RE: Survey paper

Hi Cory,

This is one of the trickiest parts of the paper. I'm on a steep learning curve for RDFa and have scracthed my head before on xlink before but I'll try to help.

The way I would put it is:

1) xlink is almost the semantic equivalent in XML to void* pointers in C programs - when you got one, you're not always sure what you can do with it because the valid recipe to handle it will depends on the subset of xlink use cases which are allowed (similarily, how to deal with void* in C programs depends largely on the age of the captain. So, any solution which does not propagate it is preferable.

2) If you have to use it, you need to look at the fine print to check what the specs says:

For example, in SVG, the xlink usage is described here: http://www.w3.org/TR/SVG/linking.html


3) In my opinion, the GML usage of xlink is incompletly described in Section 8 of http://portal.opengeospatial.org/files/?artifact_id=20509

Simon and other OGC colleagues posting on the seegrid wiki have documented the xlink usage in GML in 2006 in two wiki pages which corresponds to the two different use cases I think you are talking about:

'model reference to a ontological description' vs 'composition by inclusion of remote resources'

I think that:
GML Implementation of Features and Properties = 'composition by inclusion of remote resources'
https://www.seegrid.csiro.au/twiki/bin/view/AppSchemas/GmlImplementation

and:
"Property metadata in GML ... overloading href, and use of other xlinks" = 'model reference to a ontological description'
https://www.seegrid.csiro.au/twiki/bin/view/AppSchemas/PropertyMetadata

The major difficulty you should be aware of is that there are several (subtly) different usages in of xlinks in usage in sub-communities of OGC:
Xlink can be used to point to "fragment" of other XML or HTML files (locatable through a # beacon declaration or with the help of  Xpath expression)

The GML spec authorises 4 variants:

EXAMPLE 1 A reference to an object element in the same GML document may be encoded as:

<myProperty xlink:href="#o1"/>

EXAMPLE 2 A reference to an object element in a remote XML document using the gml:id value of that object may be encoded as:

<myProperty xlink:href="http://my.big.org/test.xml#o1"/>

EXAMPLE 3* A reference to an object element in a remote XML document (or GML object repository) using the gml:identifier property value of that object may be encoded as:

<myProperty xlink:href="http://my.big.org/test.xml#element(//gml:GeodeticCRS[./gml:identifier[@codeSpace="urn:x-ogc:def:crs:EPSG:6.3:"]="4326"])"/>

*Personally, I have never seen anyone which uses this XPath augmented style of uris ...

EXAMPLE 4 A reference to an object element with a uniform resource name may be encoded as follows (note that a URN resolver is required to resolve the URN and access the referenced object):

<myProperty xlink:href="urn:x-ogc:def:crs:EPSG:6.3:4326"/>

These 4 examples correspond to a first case of  'composition by inclusion of remote resources'

One issue is that the GML specification let the door open to all the possibility defined by the XLink spec through this sentence "the most useful of these [xlink attributes] are" .... and "For complete definitions of these and other Xlink components, including their use in extended Xlink association maps, refer to the Xlink specification."

This GML xLink Profile (07-083) by Andrew Woolf http://epubs.cclrc.ac.uk/bitstream/1851/AWO%20-%20xlink.ppt describes how some of these xlink attributes should be used in GML for a small sets of specific use cases. It has also been proposed around 2006 but to my knowledge it has not been approved or recyled in a formally approved GML spec. probably because it corresponds to a specific use case, when xlink are used to locate resources embedded in netCDF files.


I have to rush a bit here because you have exchanged 3 messages since I have started to type.

I think that the usage of xlink in SWE may be different and = 'model reference to a ontological description'

This is a different case because the xlink pointer is no longer used to point to something which would corresponds to a RDF instance (container of data) but rather to what would be a property of a class in the ontology (or in a UML model).

4) Different types of URIs

Long story (another post needed). For better scoped definitions, see the skos notation or the CURIEs spec.

5) Source of confusion (in general and in the paper)

There are many possible combinations of the above usage. You are introducing a new one which is to use xlink for annotations.

