Re: [ALL] Human-friendly syntax for communicating OWL fragments

Uschold, Michael F wrote:
> I'm not talking about mapping the abstract syntax to any triple, just to those triples that are part of OWL [expressed in RDF]. 
> 
> I thought the abstract syntax was just an syntactic variant, therefore the mapping (I would expect) should be, at least for semantic purposes, one to one.  Any differences should not interfere with human understanding, nor machine-processing. 
> 
> Am I mistaken?


Yes.

(It's very tempting to stop there ...)
Ideally if a1 and a2 are abstract syntax trees and g1 and g2 are RDF graphs 
and a1 maps to g1 and g2, and also a2 maps to g1 then at a first 
approximation they all have the same OWL DL semantics  (a1 and a2 with the 
direct semantics, g1 and g2 with the RDFS compat semantics). So that 
"should not interfere" is correct.
However, if what you want to do is to use some OWL Full feature, it is not 
clear at all whether or not the (weak and unproven) semantic guarantees 
involved in the mapping are appropriate. e.g. foaf uses Inverse Functional 
Property on DatatypeProperties.

Basically if what you want to talk about is soemthing to do with OWL Lite 
or OWL DL, then the abstract syntax is appropriate, but using it to talk 
about dc:creator say does not work because it forces you into making 
decisions that are not appropriate for dc:creator.

Jeremy


> 
> MIke
>  
> 
>  -----Original Message-----
> From: 	Jeremy Carroll [mailto:jjc@hplb.hpl.hp.com] 
> Sent:	Tuesday, March 30, 2004 11:04 PM
> To:	Jim Hendler
> Cc:	Uschold, Michael F; public-swbp-wg@w3.org
> Subject:	Re: [ALL]  Human-friendly syntax for communicating OWL fragments
> 
> As always I share both Jim's and Mike's concerns - despite them disagreeing.
> 
> It really doesn't help that the abstract syntax to triples mapping is 
> many-to-many ...
> 
> Jeremy
> 
> 
> 
> Jim Hendler wrote:
> 
> 
>>At 20:09 -0800 3/30/04, Uschold, Michael F wrote:
>>
>>
>>>I propose that by convention all OWL fragments are given using the 
>>>reader-friendly abstract syntax, rather than the parser-friendly XML 
>>>syntax.  Personaly, I cannot read OWL fragments well enough to be 
>>>motivated to ever understand the details, so I tend to skip over 
>>>them.  Even when I can get motivated, it takes way too much time.
>>
>>
>>I object to this strongly -- that syntax is neither reader friendly nor 
>>actually a part of the recommendation.  Much more importantly, we should 
>>be creating fragments that people can cut and paste into their documents 
>>(and edit) -- forcing them to figure out the mapping from the so-called 
>>human readable syntax into actually RDF or OWL (XML or N3) makes no sense.
>>
>>
>>>I would argue very strongly that any public documents published by 
>>>this WG  do use the more readable syntax. Why not get used to it when 
>>>we communicate with each other? It will also make it easier to grab 
>>>things from discussions in the archive and plunk them into documents, 
>>>instead of having to translate into the abstract syntax suitable for 
>>>the public.
>>
>>
>>I advocate use of "turtle" - which should be the first document this WG 
>>approves as a working note (if Dave is willing)
>>
>>
>>>Of course, if the discussion is about parsing, or about the syntax of 
>>>the language, then it is better to use the parser-friendly syntax, 
>>>both for internal discussions and for publised documents.
>>>
>>
>>what is parser friendly?
>>
>>
>>>What do people think about this suggestion?
>>>
>>>If the overwhelming majority of this WG actually PREFER to read the 
>>>parser-friendly syntax, then perhaps I'd best get used to it, but it 
>>>there are many like me, it makes sense to use a more reader-friendly 
>>>syntax.
>>>
>>
>>well, we could get into whether this is a voting issue for a WG (be 
>>careful when you use words like "majority" in a W3C group) but my vote 
>>is for N3 (Turtle) which is a nice compromise - or else to stick w/the 
>>RDF/XML for cut and paste reasons
>> -JH
>>p.s. Mike - have you noticed our world view doesn't always seem to align 
>>:->
> 
>

Received on Tuesday, 6 April 2004 05:58:49 UTC