From Alan.Rector at manchester.ac.uk  Wed Jan  4 19:17:01 2006
From: Alan.Rector at manchester.ac.uk (Alan Rector)
Date: Wed Jan  4 19:14:09 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <43A1717E.5060207@aifb.uni-karlsruhe.de>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu>
	<43A1717E.5060207@aifb.uni-karlsruhe.de>
Message-ID: <46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>

Peter, Danny, All

Is part of the problem that we have no way to distinguish higher  
order statements about the domain from annotations of the symbols and  
artefacts we are using to represent that domain?   Or do we?

This seems to me a common problem.  When I make an annotation on some  
bit of an symbolic artefact, I am giving information about the  
symbols in the artefact, not the domain they represents. So if I  
author a symbol "Galapagos Finch" and a symbol "Darwin"  and a symbol  
"Rector", I want to be able to say that the Symbols "Galapagos Finch"  
and "Darwin" in this ontology were created by "Rector" but I don't  
want to reason about them in the same space. (By contrast if I want  
to say that the class denoted by the symbol "Galapagos Finch" was  
described by "Darwin", I am talking about the class "Galapagos Finch"  
which is an instance of "owl:Class". Or as a more clear cut example,  
I want to say that "Galapagos Finch" is an endagered species and be  
able to ask queries or form the class of "All members of endangered  
species". Again, I am making a higher order statement about the class  
"Galapagos Finch" in the domain rather than about the symbolic  
representation of that that class. )

For annotation of the symbolic representation as opposed to higher  
order statements about the class represented, what you seem to need  
is a) a means of referring to the symbolic expressions "Galapagos  
Finch", "Darwin" etc. and the expressions using them rather than to  
their referents.

Unfortunately, this is not - at least as I understand it - what RDF  
Reification does.  As I read the Semantics document - please correct  
me if I am wrong - the semantics of reification as described by:

".. <begin quote http://www.w3.org/TR/rdf-syntax-grammar.>..

<ex:a> <ex:b> <ex:c> .

and

_:xxx rdf:type rdf:Statement .
_:xxx rdf:subject <ex:a> .
_:xxx rdf:predicate <ex:b> .
_:xxx rdf:object <ex:c> .

.... end quote"

is to create a blank node that represents ANY triple of the form

  <ex:a> <ex:b> <ex:c>

rather than the specific triple in my representation artefact.

I want to say things about the specific symbolic triple in a specific  
resource.  I don't know if this is possible by referring to the full  
URI of the rdf statement or not, but I can't see how as rdf triples  
do not appear to have URIs, even though their subject, property, and  
object all do.  I think what is being suggested is that OWL axioms,  
or at least some OWL axioms, should.

Obviously, without some kind of partitioning, allowing a statement to  
refer to another statement would allow a statement to refer to  
itself, directly or indirectly, and  would bring with it the  
paradoxes of self reference and therefore undecidability,  at least  
as soon it was used with an expressive  enough semantics to include  
negation or falsehood.

The simplest possible form of partitioning would seem to be to  
declare that some properties are annotations and do not affect  
inference - what I understood annotations did.  In fact, what I want  
to do, almost always, is query the actual representation with  
something like database (datalog) semantics, e.g. "Find all the  
axioms in ontology X that were imported from ontology A"; "Find all  
the statements in ontologies X,Y and Z asserted to be authored by  
'Rector'" .  "Find statements in ontology X for which there is no  
assertion of authorship", etc.

This seems to follow naturally from Ian Horrocks' note on Databases  
and Ontologies in which he says "DL KB doesn?t define a single model,  
it is a set of constraints that define a set of possible models...In  
contrast, DBs (and frame/rule KR systems) make assumptions such that  
DB/KB defines a single model"  This is a case where I am clearly  
interested in one possible model - the symbols and axioms actually in  
my implementation of the OWL KB and how they got there, how they fit  
together, etc.

There are a host of issues about copies and versions etc.,  but I  
can't see trying to solve them until the fundamental principles are  
clear. To start with at a minimum, I need to be able to send an  
artefact along with the description of that artefact in terms of  
provenance, usage, and editing annotations needed for somebody else  
(who shares the same annotation vocabulary) to use it in the same  
way, know where it came from, decide if they trust it, etc. (Note I  
trust artefacts not the things the artefacts represent.).

As far as I can see,  The use of the same symbol as a class and  
instance in OWL Full is a way of doing higher order representation,  
although this is not explicitly clear in anything that I can find.   
In which case the problems of fine grained annotation apply equally  
to OWL full and OWL-DL.

In summary, my contention is that there are three distinct kinds of  
statements about a) all instances of a class, b) the class itself,  
and c) the symbolic representation of the class. Further that we need  
to be able to refer to all constructs of the representation that can  
be expressed individually including individual triples, axioms,  (and  
the use of axioms to apply a restriction to a class) as well as  
classes and individuals.   My concern is that b) and c)  are often  
conflated. I would refer to b) as "higher order representation" and  
c) as" annotation". This conflation is reflected in the common use of  
the word "metadata" indiscrimately to refer to either or both.

I find it hard to see how we can resolve the issues around annotation  
definitively without a clear distinction between these two notions.   
If this is true, then there is a long term problem for which a best a  
'band aid' can be provided in OWL 1.1.

Or am I just completely confused?

