Re: Multiple itemtypes in microdata

On Tue, 18 Oct 2011, Jeni Tennison wrote:
> >
> > Surely nobody is seriously expecting random users to do searches 
> > containing actual property URLs. That's a highly advanced feature for 
> > a tiny subset of users, but it's not a real-world application.
> 
> In Sindice, use just the term 'Document' or 'News' for the class and it 
> will select classes whose URIs/labels (I'm not sure which) contain that 
> term. An alternative interface would be one like Wolfram Alpha's, where 
> labels for the classes of matches are listed for a user to choose from.

I'm not 100% sure what that means. Wouldn't this mean a viagra advert 
could just claim to be "News" and then come up in all the news searches? 
I'm trying to understand the use case here.


> > What I meant is what software consumes the non-schema.org vocabularies 
> > specifically?
> 
> Google and Yahoo! consume the rdf.data-vocabulary.org vocabulary, 
> GoodRelations and at least some parts of FOAF.

Sure, but you never need to specify that a single item uses types from two 
or more of those vocabularies. The cases where you need to give multiple 
types are when all the items have the same vocabulary (e.g. a lawyer who 
is also a teacher, or a museum that is also a shop, or whatnot).


> On the client side, as well as generic plug-ins in which vocabulary 
> isn't relevant, there are a few GoodRelations plugins for Chrome and 
> Firefox, for example, get hold of the GoodRelations Amazon Checker for 
> Chrome [1] and go to a BestBuy listing page such as [2] and you should 
> see a pop-up appear in the bottom left with links to Amazon pages for 
> those products.
> [1] https://chrome.google.com/webstore/detail/jlfealjceojnflopgcelhcmghpakmklk
> [2] http://www.bestbuy.com/site/searchpage.jsp?id=pcat17071&st=3109302%2C+3109339%2C+3109357&sc=Global&usc=All+Categories

That seems like a rather pointlessly complicated way of implementing such 
a feature. Why not just allow the user to select some text and search for 
it on Amazon?


> As far as I can tell, that's the current state of working consumers of 
> the non-schema.org vocabularies used in the use cases I listed where 
> multiple types were given for single items.

These use cases aren't very compelling. It's hard for me to understand why 
we would need to mark up a single item as being two incompatible types 
when there's not really a compelling case for there even being one type.

Looking specifically at the ones you listed:

On Sun, 16 Oct 2011, Jeni Tennison wrote:
> 
> = examiner.com =
> 
> Example: http://www.examiner.com/fall-colors-in-national
> 
> Each article on the site is marked up as being:
>   * http://schema.org/NewsArticle 
>   * http://rdfs.org/sioc/ns#Item 
>   * http://xmlns.com/foaf/0.1/Document
> 
> In this case, the three types are progressively more specific 
> descriptions of the article.

I understand why they'd be marked as http://schema.org/NewsArticle; it 
translates to prettier results on search engines. But what's the use case 
for the other two types? What software is going to use their semantics? 
(Possibly more to the point, what software would make use of their 
semantics once they have been converted to microdata vocabularies 
incompatible with schema.org's?)


> = sears.com =
> 
> Example: http://www.sears.com/shc/s/p_10153_12605_04450094000P
> 
> Each item for sale is marked up as being:
>   * http://purl.org/goodrelations/v1#SomeItems 
>   * a type from the http://www.productontology.org/id/ vocabulary, in 
>     the example http://www.productontology.org/id/Earrings
> 
> Again, the product ontology type is more specific than the GoodRelations 
> type.

What problem is this solving for them? What software is using this data?

Do Earrings have different properties than other items, in such a way that 
the items actually need two vocabularies? (e.g. do "Earrings" have an 
"ear" property with values "left" or "right" while a "Cereal" has an "ear" 
property with value "corn", or something?)


> = digifotopro.nl =
> 
> Example: http://www.digifotopro.nl/
> 
> The footer on each page describes some organisations, each of which is marked as being:
>   * http://xmlns.com/foaf/0.1/Organization
>   * http://rdf.data-vocabulary.org/#Organization
> 
> Here, there are two vocabularies with types that mean exactly the same 
> thing, but neither has clear universal adoption and the publisher is 
> hedging their bets by labelling each organisation with both types. I 
> imagine they'll add http://schema.org/Organization as well when they get 
> round to updating their site.

Why would they mark up the organisations as either type? I'm really 
failing to understand what the goal is here. What software do they expect 
to do anything with this? (Presumably not a search engine; I don't see 
what a search engine could do with the information in the footer. The page 
isn't about those organisations, so it wouldn't make sense to display it 
in the results or anything like that.)


On Tue, 18 Oct 2011, Jeni Tennison wrote:
>
> Are you also interested in vocabularies and consumers where the 
> vocabularies overlap with schema.org and there are web pages using the 
> vocabularies, but the web pages don't currently contain markup using 
> more than one type?

I'm interested in anything where a real user would actually be negatively 
impacted because data on a particular real page can't be annotated as 
using two different incompatible vocabularies, but where one user uses 
software that only supports one vocabulary and another uses software that 
only uses another vocabulary and it's unrealistic to expect the developers 
of those software products to have been unable to collaborate and develop 
a common vocabulary.

In other words, I'm interested in when the lack of syntactic support for a 
single item having properties from two vocabularies is a real problem.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Tuesday, 18 October 2011 06:33:50 UTC