http://www.w3.org/TR/2012/REC-r2rml-20120927/
http://www.w3.org/TR/2012/REC-rdb-direct-mapping-20120927/
http://semanticweb.com/transforming-relational-data-to-rdf-r2rml-becomes-official-w3c-recommendation_b32395
Thank you very much, everyone involved! A big kudos to the wonderful Editors of
Hi,
How do you include links to results of computations in Linked Data?
For instance, you publish data about entities of a given class. A property,
let's call it :expensiveProp, has this class as domain, and you know that
computing or publishing the corresponding triples is expensive. In such
Hi Francois-Paul,
how about that solution:
You publish the cheap data about your entity under
http://example.org/e0, which is the official URI of that entity:
ex:e0 owl:sameAs ex:e0expensive
ex:e0 :cheapProp ...
And under http://example.org/ex:e0expensive, you publish
ex:e0expensive
Thanks,
no, this doesn't solve the problem. A user gets ex:e0 (the cheap resource).
Though she can see that there is the link to the expensive resource, she
doesn't know the meaning of the link (it is just an owl:sameAs): she doesn't
know what this is about. (Note also that there could be
Why introduce a non-canonical NIR identifier and yet another sameAs link?
There's no need to include a complete description when you resolve the
identifier, a seeAlso IR link would suffice (wouldn't it?)
Barry
- Reply message -
From: Heiko Paulheim paulh...@ke.tu-darmstadt.de
Date:
F-P, there's a bunch of work on describing expected computed triples in a
Linked Data context; see Linked Open Services, Linked Data Services, RESTdesc,
etc.
Barry
- Reply message -
From: SERVANT Francois-Paul francois-paul.serv...@renault.com
Date: Fri, Sep 28, 2012 15:54
Subject:
Short answer is no,
linked data standards have never addressed this and many other even
basic problems(e.g. what if there are too many properties of one kind,
what kind of level of description you're supposed to get (e.g.
recourse on blank nodes?), what is a standard way to find the entry
URI
OK, I see.
Then a possible idea could be splitting the dataset into a cheap dataset
and one (or more) expensive datasets, interlinked by sameAs links. If
you include a VoID description [1] of all datasets with each of the
datasets, you can inform the user about what properties to expect when
You can have multiple documents about the same thing. Each document
can contain a different (part of the whole) description, and can link
to the other related documents.
GET /someresource
foaf:primaryTopic #thing .
#thing foaf:isPrimaryTopicOf /expensiveDescription , .
You can then provide
SERVANT Francois-Paul wrote:
Hi,
How do you include links to results of computations in Linked Data?
For instance, you publish data about entities of a given class. A property,
let's call it :expensiveProp, has this class as domain, and you know that
computing or publishing the corresponding
On 9/28/12 11:12 AM, Giovanni Tummarello wrote:
Short answer is no,
linked data standards have never addressed this and many other even
basic problems(e.g. what if there are too many properties of one kind,
what kind of level of description you're supposed to get (e.g.
recourse on blank
Hi,
may I say that the situation you describe is a bit disappointing? The
unaddressed issues that you mention had already been raised shortly after the
publishing of the linked data principles, years ago. I find it is a pity if
they remain unanswered, because this can jeopardize one of the
It's worth pointing out that there IS finally a W3C working group looking at
these issues:
http://www.w3.org/2012/ldp/charter.html
Barry
- Reply message -
From: SERVANT Francois-Paul francois-paul.serv...@renault.com
Date: Fri, Sep 28, 2012 17:54
Subject: Expensive links in Linked Data
On 9/28/12 1:02 PM, Barry Norton wrote:
It's worth pointing out that there IS finally a W3C working group looking at
these issues:
http://www.w3.org/2012/ldp/charter.html
I don't know about this group address the matter of scrollable cursors
based on partial query results as enablers of
Other parts of the charter seem more directly relevant to this thread
(since they're talking about side-effect, though the REST folks also too
often also loses focus on actual computation, rather than just moving
representations around, in my opinion) - Francois-Paul may disagree...
Barry
15 matches
Mail list logo