Regards

Alan




On 15 Dec 2005, at 13:37, Denny Vrandecic wrote:

> Hi Bijan,
>
> thanks, very good and important questions. I'll try to answer them,  
> but I am happy if others would join in resolving them as well.
>
> First, let's ask a question in return: are axioms resources? Do we  
> want to be *able* to speak about axioms? Is there anything  
> interesting to say about axioms? If you say yes to any of these,  
> they need an URI. It appears so totally wrong to me, that we  
> (supposedly) can talk about anything and that we ask anyone to  
> provide URIs for their stuff, so we can pull it into the Semantic  
> Web - but we do not offer URIs for our own statements. I'm against  
> a "You shall not annotate axioms"-law. Does anyone disagree on  
> that? - and if so, why?
>
>> When is the URI being treated as an axiom?
>
> URIs of individuals, classes, properties... and axioms are all  
> disjoint from each other, and thus, just as URIs of individuals,  
> classes, properties... etc. axiom URIs need to be either declared  
> or it must be inferrable from the context that this is an axiom URI.
>
>> Does the mere presence of an axiom URI in a KB put the axiom  
>> itself there?
>
> I don't know. This would mean that the axiom must be known to the  
> reasoner. And if it is known, isn't it already there? Hmm. Maybe  
> not. I'd prefer a "no, it doesn't", but I am not sure about the  
> implications of that.
>
>> Can I make any interesting constructions using the Axiom uri?
>
> I sure hope so, but what do you have in mind?
>
>> Should axiom equivalence entail URI equivalence?
>
> Yes. Because then I can merge ontologies and the annotations of  
> axioms merge fittingly.
>
>> (Presumably not, eh, but what about through trivial rewritings,  
>> i.e., different rdf serializations?)
>
> Huh, why not?
>
>> Are all axioms named by default, or must one coin a name to talk  
>> about them?
>
> The latter.
>
>> Are annotations on axioms restricted to axioms with names/URIs?
>
> Can you make annotations on unnamed descriptions in OWL DL?
>
>> How do we avoid the pain that RDF Reification brings (and it's not  
>> just the rdf:subject/predicate, etc. the Statement triple itself  
>> and the introduction of the URI means CONSIDERABLE pain....note  
>> the fact that no store, to my knowledge, that introduces  
>> "contexts" does so by naming all the triples).
>
> I wish I had a convincing answer for that. What about "only  
> annotated axioms actually need an URI"? Or "this is just enabling  
> you to give an axiom an URI, it is not forcing you"? Or "hmm, I  
> think we can hide the pain most of the time, by nicely extending  
> the abstract syntax or the OWL XML presentation syntax. It will  
> only remain really painful and ugly in the XML/RDF serialization"?
>
> As said, none of these answers really convince me, but maybe they  
> are a starter for a solution. Or someone else has one. Maybe there  
> is a way to give URIs to axioms without RDF reification, this would  
> be great, and I'm just not seeing it.
>
> The question is, is avoiding this pain worth the restriction of  
> disallowing to annotate axioms?
>
> Cheers,
> denny
> _______________________________________________
> OWL mailing list
> OWL@lists.mindswap.org
> http://lists.mindswap.org/mailman/listinfo/owl

-----------------------
Alan Rector
Professor of Medical Informatics
Department of Computer Science
University of Manchester
Manchester M13 9PL, UK
TEL +44 (0) 161 275 6188/6149
FAX +44 (0) 161 275 6204
www.cs.man.ac.uk/mig
www.clinical-esciences.org
www.co-ode.org

From danny.ayers at gmail.com  Thu Jan  5 10:04:14 2006
From: danny.ayers at gmail.com (Danny Ayers)
Date: Thu Jan  5 15:58:11 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu>
	<43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
Message-ID: <1f2ed5cd0601050704o41b10b01m363b7ccd54f0af69@mail.gmail.com>

On 1/5/06, Alan Rector <Alan.Rector@manchester.ac.uk> wrote:

> I want to say things about the specific symbolic triple in a specific
> resource.

fyi, see Named Graphs -
http://www.w3.org/2004/03/trix/

in particular:
Named Graphs, Provenance and Trust
Carroll, Jeremy J.; Bizer, Christian; Hayes, Patrick; Stickler, Patrick

http://www.hpl.hp.com/techreports/2004/HPL-2004-57R1.html

Cheers,
Danny.

--

http://dannyayers.com
From Alan.Rector at manchester.ac.uk  Fri Jan  6 13:53:18 2006
From: Alan.Rector at manchester.ac.uk (Alan Rector)
Date: Fri Jan  6 16:50:51 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <1f2ed5cd0601050704o41b10b01m363b7ccd54f0af69@mail.gmail.com>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu>
	<43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<1f2ed5cd0601050704o41b10b01m363b7ccd54f0af69@mail.gmail.com>
Message-ID: <1D25E509-14C1-4622-8010-62570C7999DE@manchester.ac.uk>

Danny

Yes. I meant to mention Named Graphs - apologies to Jeremy and  
yourself - but are they inside or outside existing RDF specifications/ 
standards?

Alan

On 5 Jan 2006, at 15:04, Danny Ayers wrote:

> On 1/5/06, Alan Rector <Alan.Rector@manchester.ac.uk> wrote:
>
>> I want to say things about the specific symbolic triple in a specific
>> resource.
>
> fyi, see Named Graphs -
> http://www.w3.org/2004/03/trix/
>
> in particular:
> Named Graphs, Provenance and Trust
> Carroll, Jeremy J.; Bizer, Christian; Hayes, Patrick; Stickler,  
> Patrick
>
> http://www.hpl.hp.com/techreports/2004/HPL-2004-57R1.html
>
> Cheers,
> Danny.
>
> --
>
> http://dannyayers.com

-----------------------
Alan Rector
Professor of Medical Informatics
Department of Computer Science
University of Manchester
Manchester M13 9PL, UK
TEL +44 (0) 161 275 6188/6149
FAX +44 (0) 161 275 6204
www.cs.man.ac.uk/mig
www.clinical-esciences.org
www.co-ode.org

From danny.ayers at gmail.com  Fri Jan  6 18:40:55 2006
From: danny.ayers at gmail.com (Danny Ayers)
Date: Fri Jan  6 19:50:40 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <1D25E509-14C1-4622-8010-62570C7999DE@manchester.ac.uk>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu>
	<43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<1f2ed5cd0601050704o41b10b01m363b7ccd54f0af69@mail.gmail.com>
	<1D25E509-14C1-4622-8010-62570C7999DE@manchester.ac.uk>
Message-ID: <1f2ed5cd0601061540i39295db0i77dca41b65bd5ffe@mail.gmail.com>

On 1/6/06, Alan Rector <Alan.Rector@manchester.ac.uk> wrote:
> Danny
>
> Yes. I meant to mention Named Graphs - apologies to Jeremy and
> yourself - but are they inside or outside existing RDF specifications/
> standards?

