Re: Indicating main entity / primaryTopic - proposal to use 'schema.org/about'

>
> What I think we want is a property that performs the same role as FOAF's
> 'primaryTopic': it should point to at most one entity/thing. Given
> currently popular terminology we might call it 'mainEntity' as a
> strawman.


Couldn't changing the expected value of @mainContantOfPage to Thing work
for this?

Doing so would actually help a lot of websites. I've lost count how many
times I've encountered:
<div itemprop="mainContentOfPage" itemscope itemtype="
http://schema.org/Product"> (or Article or Blog).

And by expanding the domain of @mainContentOfPage all those websites would
automagically have valid markup.


2014-05-21 21:22 GMT+02:00 Dan Brickley <danbri@google.com>:

> On 21 May 2014 19:21, Dan Scott <dan@coffeecode.net> wrote:
> > On Wed, May 21, 2014 at 02:04:20PM +0200, Jarno van Driel wrote:
> >>
> >> I was wondering, can an entity also have multiple @about properties?
>
> That's the right question to be asking. And I didn't ask it hard
> enough yesterday (probably because I wouldn't have liked the answer).
>
> The wording http://schema.org/about has currently, "The subject matter
> of the content." is awkward. The word "the" suggests a single thing is
> the subject matter, but it is vague enough that you could have several
> entities via repeated properties together capturing "the subject
> matter".
>
> What I think we want is a property that performs the same role as
> FOAF's 'primaryTopic': it should point to at most one entity/thing.
> Given currently popular terminology we might call it 'mainEntity' as a
> strawman.
>
> I was hoping we could get away with refining the interpretation of
> 'about', but I'm coming around to the view that it has been used in
> too many diverse ways over the last 3 years for that to work.
>
> >> I ask because when chaining multiple entities to a WebPageElement, to me
> >> it
> >> seems the following is the logical thing to do:
> >>
> >> <body itemscope itemtype="http://schema.org/WebPage">
> >>    ...
> >>    <div itemprop="hasPart" itemscope
> >> itemtype="http://schema.org/WPSideBar">
> >>        <div itemprop="about" itemscope
> >> itemtype="http://schema.org/ContactPoint">...</div>
> >>        <div itemprop="about" itemscope
> >> itemtype="http://schema.org/ItemList">...</div>
> >>    </div>
> >>    ...
> >> </body>
> >>
> >> Or would @hasPart or @mentions be prefered over @about?
>
> I don't think they're great examples of about-ness, except
> ContactPoint, if the page is indeed about contact info. The
> stereotypical use for 'about' is a specific person-place-or-thing that
> the content is 'about'. Sidebars and lists are structural mechanisms;
> it would be more typical to see Product, Book, Person, Place etc used.
> However your main point, that 'about' could credibly be repeated given
> its definition, is quite reasonable.
>
> >
> > I'm not going to offer any advice about whether "hasPart" or "mentions"
> > might be preferred here, but you can certainly have multiple "about"
> > properties for a single entity.
>
> Yeah. It is tempting to defend a strict reading of the word 'the' and
> claim it shouldn't _really_ be repeated; but I don't think that's
> credible.
>
> > See the example for http://schema.org/MedicalScholarlyArticle - "about"
> > is used twice, because the article is about a type of drug and
> > about a type of medical condition.
>
> quite :)
>
>
> > The cardinality of schema.org properties appears to be a FAQ dating back
> > to at least 2011 (http://www.w3.org/2011/webschema/track/issues/5); we
> > should probably add an explicit statement to
> > http://schema.org/docs/gs.html or http://schema.org/docs/faq.html (or
> > both) saying that you can, in general, repeat properties in schema.org
> > entities as necessary.
>
> There are a few (e.g. birthDate, deathDate, most boolean-valued
> properties) that have at most one sensible value. However even those
> might have several reasonable encodings. And there are some, e.g.
> iataCode hopefully, for which there should be at most one entity that
> has any given value for that property. However we've not attempted
> cataloguing these cases, partly through a concern to avoid making
> unrealistically brittle and rigid rules that will be ignored...
>
> cheers,
>
> Dan
>

Received on Wednesday, 21 May 2014 21:00:21 UTC