Re: Profiles intro

Carsten,

and I understand very well that these things are complicated... And the 
error I made on my example just shows how difficult that is:-(

However... eventually we do need a kind of an 'elevator pitch', ie, a 
short, clear, and not-too-detailed argument for end-users when and why 
they would use one profile over the other.

In some ways, we had that for  OWL-Full and DL: I know, it can be 
discussed at nauseam, it is not precise, and tastes differ a lot, but I 
remembered Dan Brickley saying something like:

 - Some application just need to express and interchange terms (with 
possible scruffiness): OWL Full is fine
 - Some applications need rigor and complex term classification with a 
guarantee offered by reasoning procedures; then OWL DL/Lite might be the 
good choice

which is good enough for the lambda SW application developer.

Let us try to agree upon some of these...

Ivan

Carsten Lutz wrote:
> On Wed, 9 Apr 2008, Ivan Herman wrote:
>>
>> Bijan Parsia wrote:
>>>
>>> It is really nice (as far as I'm concerned), but I think we can 
>>> anticipate some negative comment about the emphasis on the 
>>> "tractability" POV (e.g., from Jim Hendler). And, after all, 
>>> computational efficiency is a necessary, not sufficient, condition 
>>> for inclusion. 
>>
>> +1. No, +100...:-)
>>
>> What I am looking for are statements that make it clear in which 
>> circumstances I would choose one profile over the other (even if I 
>> have no idea of the implementation details, nor do I want to deal with 
>> those). The fact that it can be implemented in polynomial time or 
>> whatever is only one (albeit important) aspect.
> 
> I agree and certainly don't consider the intro finished. The purpose
> of my changes was only to have an intro that can be published at the
> end of this week, and I think the current one can. (the action on me
> was to put the explanations of the three fragments that I had in my
> OWLED slides, and I did even more than that).
> 
> An issue with the other aspects of fragments is that they are difficult
> to formulate without raising controversy. Let's take your example:
> 
>> I have heard arguments that say "if your ontology has a simple 
>> structure, but have a large abox, then use DL-Lite";
> 
> I am not totally happy with this formulation, and I guess Zhe is even
> less so:
> 
> - I am uncomfortable with the "ontology has a simple structure part",
>   because EL++ is also targeted at ontologies of a simple structure and
>   can also handle ABoxes, whereas I conceive DL-Lite as a constraint
>   language rather than an ontology language.
> 
> - I suppose Zhe won't like the "large ABox" part, because ontologies
>   with a simple structure and large data is precisely what OWL-R is
>   also made for.
> 
> My aim here is not to discuss these issues, but only to point out that
> it may by hard to find general rules of the kind that you imagine.
> 
>> I could imagine that a more detailed argumentation should (probably 
>> must...) be given in more details in the primer, but some of these 
>> should be added, I believe, in the profile document, too.
> 
> I agree, but we have to be careful.
> 
> greetings,
>         Carsten
> 
> -- 
> *      Carsten Lutz, Institut f"ur Theoretische Informatik, TU 
> Dresden       *
> *     Office phone:++49 351 46339171   
> mailto:lutz@tcs.inf.tu-dresden.de     *

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf

Received on Wednesday, 9 April 2008 09:06:06 UTC