Outside of the specifications, though I believe they are supported by
several of the RDF tool implementations (mostly implicitly I'd guess,
via contexts of one form or another).

Cheers,
Danny.

--

http://dannyayers.com
From jpan at csd.abdn.ac.uk  Sun Jan  8 15:05:20 2006
From: jpan at csd.abdn.ac.uk (Jeff Z. Pan)
Date: Sun Jan  8 15:05:42 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
Message-ID: <005101c6148e$df46aba0$d3d7858b@Newton>

Hi Peter, Alan and all,

After reading Alan's following email and the proposed OWL 1.1 syntax [1], it seems to me that punning is not a convincing choice for metamodeling in OWL 1.1. ( For those who are not familiar with punning - punning means that a name, like Person, can be used as both an individual and a class and a property.)  

1. It is impossible to distinguish higher order statements from annotations of symbols and  
artefacts we are using to represent that domain, as pointed out in Alan's email. The reason that they are not distinguishable is because annotations in [1] are simply syntactic sugar of individual axioms.

2. Datatype axioms, unlike other axioms in OWL 1.1 [1], cannot have annotations. This seems pretty strange, at least to me. The reason is that although individuals, object properties and classes can share names, classes and datatypes cannot. 

3.The semantics of punning is not quite intuitive. This can be shown in the following example. In the following  OWL 1.1 [1] ontology, Cat and Kitty are used as both classes and individuals. Although Cat and Kitty are the same individual and Ted is a Cat, the ontology

Class (Cat partial)
Class (Kitty partial)
SameIndividual (Cat Kitty)
Individual (Ted Cat)

does not entail that Ted is also a Kitty. This distinguishes the punning semantics from many other semantics, such as the OWL FA semantics [2], Hilog semantics [3] and RDF semantics [4].

Summary: 

The above point 2 suggests that the use of punning cannot really eliminate the need for annotations properties. Point 1 provides a good reason to distinguish annotation properties from metamodeling. Point 3 suggests that punning (as an option for metamodeling in OWL 1.1) is not quite intuitive and can be misleading. 

Greetings,
Jeff

[1] http://www-db.research.bell-labs.com/user/pfps/owl/syntax.html
[2] http://dl-web.man.ac.uk/~panz/Zhilin/pubc.php?type=epapers&id=PHS05
[3] http://dip.semanticweb.org/documents/Boris-Motik-On-the-Properties-of-Metamodeling-in-OWL.pdf
[4] http://www.w3.org/TR/rdf-mt/

--
Dr. Jeff Z. Pan (http://www.csd.abdn.ac.uk/~jpan/)
Department of Computing Science, The University of Aberdeen


----- Original Message ----- 
From: "Alan Rector" <Alan.Rector@manchester.ac.uk>
To: "Denny Vrandecic" <dvr@aifb.uni-karlsruhe.de>; "Peter Patel-Schneider" <pfps@research.bell-labs.com>
Cc: <owl@lists.mindswap.org>
Sent: Thursday, January 05, 2006 12:17 AM
Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1