Question: do you want to attach an annotation to a block of XML file which could have been replaced by a xlink pointer for the purpose of 'composition by inclusion of remote resources'  (e.g a gml:Point) or a block of the XML file which is a  'model reference to a ontological description' vs 'composition by inclusion of remote resources'
(e.g. a swe:Phenomenon). I think to you should have different mappings for the two cases.

6) To be continued ...

Laurent.

PS: The GML spec says page 20:  GML follows RDF (W3C, 1999) terminology and uses the term property rather than attribute or association role.
This translates to something like: some of the names of class or attributes used in GML have been borrowed from RDF.

But not to something like: When the user find a RDF-like concept in GML, he can be confident that the semantics of the corresponding concept or role in the RDF specs are applicable to it.


________________________________
From: public-xg-ssn-request@w3.org [mailto:public-xg-ssn-request@w3.org] On Behalf Of Cory Henson
Sent: Thursday, 13 August 2009 14:27
To: Simon Cox
Cc: Michael.Compton@csiro.au; public-xg-ssn@w3.org
Subject: Re: Survey paper

Hi Simon,

Thanks for the comment.  We are using the term semantic annotation as described in SAWSDL, as a model reference to a ontological description.  Does this conflict with the description as a 'composition by inclusion of remote resources'? As far as mapping to RDF, this is in comparison to RDFa which has a known syntactic translation from the set of annotations to RDF triples. While xlink:href maps to rdf:resource, how would the values of properties of this resource be translated to RDF? If this is not correct, or the wording is awkward, please point us in the right direction.  Thanks for your help.

-Cory


On Thu, Aug 13, 2009 at 8:01 AM, Simon Cox <simon.cox@jrc.ec.europa.eu> wrote:
Hi Mike -

A clarification relating to semantic annotations and xlink:

In GML-style XML documents, xlink:href plays the same role as rdf:resource
in an RDF/XML document.
I.e. it holds a pointer to external resource, which could be pasted inline
as an anonymous node with equivalent semantics.
This is a basic GML pattern and is explained in the GML spec
http://portal.opengeospatial.org/files/?artifact_id=20509 clause 7.2.3.
The examples involving xlink:href in Figure 2 aren't exactly 'annotations',
more 'composition by inclusion of remote resources'.

So I'm not sure if the example supports the point you are making.

You comment 'XLink has no predefined mapping to RDF.'
As mentioned above, _as used in GML documents_ xlink:href maps to
rdf:resource.

Simon Cox

European Commission, Joint Research Centre,
Institute for Environment and Sustainability,
Spatial Data Infrastructures Unit, TP 262
Via E. Fermi, 2749, I-21027 Ispra (VA), Italy
Tel: +39 0332 78 3652
Fax: +39 0332 78 6325
mailto:simon.cox@jrc.ec.europa.eu<mailto:simon.cox@jrc.ec.europa.eu>
http://ies.jrc.ec.europa.eu/simon-cox

SDI Unit: http://sdi.jrc.ec.europa.eu/
IES Institute: http://ies.jrc.ec.europa.eu/
JRC: http://www.jrc.ec.europa.eu/

-----Original Message-----
From: public-xg-ssn-request@w3.org<mailto:public-xg-ssn-request@w3.org> [mailto:public-xg-ssn-request@w3.org<mailto:public-xg-ssn-request@w3.org>] On
Behalf Of Michael.Compton@csiro.au
Sent: Thursday, 13 August 2009 13:43
To: public-xg-ssn@w3.org<mailto:public-xg-ssn@w3.org>
Subject: Survey paper

Hi all,

Sorry it's so close to the SSN'09 deadline, but with help from Cory and
Holger, I (finally) have a survey paper.  Please read, comment, etc.

(there are a couple of obvious tweaks/FIXME's yet to be made)

Michael





--
Cory Andrew Henson
Kno.e.sis Center, Wright State University
http://knoesis.wright.edu/researchers/cory/