hasBody and cnt:ContentAsText in RDFa

On 2015-07-25 21:25, Robert Sanderson wrote:
> Hi Sarven,
> One of the considerations here was that you can't have both language and
> datatype associated with the same literal, so we had to pick two
> predicates for recording the information and mint a resource for the
> subject.  We went with Content in RDF as it seemed at the time to have
> some legs and fulfilled our requirements.  In the Annotation working
> group, we've minted our own class, as it's clear that CNT is abandoned
> and will never reach recommendation status.

I may be mixing up a few things here and hope that you can clarify.

Can you please clarify what those two predicates you are referring to? 
I'm only aware of oa:hasBody. Unless you meant something completely 

oa:hasBody doesn't have a range, and talks about / 
examplifies it being purposed as an ObjectProperty. Did I understand 
this correctly, or is it possible to have oa:hasBody as a 
DatatypeProperty? Using the example from my first email, would this be okay:

<div property="oa:hasBody">

> We went with dc rather than dcterms because of the ranges.
>   dcterms:format has a range of a resource of class
> dcterms;MediaTypeOrExtent.  We thought this was overkill for the 95% use
> case of just recording the media type of the content, and dc elements
> has no such requirement.  Ditto dcterms:language and
> dcterms:LinguisticSystem, versus dc:language.
> While the information in RDFA is not very interesting to a human reader
> of the page, RDFA is not intended for humans :)  Once that annotation
> has been pulled out of the page and made available separately as its own
> resource, that information is very important so clients know how to
> render it to the user.

Aside: IMO, a good HTML+RDFa markup practice is to minimize, if not 
eliminate the differences in information that's given to both humans and 
machines. If machine-processing is the sole concern, there are other RDF 
formats, e.g., N-Triples, which are more suitable for the job. Hence, if 
one is committing to HTML+RDFa, it is arguably preferable to cut down on 
hidden markup (i.e., acting as secondary metadata) which is only 
"visible" to the machines. In a way, RDFa is closely tied to 
human-friendly documents.

> Hope that helps!
> Rob

Thanks a lot for your feedback!

> On Sat, Jul 25, 2015 at 9:16 AM, Sarven Capadisli <
> <>> wrote:
>     Hi all,
> suggests to
>     use dc:format. I was wondering whether it'd be okay to use
>     dcterms:format instead? My reasoning is simply that I predominantly
>     use dcterms, and don't wish to introduce dc (elements). Tough luck?
>     I find cnt to be (unnecessarily?) cumbersome in RDFa when used with
>     oa:hasBody and cnt:ContentAsText. Use case: representing comments on
>     a webpage which include HTML. This forces me to add markup that's
>     only for machine consumption i.e., <span property="dcterms:format"
>     content="text/html"></span></div>, and possibly even <span
>     property="cnt:characterEncoding" content="utf-8"></span>, <span
>     property="dcterms:language" content="en"></span>. I think we can
>     agree that the above information is not very "interesting" to a
>     human-reader on a webpage.
>     Leaving dcterms:format would mean that the consumer has to further
>     process - I'm okay with this if there is no sensible alternative.
>     Are there any workarounds people employing? How does the following look:
>     <div rel="oa:hasBody">
>        <div about=""
>     property="dcterms:description" datatype="rdf:HTML">
>          <p>foo</p>
>        </div>
>     </div>
>     [Note that the rdf:HTML datatype is left as non-normative in RDF
>     1.1: ]
>     Thoughts?
>     Thanks,
>     -Sarven
> --
> Rob Sanderson
> Information Standards Advocate
> Digital Library Systems and Services
> Stanford, CA 94305


Received on Monday, 3 August 2015 14:05:20 UTC