> Peter, Danny, All
> 
> Is part of the problem that we have no way to distinguish higher  
> order statements about the domain from annotations of the symbols and  
> artefacts we are using to represent that domain?   Or do we?
> 
> This seems to me a common problem.  When I make an annotation on some  
> bit of an symbolic artefact, I am giving information about the  
> symbols in the artefact, not the domain they represents. So if I  
> author a symbol "Galapagos Finch" and a symbol "Darwin"  and a symbol  
> "Rector", I want to be able to say that the Symbols "Galapagos Finch"  
> and "Darwin" in this ontology were created by "Rector" but I don't  
> want to reason about them in the same space. (By contrast if I want  
> to say that the class denoted by the symbol "Galapagos Finch" was  
> described by "Darwin", I am talking about the class "Galapagos Finch"  
> which is an instance of "owl:Class". Or as a more clear cut example,  
> I want to say that "Galapagos Finch" is an endagered species and be  
> able to ask queries or form the class of "All members of endangered  
> species". Again, I am making a higher order statement about the class  
> "Galapagos Finch" in the domain rather than about the symbolic  
> representation of that that class. )
> 
> For annotation of the symbolic representation as opposed to higher  
> order statements about the class represented, what you seem to need  
> is a) a means of referring to the symbolic expressions "Galapagos  
> Finch", "Darwin" etc. and the expressions using them rather than to  
> their referents.
> 
> Unfortunately, this is not - at least as I understand it - what RDF  
> Reification does.  As I read the Semantics document - please correct  
> me if I am wrong - the semantics of reification as described by:
> 
> ".. <begin quote http://www.w3.org/TR/rdf-syntax-grammar.>..
> 
> <ex:a> <ex:b> <ex:c> .
> 
> and
> 
> _:xxx rdf:type rdf:Statement .
> _:xxx rdf:subject <ex:a> .
> _:xxx rdf:predicate <ex:b> .
> _:xxx rdf:object <ex:c> .
> 
> .... end quote"
> 
> is to create a blank node that represents ANY triple of the form
> 
>  <ex:a> <ex:b> <ex:c>
> 
> rather than the specific triple in my representation artefact.
> 
> I want to say things about the specific symbolic triple in a specific  
> resource.  I don't know if this is possible by referring to the full  
> URI of the rdf statement or not, but I can't see how as rdf triples  
> do not appear to have URIs, even though their subject, property, and  
> object all do.  I think what is being suggested is that OWL axioms,  
> or at least some OWL axioms, should.
> 
> Obviously, without some kind of partitioning, allowing a statement to  
> refer to another statement would allow a statement to refer to  
> itself, directly or indirectly, and  would bring with it the  
> paradoxes of self reference and therefore undecidability,  at least  
> as soon it was used with an expressive  enough semantics to include  
> negation or falsehood.
> 
> The simplest possible form of partitioning would seem to be to  
> declare that some properties are annotations and do not affect  
> inference - what I understood annotations did.  In fact, what I want  
> to do, almost always, is query the actual representation with  
> something like database (datalog) semantics, e.g. "Find all the  
> axioms in ontology X that were imported from ontology A"; "Find all  
> the statements in ontologies X,Y and Z asserted to be authored by  
> 'Rector'" .  "Find statements in ontology X for which there is no  
> assertion of authorship", etc.
> 
> This seems to follow naturally from Ian Horrocks' note on Databases  
> and Ontologies in which he says "DL KB doesn?t define a single model,  
> it is a set of constraints that define a set of possible models...In  
> contrast, DBs (and frame/rule KR systems) make assumptions such that  
> DB/KB defines a single model"  This is a case where I am clearly  
> interested in one possible model - the symbols and axioms actually in  
> my implementation of the OWL KB and how they got there, how they fit  
> together, etc.
> 
> There are a host of issues about copies and versions etc.,  but I  
> can't see trying to solve them until the fundamental principles are  
> clear. To start with at a minimum, I need to be able to send an  
> artefact along with the description of that artefact in terms of  
> provenance, usage, and editing annotations needed for somebody else  
> (who shares the same annotation vocabulary) to use it in the same  
> way, know where it came from, decide if they trust it, etc. (Note I  
> trust artefacts not the things the artefacts represent.).
> 
> As far as I can see,  The use of the same symbol as a class and  
> instance in OWL Full is a way of doing higher order representation,  
> although this is not explicitly clear in anything that I can find.   
> In which case the problems of fine grained annotation apply equally  
> to OWL full and OWL-DL.
> 
> In summary, my contention is that there are three distinct kinds of  
> statements about a) all instances of a class, b) the class itself,  
> and c) the symbolic representation of the class. Further that we need  
> to be able to refer to all constructs of the representation that can  
> be expressed individually including individual triples, axioms,  (and  
> the use of axioms to apply a restriction to a class) as well as  
> classes and individuals.   My concern is that b) and c)  are often  
> conflated. I would refer to b) as "higher order representation" and  
> c) as" annotation". This conflation is reflected in the common use of  
> the word "metadata" indiscrimately to refer to either or both.
> 
> I find it hard to see how we can resolve the issues around annotation  
> definitively without a clear distinction between these two notions.   
> If this is true, then there is a long term problem for which a best a  
> 'band aid' can be provided in OWL 1.1.
> 
> Or am I just completely confused?
> 
> Regards
> 
> Alan
> 
> 
> 
> 
> On 15 Dec 2005, at 13:37, Denny Vrandecic wrote:
> 
>> Hi Bijan,
>>
>> thanks, very good and important questions. I'll try to answer them,  
>> but I am happy if others would join in resolving them as well.
>>
>> First, let's ask a question in return: are axioms resources? Do we  
>> want to be *able* to speak about axioms? Is there anything  
>> interesting to say about axioms? If you say yes to any of these,  
>> they need an URI. It appears so totally wrong to me, that we  
>> (supposedly) can talk about anything and that we ask anyone to  
>> provide URIs for their stuff, so we can pull it into the Semantic  
>> Web - but we do not offer URIs for our own statements. I'm against  
>> a "You shall not annotate axioms"-law. Does anyone disagree on  
>> that? - and if so, why?
>>
>>> When is the URI being treated as an axiom?
>>
>> URIs of individuals, classes, properties... and axioms are all  
>> disjoint from each other, and thus, just as URIs of individuals,  
>> classes, properties... etc. axiom URIs need to be either declared  
>> or it must be inferrable from the context that this is an axiom URI.
>>
>>> Does the mere presence of an axiom URI in a KB put the axiom  
>>> itself there?
>>
>> I don't know. This would mean that the axiom must be known to the  
>> reasoner. And if it is known, isn't it already there? Hmm. Maybe  
>> not. I'd prefer a "no, it doesn't", but I am not sure about the  
>> implications of that.
>>
>>> Can I make any interesting constructions using the Axiom uri?
>>
>> I sure hope so, but what do you have in mind?
>>
>>> Should axiom equivalence entail URI equivalence?
>>
>> Yes. Because then I can merge ontologies and the annotations of  
>> axioms merge fittingly.
>>
>>> (Presumably not, eh, but what about through trivial rewritings,  
>>> i.e., different rdf serializations?)
>>
>> Huh, why not?
>>
>>> Are all axioms named by default, or must one coin a name to talk  
>>> about them?
>>
>> The latter.
>>
>>> Are annotations on axioms restricted to axioms with names/URIs?
>>
>> Can you make annotations on unnamed descriptions in OWL DL?
>>
>>> How do we avoid the pain that RDF Reification brings (and it's not  
>>> just the rdf:subject/predicate, etc. the Statement triple itself  
>>> and the introduction of the URI means CONSIDERABLE pain....note  
>>> the fact that no store, to my knowledge, that introduces  
>>> "contexts" does so by naming all the triples).
>>
>> I wish I had a convincing answer for that. What about "only  
>> annotated axioms actually need an URI"? Or "this is just enabling  
>> you to give an axiom an URI, it is not forcing you"? Or "hmm, I  
>> think we can hide the pain most of the time, by nicely extending  
>> the abstract syntax or the OWL XML presentation syntax. It will  
>> only remain really painful and ugly in the XML/RDF serialization"?
>>
>> As said, none of these answers really convince me, but maybe they  
>> are a starter for a solution. Or someone else has one. Maybe there  
>> is a way to give URIs to axioms without RDF reification, this would  
>> be great, and I'm just not seeing it.
>>
>> The question is, is avoiding this pain worth the restriction of  
>> disallowing to annotate axioms?
>>
>> Cheers,
>> denny
>> _______________________________________________
>> OWL mailing list
>> OWL@lists.mindswap.org
>> http://lists.mindswap.org/mailman/listinfo/owl
> 
> -----------------------
> Alan Rector
> Professor of Medical Informatics
> Department of Computer Science
> University of Manchester
> Manchester M13 9PL, UK
> TEL +44 (0) 161 275 6188/6149
> FAX +44 (0) 161 275 6204
> www.cs.man.ac.uk/mig
> www.clinical-esciences.org
> www.co-ode.org
> 
> _______________________________________________
> OWL mailing list
> OWL@lists.mindswap.org
> http://lists.mindswap.org/mailman/listinfo/owl
> 
>

