Re: vCard RDF merge....

[Toby Inkster-4]

On Thu, 2009-09-17 at 10:36 +0200, Martin Hepp (UniBW) wrote:
> I don't know whether I am in the Linked Data scene ;-), but I am
> convinced that bNodes can be useful.

As far as I'm concerned a bNode is just a resource that nobody's
bothered to give a URI... yet.

I don't think that the vCard vocab should be *insisting* that these are
bNodes rather than URIs. e.g. the following should be essentially the
same as far as vCard vocab consumers are concerned:

        <#me> v:tel [ a v:Home ; rdf:value "123456789" ] .

and:

        <#me> v:tel <#homePhone> .
        <#homePhone> a v:Home ; rdf:value "123456789" .

Indeed, it annoys me that [...] in SPARQL doesn't match nodes which have
URIs. It's unintuitive.

I'm CCing public-rdf-dawg-comments.

--
Toby A Inkster
<mailto:mail@...>
<http://tobyinkster.co.uk>

[Seaborne, Andy]

> -----Original Message-----
> From: public-rdf-dawg-comments-request@... [mailto:public-rdf-dawg-
> comments-request@...] On Behalf Of Toby Inkster
> Sent: 17 September 2009 11:23
> To: martin.hepp@...
> Cc: Dan Brickley; Harry Halpin; Renato Iannella; Norman Walsh; Peter Mika;
> Brian Suda; www-archive@...; public-rdf-dawg-comments@...
> Subject: Re: vCard RDF merge....
>
> On Thu, 2009-09-17 at 10:36 +0200, Martin Hepp (UniBW) wrote:
> > I don't know whether I am in the Linked Data scene ;-), but I am
> > convinced that bNodes can be useful.
>
> As far as I'm concerned a bNode is just a resource that nobody's
> bothered to give a URI... yet.
>
> I don't think that the vCard vocab should be *insisting* that these are
> bNodes rather than URIs. e.g. the following should be essentially the
> same as far as vCard vocab consumers are concerned:
>
> <#me> v:tel [ a v:Home ; rdf:value "123456789" ] .
>
> and:
>
> <#me> v:tel <#homePhone> .
> <#homePhone> a v:Home ; rdf:value "123456789" .
>
> Indeed, it annoys me that [...] in SPARQL doesn't match nodes which have
> URIs. It's unintuitive.

Toby - what makes you say that?  

Blank nodes in a SPARQL query are variables and, as such, should match nodes in the graph being matched with URIs or literal or blank nodes in the data.

(Note for your query to match, query and data must have the same base URI for <#..> to match.)

        Andy

>
> I'm CCing public-rdf-dawg-comments.
>
> --
> Toby A Inkster
> <mailto:mail@...>
> <http://tobyinkster.co.uk>
>

[Toby Inkster-4]

On Thu, 2009-09-17 at 11:28 +0000, Seaborne, Andy wrote:
> Blank nodes in a SPARQL query are variables and, as such, should match
> nodes in the graph being matched with URIs or literal or blank nodes
> in the data.

Drat! You are right, of course. And that's a good thing really.

Perhaps it was a bad experience in one particular implementation that
made me think that [...] wouldn't match URI nodes. Or more likely some
other error in my query caused it to fail and I ended up blaming the
square brackets! But certainly ARQ seems to get things right.

PS: Andy, you may be interested in running 'SELECT * WHERE
{ <http://demiblog.org/vocab/oauth#> rdfs:label ?label }'
at <http://demiblog.org/_sparql> and requesting the result in
text/turtle. It uses your result-set vocab.

--
Toby A Inkster
<mailto:mail@...>
<http://tobyinkster.co.uk>

[Peter Mika-2]

Hi All,

I'm following some of the discussion, but not all, unfortunately :(

> <#me> v:tel [ a v:Home ; rdf:value "123456789" ] .
>  

I would only go for this if this was as easy to write in RDFa as

<#me> v:homeTel "123456789" .


But I'm afraid that's not the case.

Peter

[Harry Halpin]

On Thu, Sep 17, 2009 at 6:57 PM, Peter Mika <pmika@...> wrote:

> Hi All,
>
> I'm following some of the discussion, but not all, unfortunately :(
>
>>        <#me> v:tel [ a v:Home ; rdf:value "123456789" ] .
>>
>
> I would only go for this if this was as easy to write in RDFa as
>
> <#me> v:homeTel "123456789" .
>

Well, what I was proposing is that we have a "blank node" style that
lets people type v:tel properties, and a shortcut property for
commonly-used properties like v:homeTel. The advantages of this
proposal is that:

1) It's backwards compatible with vCard 2006
2) But lets people express (and so round-trip) with actually existing
vCards that are more complicated, as earlier VCard let us do.

I thought we had agreement on this. While we can't make it logically
equal using RDF semantics, we can make the namespace document explain
very clearly's what's going on.

Also, +1 on not using URIs for telephone numbers, but just plain literals.

>
> But I'm afraid that's not the case.
>
> Peter
>

[Peter Mika-2]

Hi Harry,

Agreed and agreed!

Sorry for not having followed the discussion.

Peter

Harry Halpin wrote:

> On Thu, Sep 17, 2009 at 6:57 PM, Peter Mika <pmika@...> wrote:
>  
>> Hi All,
>>
>> I'm following some of the discussion, but not all, unfortunately :(
>>
>>    
>>>        <#me> v:tel [ a v:Home ; rdf:value "123456789" ] .
>>>
>>>      
>> I would only go for this if this was as easy to write in RDFa as
>>
>> <#me> v:homeTel "123456789" .
>>
>>    
>
> Well, what I was proposing is that we have a "blank node" style that
> lets people type v:tel properties, and a shortcut property for
> commonly-used properties like v:homeTel. The advantages of this
> proposal is that:
>
> 1) It's backwards compatible with vCard 2006
> 2) But lets people express (and so round-trip) with actually existing
> vCards that are more complicated, as earlier VCard let us do.
>
> I thought we had agreement on this. While we can't make it logically
> equal using RDF semantics, we can make the namespace document explain
> very clearly's what's going on.
>
> Also, +1 on not using URIs for telephone numbers, but just plain literals.
>
>  
>> But I'm afraid that's not the case.
>>
>> Peter
>>
>>