Re: [docbook] RDFa Lite inside <info>

[Thomas Schraitle <tom_schr@web.de>]

Hi Norman,

Am 02.07.21 um 11:00 schrieb Norm Tovey-Walsh:
> [...]
>
> Iâd be inclined to use DocBook markup for the
> specific properties youâre interested in:
>
>   <article audience="â">
>     <info>
>       <keywordset>
>         <keyword>â</keyword>
>       </keywordset>
>       <!-- or subjectset if you have a controlled vocabulary -->
>       <abstract>
>         <para>aboutâ</para>
>       </abstract>
>     </info>

Yes, I'm aware and was about to mention that in my original text. Just
tried to simplify my text. :)

Sure, in that case you can use the <keyword> element or use the audience
attribute. However, for my taste, it creates two distinct sets of
metadata: one from DocBook and the other from the vocabulary (in my
case, schema.org).


> I imagine youâll want to run some sort of extraction
> process to get metadata out of your documents (to put in
> a linked data store, for example), and thereâs no reason
> not to map from DocBook metadata to properties.

Yes, the information should appear in the HTML source. If I'm not
mistaken, RDFa Lite is currently not supported for the XSLT 1.0
stylehsheets. This would need a customization layer (which I already have).


>> However, that feels like abusing the <annotation> element.
>
> Yes, it does.

:-)


>> Maybe DocBook needs a general <meta> element soley for
>> information[...]
>
> DocBook 5.2 introduces such an element which can either
> contain content or be empty:
>
>   https://tdg.docbook.org/tdg/5.2/meta.content.html
>   https://tdg.docbook.org/tdg/5.2/meta.empty.html

Ahh, I haven't looked into the latest DocBook version. Cool, thanks!


> Itâs perhaps geared a little too strongly towards the HTML
> style of meta element. It could probably be relaxed a
> little bit, so that name was optional for example, to make
 it fit this use case a little better.

If I read the TDG correctly, "name" is a required attribute, right?
Wouldn't that make this attribute somehow unnecessary in the light of
RDFa Lite?

I was thinking about my example:

 <info>
   <meta vocab="http://schema.org/"; typeof="Guide">
     <meta property="audience">...</phrase>
     <meta property="keywords">...</phrase>
     <meta property="about">...</phrase>
   </meta>
 </info>

This would mean, I'm forced to use "name":

 <info>
   <meta vocab="http://schema.org/"; typeof="Guide">
     <meta property="audience" name="audience">...</phrase>
   </meta>
 </info>

Is that correct? That would mean double information which is always a
bad sign.

Not sure if the content model should be adapted. Something like a
"HTML-like" <meta> element with name and a "RDFa Lite"-like <meta> with
all other RDFa attributes. Would that make sense?

Thanks for all the insights. Much appreciated! :)


--
GruÃ/Regards
  Thomas Schraitle