From franconi at inf.unibz.it  Sun Jan  8 15:29:50 2006
From: franconi at inf.unibz.it (Enrico Franconi)
Date: Sun Jan  8 15:30:11 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <005101c6148e$df46aba0$d3d7858b@Newton>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
Message-ID: <6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>

On 8 Jan 2006, at 21:05, Jeff Z. Pan wrote:
> After reading Alan's following email and the proposed OWL 1.1  
> syntax [1], it seems to me that punning is not a convincing choice  
> for metamodeling in OWL 1.1. ( For those who are not familiar with  
> punning - punning means that a name, like Person, can be used as  
> both an individual and a class and a property.)

(...)

> 3.The semantics of punning is not quite intuitive. This can be  
> shown in the following example. In the following  OWL 1.1 [1]  
> ontology, Cat and Kitty are used as both classes and individuals.  
> Although Cat and Kitty are the same individual and Ted is a Cat,  
> the ontology
>
> Class (Cat partial)
> Class (Kitty partial)
> SameIndividual (Cat Kitty)
> Individual (Ted Cat)
>
> does not entail that Ted is also a Kitty. This distinguishes the  
> punning semantics from many other semantics, such as the OWL FA  
> semantics [2], Hilog semantics [3] and RDF semantics [4].

I'm not sure that the mentioned alternative to punning are really  
better:

- OWL-FA semantics does not exist yet: the last time I saw it was  
severely bugged; it would break wrt normative RDF; it is not  
compatible with punning.

- Hilog captures more expected inferences than punning but (a) Peter  
didn't say that it still leave some expected interesting inferences  
out (see Boris' paper for an example), and (b) it requires a re- 
implementation of the OWL reasoners, as Peter noticed.

- The only approach that would capture exactly meta-modelling with  
OWL-DL is undecidable (see Boris' paper) - this fact was hidden by  
Peter...

As we proved in [5], normative RDF semantics is completely equivalent  
to punning semantics, so there is no break wrt normative RDF. In [5],  
we give a complete account of punning semantics with OWL-DL and SPARQL.

cheers
--e.

[5] Jos de Bruijn, Enrico Franconi, Sergio Tessaris (2005). Logical  
Reconstruction of normative RDF. Proc. of the Workshosp on OWL  
Experiences and Directions (OWLED 2005), Galway, Ireland, November 2005.


From bparsia at isr.umd.edu  Sun Jan  8 15:54:47 2006
From: bparsia at isr.umd.edu (Bijan Parsia)
Date: Sun Jan  8 15:55:14 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
Message-ID: <428d71011b3c23eb1eddf878897035f2@isr.umd.edu>

On Jan 8, 2006, at 3:29 PM, Enrico Franconi wrote:

> On 8 Jan 2006, at 21:05, Jeff Z. Pan wrote:
[snip]
>> This distinguishes the punning semantics from many other semantics, 
>> such as the OWL FA semantics [2], Hilog semantics [3] and RDF 
>> semantics [4].
>
> I'm not sure that the mentioned alternative to punning are really 
> better:
[snip]
> - Hilog captures more expected inferences than punning but (a) Peter

s/Peter/Jeff/ ?

>  didn't say that it still leave some expected interesting inferences 
> out (see Boris' paper for an example), and (b) it requires a 
> re-implementation of the OWL reasoners, as Peter noticed.

Ibid?

> - The only approach that would capture exactly meta-modelling with 
> OWL-DL is undecidable (see Boris' paper) - this fact was hidden by 
> Peter...
[snip]

Ibid? Or am I missing an email from Peter?

 From various conversation with people who use OWL Full, and some 
introspection, I see two primary, if only current, uses of higher order 
like constructs (be they annotations, punning, or some more full blown 
species of metamodeling): Metadata about the "symbolic artefact", e.g., 
who wrote these axioms, when, when last modified, etc. and for ontology 
alignment (e.g., I modeled Wines as a class and you as an instance). I 
am not saying that these are the *only* uses of higher order like 
constructs, but they are *in my experience* what get mentioned. Only 
the latter has potentially interesting modeling impact, and, in 
practice, people are just happy to be able to *mark* these alignments 
and let some other piece of software (usually not a reasoner!) take 
care of, e.g., conversions of data between ontologies.

Since neither of these involves significant cross level entailments, it 
seems like some version of punning suffices conceived as a 
liberalization of annotation properites.

(This also has the advantage of typically only requiring modification 
of the input and output of tools, esp. reasoners.)

