[higgins-dev] [IdAS] filter language: XPath/SPARQL

[Greg Byrd <gbyrd@xxxxxxxx>]

I've been thinking about the filter language, and (as usual) have come 
up with more questions than answers.  My plan was to use the 
recently-posted Example Ontology 
(http://spwiki.editme.com/ExampleContextOntology) as a springboard to 
compare XPath expressions and SPARQL expressions.   Didn't quite get 
there...

For the following discussion, I will ONLY be considering the filter used 
in Context.getSubjects(filter). 

-----
Question 1:  Is the filter expression intended to apply to (a) the 
entire Context (i.e., the result of the filter should be the set of 
DigitalSubjects that match), or (b) to a DigitalSubject (i.e., look at 
each DigSub and add it to the list if the filter is true)?

If (a) then we might have matches to nodes (objects) that aren't 
DigSubs.  If (b), we can't take advantage of SPARQL expressions that let 
you control the ordering, number, etc., of the results.  Going forward, 
I'll assume that it's (b).


-----
Question 2:  Is there a canonical RDF/XML representation of any given 
RDF graph?  The Primer (http://www.w3.org/TR/rdf-primer/) lists many 
abbreviations that can be used, and so there are many possible XML 
representation of a given graph.  For example, let's assume that the 
example Person type only has two elements: firstname and surname.  The 
subject could be represented either as:

<rdf:Description rdf:about="urn:jim-sermersheim">
      <jim:firstname>Jim</jim:firstname>
</rdf:Description>

<rdf:Description rdf:about="urn:jim-sermersheim">
      <jim:surname>Sermersheim</jim:surname>
</ref:Description>

Or as the following:

<rdf:Description rdf:about="urn:jim-sermersheim">
      <jim:firstname>Jim</jim:firstname>
      <jim:surname>Sermersheim</jim:surname>
</ref:Description>

In the second case, an XPath expression '.[jim:firstname="Jim" and 
jim:surname="Sermersheim"]' would work, but in the first case it 
wouldn't.  (I apologize if I don't get the XPath syntax quite right -- 
I'm a newbie.)  Is there a way to handle the first case in XPath?  If 
the "canonical" RDF/XML representation combines all nodes with the same 
rdf:about attribute, then maybe this is OK.

In both cases, the following SPARQL query would be fine: { ?x 
jim:firstname "Jim" . ?x jim:surname "Sermersheim" }.

Another case is the representation of type.  In Paul's example, we say 
that urn:jim-sermersheim is a Person as follows:

<rdf:Description rdf:about="urn:jim-sermersheim">
      ...
      <rdf:type>
            <rdf:Description rdf:about="&jim;Person>
      </rdf:type>
</rdf:Description>

But in the Primer, they tend to use the following:

<rdf:Description rdf:about="urn:jim-sermersheim">
      ...
      <rdf:type rdf:resource="http://www.novell.com/jim#Person"; />
</rdf:Description>


In one case the type is an element, and in the other it's an attribute.  
How would you use XPath in this case?


-----
Question 3:  In the example, hasPostalAddress is a separate object.  The 
postal address information is not an XML descendent of the 
jim-sermersheim node, and there's no link back to the jim-sermersheim 
node from the jims-address node.

Given this schema, how would one search for all DigSubs from a 
particular city?

In SPARQL, you'd say: { ?x jim:hasPostalAddress ?addr . ?addr jim:city 
"Provo" }.

Is there a way to do this in XPath?  I wasn't able to figure it out, 
based on my limited research.