Re: It is bad practice to consume Linked Data and not publish Linked Data

Excellent, Sergio, thanks for commenting.

On 4 Apr 2013, at 07:26, Sergio Fernández <sergio.fernandez@salzburgresearch.at> wrote:

> On 03/04/13 23:32, Hugh Glaser wrote:
>> Because it is. :-)
> 
> Because it is not ;-)
> 
> The problem is when we restrict ourselves to this little world of self-consumption. But it we think in broader terms, let's say REST APIs on any service, what's the ratio of providers vs. consumers?
I am definitely not thinking of a self-consuming world.
I am thinking of where I help others to consume my outputs by making them available in other (structured) forms, such as RDF and especially JSON.
If we prescribe a cosy little world of Linked Data exchange, with a circumference of sites that just consume it and publish web pages, it really isn't very exciting (and won't ever be very big).
> 
>> Along with Kingsley's Crime #2 Against Linked Data, I think this is Crime #1 Against Linked Data.
> 
> So, sorry, I don't see where is the crime…
Typical situation from a published file (that happened):
uri:building-one uri:has-architect "Hugh Casson" .
So all I can really do with this is print the name "Hugh Casson" on a page about the building.
If there is also
uri:building-two uri:has-architect "Hugh Casson" .
I can only print the "Hugh Casson" twice - there is no connection between the actual architects.
(A/The whole point of Semantic Web technologies is that we don't do risky things like assuming similar string resources identify the same implied real-world things.)
Even if I have worked out that they are the same "Hugh Casson" agent, I have no simple way of representing or stating this.
And of course if there is another triple:
uri:building-three uri:has-architect "Hugh Casson Partners" .
I am completely stuffed in terms of adding any value to the knowledge (other than asserting loads of triples to say what I want).

Of course, the data was fine for the publisher - all they wanted to do was annotate the text in an html page with the name of the architect.
Since they were new into Linked Data, it took me a little while to get across that other people might want to do other things (such as list all the buildings by architect).

Strangely enough, I ran into this on http://data.semanticweb.org a couple of days ago.
The keywords for papers are:
At the moment, pages such as
http://data.semanticweb.org/conference/iswc/2012/paper/inuse-51
have things like
<dc:subject>Provenance</dc:subject>
<dc:subject>Linked Data</dc:subject>
and also
<dcterms:subject>Linked Data</dcterms:subject>

This means that I can't easily do linkage.
I'm actually trying to find out if anyone currently loves the site and can fix it - does anyone know? :-)

Cheers
> 
> Cheers,
> 
> -- 
> Sergio Fernández
> Salzburg Research
> +43 662 2288 318
> Jakob-Haringer Strasse 5/II
> A-5020 Salzburg (Austria)
> http://www.salzburgresearch.at
> 

Received on Thursday, 4 April 2013 11:12:54 UTC