Now, I'm not sure that this is *ideal*. There may be some form of 
metamodeling with the legs to make it worth adopting even at the cost 
of revamping tooling. This doesn't seem to be 1.1 material though.

It would be nice if whatever we slide into 1.1 was neatly extendable to 
the Next Great Metamodeling Thing, but eh. We do what we can :) (And 
actually, most can *probably* by tying more interesting semantics to 
the presence of certain bridging axioms. This suggest we should *not* 
overload things like SameIndividual for alignment purposes, but instead 
introduce new sytnax (e.g., correspondsTo).)

Cheers,
Bijan.

From franconi at inf.unibz.it  Sun Jan  8 18:10:24 2006
From: franconi at inf.unibz.it (Enrico Franconi)
Date: Sun Jan  8 18:10:30 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <428d71011b3c23eb1eddf878897035f2@isr.umd.edu>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<428d71011b3c23eb1eddf878897035f2@isr.umd.edu>
Message-ID: <97FC81BA-D4E5-4414-825E-65B74C6DCAB3@inf.unibz.it>

On 8 Jan 2006, at 21:54, Bijan Parsia wrote:
> From various conversation with people who use OWL Full, and some  
> introspection, I see two primary, if only current, uses of higher  
> order like constructs (be they annotations, punning, or some more  
> full blown species of metamodeling): Metadata about the "symbolic  
> artefact", e.g., who wrote these axioms, when, when last modified,  
> etc. and for ontology alignment (e.g., I modeled Wines as a class  
> and you as an instance). I am not saying that these are the *only*  
> uses of higher order like constructs, but they are *in my  
> experience* what get mentioned. Only the latter has potentially  
> interesting modeling impact, and, in practice, people are just  
> happy to be able to *mark* these alignments and let some other  
> piece of software (usually not a reasoner!) take care of, e.g.,  
> conversions of data between ontologies.

Mmmhh, you are missing the *real* usages in the two biggest  
communities in informations system. In conceptual modelling, people  
do use metamodelling to characterise their object languages (I really  
don't like this, but this is a fact): see UML! In database systems,  
pleople use aggregation functions to characterise the values of  
properties of sets of tuples: see SQL! That is, you can have a  
property of a set of tuples defined in some way (e.g., average of the  
values of a property of all elements of the set of all tuples having  
some other property, etc).

--e.
From bparsia at isr.umd.edu  Sun Jan  8 21:37:08 2006
From: bparsia at isr.umd.edu (Bijan Parsia)
Date: Sun Jan  8 21:37:34 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <97FC81BA-D4E5-4414-825E-65B74C6DCAB3@inf.unibz.it>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<428d71011b3c23eb1eddf878897035f2@isr.umd.edu>
	<97FC81BA-D4E5-4414-825E-65B74C6DCAB3@inf.unibz.it>
Message-ID: <7a15980d679896c43d10c36f80a1516f@isr.umd.edu>

On Jan 8, 2006, at 6:10 PM, Enrico Franconi wrote:

> On 8 Jan 2006, at 21:54, Bijan Parsia wrote:
>> From various conversation with people who use OWL Full, and some 
>> introspection, I see two primary, if only current, uses of higher 
>> order like constructs (be they annotations, punning, or some more 
>> full blown species of metamodeling): Metadata about the "symbolic 
>> artefact", e.g., who wrote these axioms, when, when last modified, 
>> etc. and for ontology alignment (e.g., I modeled Wines as a class and 
>> you as an instance). I am not saying that these are the *only* uses 
>> of higher order like constructs, but they are *in my experience* what 
>> get mentioned. Only the latter has potentially interesting modeling 
>> impact, and, in practice, people are just happy to be able to *mark* 
>> these alignments and let some other piece of software (usually not a 
>> reasoner!) take care of, e.g., conversions of data between 
>> ontologies.
>
> Mmmhh, you are missing the *real* usages in the two biggest 
> communities in informations system.

That may be true, but I apparently don't talk to such people :)

> In conceptual modelling, people do use metamodelling to characterise 
> their object languages (I really don't like this, but this is a fact): 
> see UML!

This is true, but they aren't, in my experience, using OWL or OWL Full. 
That could change.

> In database systems, pleople use aggregation functions to characterise 
> the values of properties of sets of tuples: see SQL! That is, you can 
> have a property of a set of tuples defined in some way (e.g., average 
> of the values of a property of all elements of the set of all tuples 
> having some other property, etc).

Yep. But are these people wanting to use OWL for anything? UML maybe, 
though I've not seen *UMLers* step up and request this stuff. You're 
experience might vary. If there were a clear UML application (ICOMM? 
:)) that we were targeting as a kind of killer app for owl, I would 
find that example more compelling. But then I'd want a plan for 
sweeping the UML folks off their feet.

Or more simply, I don't know what the dominant set of users for robust 
metamodeling are thus have a hard time assessing the utility of 
standardizing various proposals. Annotations (with full punning) are 
ubiquitous and highly desired and (mostly) easy. So I find them 
reasonable for 1.1.

Cheers,
Bijan.

From pfps at research.bell-labs.com  Mon Jan  9 03:02:09 2006
From: pfps at research.bell-labs.com (Peter F. Patel-Schneider)
Date: Mon Jan  9 03:02:23 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <005101c6148e$df46aba0$d3d7858b@Newton>
References: <43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
Message-ID: <20060109.030209.64527517.pfps@research.bell-labs.com>

From: "Jeff Z. Pan" <jpan@csd.abdn.ac.uk>
Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1
Date: Sun, 8 Jan 2006 20:05:20 -0000

> Hi Peter, Alan and all,
> 
> After reading Alan's following email and the proposed OWL 1.1 syntax [1], it
> seems to me that punning is not a convincing choice for metamodeling in OWL
> 1.1. ( For those who are not familiar with punning - punning means that a name,
> like Person, can be used as both an individual and a class and a property.)

