Re: [xml-dev] ANN: White Paper - "Using OWL to Avoid Syntactic Rigor Mortis"

["Rick Jelliffe" <ricko@allette.com.au>]

Interesting to couple this with Abstract Patterns in Schematron 1.6 (the code for which 
has been released in dribs: I have promised an official release this week). Abstract
patterns can be implemented as simple macros. The following Schema specifies which 
are required (a real schema would have more adventurous constraints.)


<sch:schema xmlns:sch="http://www.ascc.net/xml/schematron";  version="1.6">
 <sch:title>Camera Details</sch:title>

 <!-- THE ABSTRACT PATTERN FOR A CAMERA SPECIFICATION --> 
 <sch:pattern abstract="true" name="Camera">
  <sch:rule context=" $Camera ">
   <sch:assert test=" $ViewFinder ">A camera spec should have a view finder</sch:assert>
   <sch:assert test=" $FocalLength ">A camera spec should have a focal length for its lens</sch:assert>
   <sch:assert test=" $Aperture ">A camera spec should have an apperture</sch:assert>
   <sch:assert test=" $ShutterSpeed ">A camera spec should have a shutter speed</sch:assert>   
  </sch:rule>
 </sch:pattern>

  <!-- ROGER'S FIRST FORMAT --> 
 <sch:pattern name="SLR-format"   is-a="Camera" 
    see="http://www.xfront.com/avoiding-syntactic-rigor-mortis.html";>
  <sch:param name="Camera"  value="SLR" />
  <sch:param name="ViewFinder"  value="ViewFinder" />
  <sch:param name="FocalLength"  value="optics/Lens/focal-length" />
  <sch:param name="Aperture"  value="optics/Lens/f-stop" />      
  <sch:param name="ShutterSpeed"  value="shutter-speed" />        
 </sch:pattern>
 
 <!-- ROGER'S SECOND FORMAT -->
 <sch:pattern name="Alternative-Format"  is-a="Camera" 
    see="http://www.xfront.com/avoiding-syntactic-rigor-mortis.html";>
  <sch:param name="Camera"  value="Camera" />
  <sch:param name="ViewFinder"  value="ViewFinder" />
  <sch:param name="FocalLength"  value="optics/Lens/size" />
  <sch:param name="Aperture"  value="optics/Lens/aperture" />      
  <sch:param name="ShutterSpeed"  value="shutter-speed" />        
 </sch:pattern>
 
 <!-- AND ANOTHER FORMAT -->
  <sch:pattern name="SLR-table-format">
  <sch:param name="Camera"  value="camera-table/data/row" />
  <sch:param name="ViewFinder"  value="*[position()=../names/name[.="ViewFinder"]/position()" />
  <sch:param name="FocalLength"  value="*[position()=../names/name[.="LensSize"]/position()" " />
  <sch:param name="Aperture"  value="*[position()=../names/name[.="LensApperture"]/position()" " />      
  <sch:param name="ShutterSpeed"  value="*[position()=../names/name[.="ShutterSpeed"]/position()" " />        
 </sch:pattern>
 
</sch:schema>



What is that last entry? It copes with tabular data that looks like this:

<camera-table>
<names>
    <name>Model</name><name>ViewFinder</name><name>LensSize</name>
    <name>LensAperture</name><name>ShutterSpeed</name>
</names>
<data>
<row><d>Olympus-OM-10</d><d>twin mirror</d><d>75-300mm zoom</</d>
        <d>4.0-4.5</d><d>1/2000 sec. to 10 sec.</</d></row>
<row><d>Olympodes-OMS-10</d><d>twin mirror</d><d>76-290mm zoom</</d>
        <d>4.0-4.5</d><d>1/100 sec. to 1 sec.</</d></row>
...

</data>
</camera-table>

If I have the (untested) XPaths right, it is even independent of the number of
columns or their order. Open.

I haven't followed OWL: it would be interesting to figure out how Schematron
and OWL stand in relation to each other. Schematron is interested in patterns
(i.e. views of structures, especially non-hierarchical structures) though it does 
allow some labelling of nodes. 

Cheers
Rick Jelliffe