Re: Article Proposal

On Thu, Dec 12, 2013 at 6:58 PM, Niklas Lindström <lindstream@gmail.com> wrote:
> Hello again,
>
> The microdata is now fixed in the following ways:
>
> * corrected some itemtype values, which microdata requires to be full URLs
> (unlike RDFa, which can use @vocab to avoid repetition)
> * the items are linked together using isPartOf
> * the same entities are described throughout (instead of six disjoint
> entities), using @itemid
>
> I used @itemid this time, since at least microdata parsers producing RDF get
> the data right. Unfortunately, @itemid is (also) required to be a full URL
> in microdata, and it is only allowed if both itemscope and itemtype are also
> present. It's either that or using @itemref, which as I showed earlier [1]
> is also somewhat cumbersome (it requires you to sprinkle in @id and glue
> items together from disparate parts). Though if anyone more versed in
> microdata can clean it up, please do.
>
> I also added an RDFa version (which I find to be less verbose). I really
> recommend to paste that into RDFa Play [2].
>
> The examples are verified (using RDFLib) to produce the also added Turtle
> example (minus some web page related details).
>
> (Apart from considering the weight of the markup (which gets heavy with this
> much granularity in once place), the Turtle is what I usually focus on when
> I reason about the merits and flaws of various properties, types and uses
> thereof.)

Awesome, thanks for fixing this up, Niklas! I was enjoying a visit
with Santa at our local public library. Well, my kids enjoyed it too
:)

> I also added a variant with less verbose precision (but using the same
> properties of course): just an Article linked to a PeriodicalIssue (skipping
> the volume and periodical). Notice that name, volumeNumber and issn is used
> on the PeriodicalIssue, indicating that those are, scruffily, "inherited"
> from the collections above. That's the kind of flexibility I believe we're
> after.

Hmm. I think that might be _too_ scruffy; that example hangs the issn,
volumeNumber, and periodical name off of PublicationIssue, which is
not valid according to the proposal that we're putting forward, and
therefore wouldn't be expected to be parsed correctly by the search
engines, right? (I took a quick stab at sorting it out but then
realized that the result was going to pretty much mirror the core
example...).

Thanks,
Dan

Received on Friday, 13 December 2013 04:20:46 UTC