Punning may not be the ideal way to proceed.  However, it is an easy way to go,
and appears to be at least somewhat useful.

> 1. It is impossible to distinguish higher order statements from annotations of
> symbols and
> artefacts we are using to represent that domain, as pointed out in Alan's
> email. The reason that they are not distinguishable is because annotations in
> [1] are simply syntactic sugar of individual axioms.

Yes, but what proposal does distinguish between higher-order statements and
annotations of symbols?

> 2. Datatype axioms, unlike other axioms in OWL 1.1 [1], cannot have
> annotations. This seems pretty strange, at least to me. The reason is that
> although individuals, object properties and classes can share names, classes
> and datatypes cannot.

>From the OWL 1.1 syntax document [1], recapitulating the OWL DL syntax:

axiom ::= 'DatatypeProperty(' datavaluedPropertyID ['Deprecated'] { annotation } 
                { 'super(' datavaluedPropertyID ')'} ['Functional']
                { 'domain(' description ')' } { 'range(' dataRange ')' } ')'

This sure looks as if datatype axioms can have annotations.

It is true that classes and datatypes cannot share names.  This could be
changed, but I don't see that it has any impact on annotations for datatype
axioms.

Perhaps you are referring to a slight ambiguity (or bug) in the OWL 1.1
syntax.  The fix would be to be more explicit about the overloading of names,
perhaps something like:

	In OWL 1.1 a name can be used as a class or datatype as well as a
	property as well as an individual.

> 3.The semantics of punning is not quite intuitive. This can be shown in the
> following example. In the following OWL 1.1 [1] ontology, Cat and Kitty are
> used as both classes and individuals. Although Cat and Kitty are the same
> individual and Ted is a Cat, the ontology
> 
> Class (Cat partial)
> Class (Kitty partial)
> SameIndividual (Cat Kitty)
> Individual (Ted Cat)
> 
> does not entail that Ted is also a Kitty. This distinguishes the punning
> semantics from many other semantics, such as the OWL FA semantics [2], Hilog
> semantics [3] and RDF semantics [4].

I do agree that the punning semantics is weaker than other semantics.  This may
make it somewhat less useful than other semantics, but does have the "benefit"
that it could be strengthened at some later date.

I disagree, however, that the semantics of punning is unintuitive.  It is just
the situation of FOL.

> Summary: 
> 
> The above point 2 suggests that the use of punning cannot really eliminate the
> need for annotations properties. 

I don't think that this is the case, and it certainly isn't a necessary feature
of OWL 1.1.

> Point 1 provides a good reason to distinguish
> annotation properties from metamodeling. 

Well, yes, but what proposal addresses this?

> Point 3 suggests that punning (as an
> option for metamodeling in OWL 1.1) is not quite intuitive and can be
> misleading.

This can be argued either way.

> Greetings,
> Jeff
> 
> [1] http://www-db.research.bell-labs.com/user/pfps/owl/syntax.html
> [2] http://dl-web.man.ac.uk/~panz/Zhilin/pubc.php?type=epapers&id=PHS05
> [3]
> http://dip.semanticweb.org/documents/Boris-Motik-On-the-Properties-of-Metamodeling-in-OWL.pdf
> [4] http://www.w3.org/TR/rdf-mt/
> 
> --
> Dr. Jeff Z. Pan (http://www.csd.abdn.ac.uk/~jpan/)
> Department of Computing Science, The University of Aberdeen


peter
From franconi at inf.unibz.it  Mon Jan  9 03:55:43 2006
From: franconi at inf.unibz.it (Enrico Franconi)
Date: Mon Jan  9 03:55:51 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <7a15980d679896c43d10c36f80a1516f@isr.umd.edu>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<428d71011b3c23eb1eddf878897035f2@isr.umd.edu>
	<97FC81BA-D4E5-4414-825E-65B74C6DCAB3@inf.unibz.it>
	<7a15980d679896c43d10c36f80a1516f@isr.umd.edu>
Message-ID: <8460614E-8F29-45C5-B9CB-2A6E4BC5700E@inf.unibz.it>

On 9 Jan 2006, at 03:37, Bijan Parsia wrote:

> On Jan 8, 2006, at 6:10 PM, Enrico Franconi wrote:
>
>> In database systems, pleople use aggregation functions to  
>> characterise the values of properties of sets of tuples: see SQL!  
>> That is, you can have a property of a set of tuples defined in  
>> some way (e.g., average of the values of a property of all  
>> elements of the set of all tuples having some other property, etc).
>
> Yep. But are these people wanting to use OWL for anything? UML  
> maybe, though I've not seen *UMLers* step up and request this  
> stuff. You're experience might vary. If there were a clear UML  
> application (ICOM? :)) that we were targeting as a kind of killer  
> app for owl, I would find that example more compelling. But then  
> I'd want a plan for sweeping the UML folks off their feet.

Aggregation is ubiquitous in databases and SQL. It is not present in  
standard conceptual modelling languages (e.g., EntityRelationship E/R  
data model) for historical reasons (they came before SQL). It has a  
minor presence in UML (the association with a diamond at one end) -  
but UML was not conceived as a database conceptual modelling  
language. There is *plenty* of proposals for extension of the E/R  
data model (and for UML) to model conceptually aggregations, and all  
the conceptual modelling tools I know about have some (non-standard)  
hook to model aggregation. DL people (me, Uli, et al) did study  
extensions of DLs with aggregations, and I came up with an extension  
of E/R (and UML) which (a) is implemented in ICOM (the DL based  
conceptual modelling tool), and (b) deserved a chapter in a  
Datawarehouse design book :-).
I also mentioned aggregation since I'd say it is the typical use case  
for meta-modelling.
If OWL starts to catch up to model databases and more generally  
information integration (which we really hope!), then aggregation  
will become ubiquitous as well. I wonder whether people in the SWBP  
WG came up with such a modelling requisite.

