W3C home > Mailing lists > Public > public-media-annotation@w3.org > September 2010

RE: OWL FULL or DL?

From: Evain, Jean-Pierre <evain@ebu.ch>
Date: Thu, 23 Sep 2010 12:28:56 +0200
To: "Evain, Jean-Pierre" <evain@ebu.ch>, 'Tobias Bürger' <tobias.buerger@salzburgresearch.at>
CC: 'Raphaël Troncy' <raphael.troncy@eurecom.fr>, 'Yves Raimond' <yves.raimond@gmail.com>, "public-media-annotation@w3.org" <public-media-annotation@w3.org>
Message-ID: <7D1656F54141C042A1B2556AE5237D60010CEEB7F297@GVAMAIL.gva.ebu.ch>
In fact inverse relations between properties are fine. Only disjoints were causing some problems.

-----Original Message-----
From: Evain, Jean-Pierre 
Sent: jeudi, 23. septembre 2010 12:16
To: Evain, Jean-Pierre; 'Tobias Bürger'
Cc: 'Raphaël Troncy'; 'Yves Raimond'; public-media-annotation@w3.org
Subject: RE: OWL FULL or DL?

Actually the current version needs the following corrections to be DL:
- use rdf:XMLLiteral instead of punning
- remove disjoints
- Remove inverse relations between properties (but leave the properties in)

This being done it validates as DL.

I am not sending this as we are still discussing it with Tobias as I added a number of missing properties (the result of my homework on the mapping tables and the review of the ma-ont list of core properties and semantics)

Regards,

Jean-Pierre

-----Original Message-----
From: public-media-annotation-request@w3.org [mailto:public-media-annotation-request@w3.org] On Behalf Of Evain, Jean-Pierre
Sent: jeudi, 23. septembre 2010 11:06
To: 'Tobias Bürger'
Cc: 'Raphaël Troncy'; 'Yves Raimond'; public-media-annotation@w3.org
Subject: RE: OWL FULL or DL?

Thanks for the link Tobias.  This is actually very informative and tells a lot on owl-2.

As for the version we have now it is DL compliant I would think.  As you know several properties are missing and we need to find a way to accommodate some of the requirements mentioned in the semantics of the terms (e.g. use of rdfs:literal). We are otherwise not using inverse properties so far.  

The only complexity in the current version is the use of sub-classes and some sub-properties.

The issues were effectively the use of FOAF (a DL profile / version) would solve most of the problems but we can go our own route (hence what would be the point of developing and ontology but I know you know) and property puns.

JP

-----Original Message-----
From: Tobias Bürger [mailto:tobias.buerger@salzburgresearch.at] 
Sent: jeudi, 23. septembre 2010 10:49
To: Evain, Jean-Pierre
Cc: 'Raphaël Troncy'; 'Yves Raimond'; public-media-annotation@w3.org
Subject: Re: OWL FULL or DL?

  Dear all,

I missed the whole discussion due to traveling, but my two cents which I 
would like to share on this issue are:

If the ontology shall only be used as a vocabulary then we need not to 
care about OWL Full vs. OWL DL. Then we could make it fairly simple and 
just revert back to the RDF(S) version which was the first I have 
developed for the group. If we want to define a lot of restrictions in 
the usage of our vocabulary we have to go for OWL.
Wrt to including FOAF directly in our ontology: Many others do slightly 
change/adapt the FOAF ontology to be OWL-DL compliant (if they care 
because of their application scenario about it). There are a lot of 
guidelines and discussions on how to do this on the Web.

The lack of tools should not restrain us from going for an OWL2 profile. 
You can check this page for tools providing OWL2 support: 
http://www.w3.org/2007/OWL/wiki/Implementations

OWL2 vs OWL1: I do not care too much at the moment about this choice, 
unless we think that we need the newly defined functionality in OWL2.

Best,

Tobias

Am 23.09.2010 09:36, schrieb Evain, Jean-Pierre:
> What is ma-ont (not me, isn't it) trying to achieve? That is the question !
>
> My question was of course asked in this context (public-media-annotation@w3.org?). I hope experts in this group have views on this.
>
> If the task is about searching content of interest through different angles reflected by the ontology properties across different namespaces linked to ma-ont through mapping, then I interpret from your response that FULL should maybe not be used to maximise searchability resulting in more positive accurate hits.
>
> We all know this can be avoided by not using certain properties. The use of rdfs:literal may also be an alternative to punning leaving the choice to enter e.g. a URI (link to a SKOS concept although there is more to say about not using a concept class) or string.
>
> DL it is?
>
> I had a look at OWL-2 and its different profiles. QL would seem to be the right profile. However, I still need to look at some of the restrictions on classes in more details.  I also would need to make sure that tools exist to work on this.
>
> Jean-Pierre
>
>
>
> -----Original Message-----
> From: Raphaël Troncy [mailto:raphael.troncy@eurecom.fr]
> Sent: mercredi, 22. septembre 2010 23:39
> To: Evain, Jean-Pierre
> Cc: 'Yves Raimond'; Tobias Bürger; public-media-annotation@w3.org
> Subject: Re: OWL FULL or DL?
>
> I certainly didn't want to make this personal, so let's de-passionate
> the debate and talking only on the core matters. You first told us about
> supposed behaviors of FOAF that we have never seen, looking at only what
> is important, the machine readable version of the FOAF ontology (and not
> a particular rendering in a particular tool).
>
>> The point I made is disconnected from this as although this is a old
>> debate it seems there are still doubts on what should preferably be
>> used for operational implementation.
> I point you to precise questions. There are no doubts about operational
> implementation as soon as you know how the ontology will be used, what
> are its purpose. I would not recommend to develop OWL Full ontologies if
> the core matters is to do complex reasoning or if it is important to
> have all the good answers to a query. That's why, I asked you what was
> the purpose of the ontology.
>
>> This is a question for which I
>> have much more sympathy than floating in cyberspace using 'cool'
>> tools.
> Who talked about 'cool' tools or features or whatever?
>
>> Still you neither said anything I didn't know
>> nor brought a convincing argument in favour of FULL.
> I simply don't understand what are your issues. The question you are
> asking has no general answer, or if you prefer, the answer is "it
> depends" ... thus again, asking clarification, see my questions.
> Their only purpose is not to bother you, but just identifying precisely
> what you need to adopt the best technology ... this is also why we have
> flavors and profiles for OWL and OWL2. And by the way, this is the first
> thing that ontology engineering best practices recommend to do (not me
> saying this :-).
>
>     Raphaël
>

-- 
================================================================
Dr. Tobias Bürger         Knowledge and Media Technologies Group
Salzburg Research                           FON +43.662.2288-415
Forschungsgesellschaft                      FAX +43.662.2288-222
Jakob-Haringer-Straße 5/III   tobias.buerger@salzburgresearch.at
A-5020 Salzburg | AUSTRIA         http://www.salzburgresearch.at
Received on Thursday, 23 September 2010 10:45:11 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 23 September 2010 10:45:12 GMT