Ethan, it is true that probably a majority of RDF "sets" in the cloud
are exports from a non-RDF format. Yet if you look at the page I cited,
you will see that there are major players (including Google) working
with triple-stores and doing validation on them using SPARQL. So
validation of RDF has use cases, and those use cases appear to be
growing as more users move to native or near-native RDF.
Europeana uses XSD/schematron in their implementation, but apparently
would prefer a better solution. (See talk by Antoine Isaac, Day 1).
kc
On 9/12/13 8:34 AM, Ethan Gruber wrote:
RDF is not the be all end all for representing information, so I don't know
if there is a point to defining a validation schema which can also be
represented in RDF since requirements vary from model to model, project to
project. If you were creating RDF/XML, you could enforce complex
validation through schematron. XForms 2.0 will support JSON and other
non-XML data models, so you could enforce complex validation through XForms
bindings since XPath 3 will support parsing JSON, thus JSON-LD.
Our project consists of (at the moment) tens of thousands of concepts
defined at URIs and represented by XHTML+RDFa fragments. These bits of
XHTML are edited in XForms, so the validation is pretty tight. The
XHTML+RDFa is transformed into RDF proper upon file save and posted into
our endpoint with the SPARQL/Update mechanism.
But my broader point is: RDF (typically) is a derivative resource of a more
detailed data model. In the case where the RDF is derivative of a
canonical resource/document, validation can be applied more consistently
during the editing process of the canonical resource.
Ethan
On Thu, Sep 12, 2013 at 11:19 AM, Karen Coyle <[email protected]> wrote:
I followed the W3C RDF Validation Workshop [1] over the last two days. The
web page has both written papers and slides from each presentation.
The short summary is that a number of users of RDF have found a need to do
traditional style validation (required, one or more, must be numeric/from a
list, etc.) on their RDF metadata. There is currently no RDF-based standard
for defining validation rules, so each of these is an ad hoc solution which
cannot be easily exchanged. [2]
The actual technology of validation in all cases is SPARQL. Whether or not
this really scales is one of the questions, but it seems pretty clear that
SPARQL will continue to be the solution for the near future.
I will try to write up a blog post that will give some more information.
kc
[1]
https://www.w3.org/2012/12/**rdf-val/agenda<https://www.w3.org/2012/12/rdf-val/agenda>
[2] nota bene: Although OWL appears to provide validation rules, the OWL
rules only support inferencing. OWL cannot be used to constrain your data
to valid values.
--
Karen Coyle
[email protected] http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet
--
Karen Coyle
[email protected] http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet
