W3C home > Mailing lists > Public > www-webont-wg@w3.org > September 2003

Re: best practices, Re: test results ontology

From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
Date: Wed, 17 Sep 2003 11:57:22 -0400 (EDT)
Message-Id: <20030917.115722.02613014.pfps@research.bell-labs.com>
To: sandro@w3.org
Cc: www-webont-wg@w3.org

From: Sandro Hawke <sandro@w3.org>
Subject: best practices, Re: test results ontology 
Date: Mon, 15 Sep 2003 04:59:13 -0400

> 
> 
> > >      http://www.w3.org/2002/03owlt/resultsOntology
> 
> > I suggest not using rdf:parsetype="Literal" - instead use xsd:string.
> 
> Hm.   Why xsd:string instead of plain literals?
> 
> I used rdf:parsetype="Literal" because:
> 
>     1.  It's good ("best") practice to allow html/markup, when
>         possible, in human-readable strings.  (Sorry, I can't find a
>         concise reference for this.)

Hmm.  How then should I human-read

	<ex:foo rdf:parsetype="Literal">Hi! 
		<bar comment="Don't read this."	/>
		<bar>Read this!</bar>
		Bye!
	</ex:foo>

>     2.  Given that the XMLLiteral "hello" is not the same thing as the
>         plain literal "hello", changing from 
> 
>            <rdfs:comment>I like cheese</rdfs:comment>
> 
>         to
> 
>            <rdfs:comment rdf:parsetype="Litera">I <em>like</em>
>                                                   cheese</rdfs:comment> 
> 
>         involves a change in two dimensions, which somehow seems
>         inelegant.  
> 
> Point 2 is quite weak.  I guess all I'm really attached to is making
> sure *some* comments have markup and that applications can handle that
> gracefully. 

The problem I see is that parsetype="Literal" is not very appropriate for
html/markup.

> > I suggest not limiting the output of a test run to be a document.  It is
> > entirely possible that the output of a test run is a fragment of a
> > document, or some other kind of entitity.  
> 
> What's important to me is that the URI given to name the output is one
> which works nicely in browsers, so that links in the table are
> helpful.  foaf:Document was the best I could find, but I agree
> something which allowed HTML anchors/fragments would be better.
> 
> I suppose I can drop the foaf:Document bit and just use a comment, but
> that smells a little like a failure of OWL.   What would you suggest
> as an rdfs:range here?

Well that is a good question.  There is no real problem with making our own
URI up for this, in the absence of a suitable existing one.

> > I suggest ensuring that the ontology is in OWL Lite.  I think that the only
> > reason that the ontology is not in OWL Lite is that
> > http://xmlns.com/foaf/0.1/Document is not an OWL class.
> > 
> > 	./owlParse http://www.w3.org/2002/03owlt/resultsOntology
> > 	Reading http://www.w3.org/2002/03owlt/resultsOntology
> > 	Fatal error: exception Owl.Syntax("Non-class uri for description: <http
> > ://xmlns.com/foaf/0.1/Document>
> > 	")
> 
> In your opinion, to fix that (assuming I wanted to keep using
> foaf:Document), I would need to get the FOAF folks to change their
> ontology to be OWL Lite and then I would have to use owl:imports,
> right?   Or could I leave out the imports part?

All that would be needed would be to include something that results in a
	<http://xmlns.com/foaf/0.1/Document> rdf:type owl:Class .
triple.

Imports is not needed.

> > It is not possible to correctly use xsd:duration in OWL (and RDF) as the
> > value space for xsd:duration is not well-defined.  This is only mentioned
> > in a comment, but I suggest that even this use is not appropriate.
> 
> Are you opposed to supporting the reporting of durations?  I've been
> toying with displaying the duration on the results page at an
> order-of-magnitude level, because I (as a user) would really like to
> have some clues there.   

Well, I'm certainly not proposing to support xsd:duration because I don't
know how it might work. 

> If not, do you have a suggestion for something instead of xsd:duration?

Well, again, if there is no suitable existing URI reference, I don't have a
problem with coining one.  If a well-known suitable URI reference is later
produced, the fresh-minted one can be stated to be equivalent to the
well-known one.

[...]

> > How does one report success in passing the syntax part of a test?
> 
> What DanC and I came up with was this:
> 
> <tres:test rdf:parseType="Resource">
>     <tres:syntacticLevelTestFrom rdf:resource="http://www.w3.org/2002/03owlt/description-logic/Manifest601#test"/>
>     </tres:test>
>     <tres:system rdf:resource="#owlp"/>
>     <tres:output rdf:resource="http://lists.w3.org/Archives/Public/public-webont-comments/2003Aug/0019.html"/>
> </tres:PassingRun>
> 
> where tres:syntacticLevelTestFrom is an owl:InverseFunctionalProperty
> from a "DerivedSyntacticLevelTest" to another test it's derived from.
> The derived test involves correctly identifying the level of ALL the
> documents use in the other test.  
> 
> I didn't document this in the test-results-ontology because it seemed
> too owl-specific; in fact, it seems like it belongs in the
> testOntology.  Jeremy?

This is also appropriate for RDF tests.  It might be better to have a
different category for passing the syntactic portion of a test.  So that
one would report directly on the test.

>      -- sandro

peter
Received on Wednesday, 17 September 2003 11:57:32 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:02 GMT