W3C home > Mailing lists > Public > public-vocabs@w3.org > July 2012

Re: Vocabularies for Technical Publishing

From: Joshua Wulf <jwulf@redhat.com>
Date: Fri, 13 Jul 2012 09:52:35 -0400 (EDT)
To: Kenley Lamaute <kenleyl@microsoft.com>
Cc: public-vocabs@w3.org
Message-ID: <e3187dda-6eee-46f0-bc03-c02cb3b50123@zmail16.collab.prod.int.phx2.redhat.com>

----- Original Message -----
> From: "Kenley Lamaute" <kenleyl@microsoft.com>
> To: public-vocabs@w3.org
> Sent: Friday, July 13, 2012 6:05:44 PM
> Subject: Re: Vocabularies for Technical Publishing
> 2. I see the benefit of using a URI reference for currentModel so
> that statically built pages could benefit from it. I understand this
> as being a pointer to the value rather than emitting the actual
> value. It seems that this puts the heavy lifting on search engines
> to follow the pointer and read-in the value.
> I see many examples with URL data type; however, I haven’t found a
> practical example of how an URI reference is currently being used in
> this specific way… Please reply with an active example if you know
> of one.
> There is strong merit in having the option to specify currentModel.
> The value would be fairly straight forward to maintain for
> dynamically built content (php, aspx, etc.). If using a URI
> reference is a viable alternative and there are practical examples
> in use today, then I’ll update the proposal with an included
> example.

Please let me know if I'm misunderstanding this: the currentModel attribute acts as a flag that indicates "This content is/is not relevant to the latest version of a product".

Its likely use is that when a user requests information about a product, but does not specify a version, the search engine would compare the value of search result pages' "Product model" to their "Product currentModel", and favour results where the model and currentModel are the same. This would favour the most recent material.

So currentModel would need to, ultimately, be type-compatible with model: a string literal.

I don't know if there is an existing schema.org pattern to deal with this, but something like the xi:include pattern of xml seems like a good fit. A URI is given, and the content returned by dereferencing that URI is accepted as the value. So in this case, you request the URI, you get back a string, and you take the string as the value.

I would say that an approach such as this - the option to specify the metadata inline or as a reference that is read from somewhere else - is an appropriate optional pattern where a page includes metadata about something other than itself or its content; and it's something that affects more than just the Technical Article schema.

Also, as you indicated, such an approach is useful in practice only if search engines dereference the URI to get the value.

Is there any traction for an xi:include like mechanism for metadata injection via URI reference* in schema.org schemas? Or is it better to say: "Don't populate this value if you can't guarantee that you'll update it"? Because the value of the data decreases when it's unreliable.

- Josh

* Which is essentially the same pattern as implementing it in javascript when server-side dynamic assembly is not available. 

> -Kenley
> Joshua Wulf < jwulf@redhat.com >
> Date : Fri, 22 Jun 2012 18:24:53 -0400 (EDT)
> To : public-vocabs@w3.org
> Message-ID :
> <b3c509c0-1d16-4eae-8b6a-e9ffe67c34ac@zmail16.collab.prod.int.phx2.redhat.com>
> Great initiative, and I'm keenly interested in its development. I
> have a couple of questions. 1. Can you put multiple product versions
> inside the same declaration? The same procedure might be valid for
> several different iterations of a product (and not valid for other
> specific iterations, obviously). So, for example: <div
> itemprop="about" itemscope itemtype=" http://schema.org/Product ">
> <p> <strong>Applies to:</strong> <span itemprop="name">Microsoft SQL
> Server 2008 R2</span> <span itemprop="name">Microsoft SQL Server
> 2008 R3</span> <span itemprop="name">Microsoft SQL Server 2008
> R4</span> </p> <meta itemprop="model" content="2008 R2"/> <meta
> itemprop="model" content="2008 R3"/> <meta itemprop="model"
> content="2008 R4"/> <meta itemprop="currentModel" content="2012"/>
> </div> 2. Is it possible to specify the currentModel as a URI
> reference? So then you only have to update it in one place and
> voila, all your htmlz are updated - without having to republish
> everything or implement dynamic injection. It seems like the other
> metadata is data about the page or entities described on or involved
> in production of the page; but the currentModel item is metadata
> about another entity - albeit related - which doesn't necessarily
> exist when the page is produced, /and/ changes over time. - Josh
Received on Friday, 13 July 2012 13:53:06 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:48:47 UTC