Re: Linked Data Aspects of R2RML

* Juan Sequeda <juanfederico@gmail.com> [2010-01-14 18:51-0500]
> Michael and all,
> 
> I have a question about reusable identifiers.
> 
> If I have my movie rental company relational database and I want to expose
> it all as Linked Data. What should be the identifier for "Breakfast at
> Tiffany's?"
> 
> http://dbpedia.org/resource/Breakfast_at_Tiffany%27s_%28film%29
> http://data.linkedmdb.org/resource/film/71
> okkam identifier for Breakfast at Tiffany's (if it exists)
> 
> or should it be a
> 
> http://myrentalstore.com/resource/film/123 owl:samesAs
> http://dbpedia.org/resource/Breakfast_at_Tiffany%27s_%28film%29

While it's fabulous to use URLs for both unambiguity *and*
exploration, the two are in conflict in most publication scenarios.
IF you use dbpedia and *everyone* uses dbpedia, data integration
becomes trivial; people who used to write mashups can now just write
SPARQL queries.

The downside is that you lose the appealing aspect of controlling what
content the user sees, which means, you don't get to pepper it with
links like <rent it> or <if you liked this, you'll also like...>.

The fallback is to have myrentalstore links and some owl:sameAs. The
user sees what you want them to see, and sparql query authors have a
small additional burden of including the sameAs in their mashup query.
(You could have a SPARQL query translator that stuck sameAs in willy
nilly, but the cost of that query is much higher.)

This problem is apparent in most LOD sites which are actually large
warehouses. One possible solution is to have a bit more protocol so
that when you start out using your generic RDF browser on
myrentalstore, you can follow a link to dbpedia and have the browser
do an extra query on the myrentalstore SPARQL server to supplement the
info. Another is to divide data from display:
  <p><a href="http://myrentalstore.com/resource/film/123">http://dbpedia.org/resource/Breakfast_at_Tiffany%27s_%28film%29</a> a <a href="http://myrentalstore.com/resource/Classics">http://dbpedia.org/resource/Film_Classic</a> .</p>

Most of these solutions have some drawbacks; time will tell what we
choose and what would should have chosen.

> Audrey Hepburn, can be considered a well known entity with URIs in dbpedia,
> freebase, etc. We know that Audrey Hepburn acted in Breakfast at Tiffany's.
> So should I have my own URI for the movie and reuse an identifier for Audrey
> Hepburn?
> 
> I agree that we need to offer the possibility in the language to reuse the
> identifiers, I'm just wondering what is the use case.
> 
> 
> 
> Juan Sequeda, Ph.D Student
> Dept. of Computer Sciences
> The University of Texas at Austin
> www.juansequeda.com
> www.semanticwebaustin.org
> 
> 
> On Thu, Jan 14, 2010 at 5:48 AM, Michael Hausenblas <
> michael.hausenblas@deri.org> wrote:
> 
> >
> > All,
> >
> > I've put my initial thoughts re the Linked Data aspects of R2RML on our
> > Wiki
> > [1]. Please read, review & comment (preferably in the Wiki; saves us all
> > time ;).
> >
> > Note that this will be the main discussion point for our upcoming telco on
> > 2010-01-19.
> >
> > Cheers,
> >      Michael
> >
> > [1] http://www.w3.org/2001/sw/rdb2rdf/wiki/LinkedDataAspects
> >
> > --
> > Dr. Michael Hausenblas
> > LiDRC - Linked Data Research Centre
> > DERI - Digital Enterprise Research Institute
> > NUIG - National University of Ireland, Galway
> > Ireland, Europe
> > Tel. +353 91 495730
> > http://linkeddata.deri.ie/
> > http://sw-app.org/about.html
> >
> >
> >
> >

-- 
-ericP

Received on Friday, 15 January 2010 12:47:49 UTC