> Or more simply, I don't know what the dominant set of users for  
> robust metamodeling are thus have a hard time assessing the utility  
> of standardizing various proposals. Annotations (with full punning)  
> are ubiquitous and highly desired and (mostly) easy. So I find them  
> reasonable for 1.1.

The reason why they are ubiquitous in OWL is because they are already  
as a partial feature in OWL.

cheers
--e.
From matthew.pocock at ncl.ac.uk  Mon Jan  9 06:30:33 2006
From: matthew.pocock at ncl.ac.uk (Matthew Pocock)
Date: Mon Jan  9 06:31:31 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <8460614E-8F29-45C5-B9CB-2A6E4BC5700E@inf.unibz.it>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu>
	<7a15980d679896c43d10c36f80a1516f@isr.umd.edu>
	<8460614E-8F29-45C5-B9CB-2A6E4BC5700E@inf.unibz.it>
Message-ID: <200601091130.33786.matthew.pocock@ncl.ac.uk>

On Monday 09 January 2006 08:55, Enrico Franconi wrote:
> Aggregation is ubiquitous in databases and SQL. 

Just to provide another data-point, in bioinformatics modelling, agregation is 
a continually usefull concept.

Matthew
From franconi at inf.unibz.it  Mon Jan  9 06:46:12 2006
From: franconi at inf.unibz.it (Enrico Franconi)
Date: Mon Jan  9 06:46:22 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <200601091130.33786.matthew.pocock@ncl.ac.uk>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu>
	<7a15980d679896c43d10c36f80a1516f@isr.umd.edu>
	<8460614E-8F29-45C5-B9CB-2A6E4BC5700E@inf.unibz.it>
	<200601091130.33786.matthew.pocock@ncl.ac.uk>
Message-ID: <CC5A093A-CAA3-4BA3-896E-8EC10BA830A3@inf.unibz.it>


On 9 Jan 2006, at 12:30, Matthew Pocock wrote:

> On Monday 09 January 2006 08:55, Enrico Franconi wrote:
>> Aggregation is ubiquitous in databases and SQL.
>
> Just to provide another data-point, in bioinformatics modelling,  
> agregation is
> a continually usefull concept.

Oh yeah: for example, all the granularity issues, the  
characterisation of pathways, etc...
--e.
From matthew.pocock at ncl.ac.uk  Mon Jan  9 06:55:48 2006
From: matthew.pocock at ncl.ac.uk (Matthew Pocock)
Date: Mon Jan  9 06:56:39 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <20060109.030209.64527517.pfps@research.bell-labs.com>
References: <43A1717E.5060207@aifb.uni-karlsruhe.de>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<20060109.030209.64527517.pfps@research.bell-labs.com>
Message-ID: <200601091155.49279.matthew.pocock@ncl.ac.uk>

So let's say I want to use OWL 1.1 meta-modelling facilities to model some 
realy simple generic data-structure constructor - let's choose a List (ala 
slingly-linked list, like every functional language has), with nill being the 
empty list. I've no idea of the syntax we may be using, so I hope this is 
intelligable. I've muddled up meta-level and object-level assertions, but I 
hope it's clear from context what my intention was.

To model a List type, I need four concepts:

Meta-List:
    1 x sameClassAs OWL:Class as LT
    1 x hasValueType OWL:Class as VT
    1 x hasPrevious LT
    1 x hasValue VT

So, we could have an instance:

class(PeopleList
  union(
    intersection(
      all previous PeopleList
      all person People),
    oneOf(nill)))

instance(PeopleList Meta-List)

Then, using OWL I should be able to

1) find that these declarations are consistent with the meta-model of a list
2) discover that People fills the role hasValueType, person fills the role 
hasValue and previous fills the role hasPrevious.

Is this in any way the sort of meta-modelling that OWL 1.1 is seeking to 
address? It would be nice if it can be done, as at this point we can do 
indexed types e.g. something like:

class(ListLength2
  Meta-List
  hasPrevious hasPrevious oneOf(nill) )

class(ListLength>2
  Meta-List
  hasPrevious hasPrevious LT)

Of course, the way I've written this is bad, because I've not realy 
distinguished between the properties in the meta-layer and object-layer, but 
I guess that's half the discussion we are having anyway. But, I could use 
this sort of expressivity now for a) flexible mappings of OWL data into 
end-user languages (sql, Java, ...) b) validation of systematic design of 
ontologies.

Matthew
From bparsia at isr.umd.edu  Mon Jan  9 08:12:54 2006
From: bparsia at isr.umd.edu (Bijan Parsia)
Date: Mon Jan  9 08:13:10 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <8460614E-8F29-45C5-B9CB-2A6E4BC5700E@inf.unibz.it>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<428d71011b3c23eb1eddf878897035f2@isr.umd.edu>
	<97FC81BA-D4E5-4414-825E-65B74C6DCAB3@inf.unibz.it>
	<7a15980d679896c43d10c36f80a1516f@isr.umd.edu>
	<8460614E-8F29-45C5-B9CB-2A6E4BC5700E@inf.unibz.it>
Message-ID: <3d31cb13c6b49edf648da1726c5d3522@isr.umd.edu>

On Jan 9, 2006, at 3:55 AM, Enrico Franconi wrote:

> On 9 Jan 2006, at 03:37, Bijan Parsia wrote:
>
>> On Jan 8, 2006, at 6:10 PM, Enrico Franconi wrote:
>>
>>> In database systems, pleople use aggregation functions to 
>>> characterise the values of properties of sets of tuples: see SQL! 
>>> That is, you can have a property of a set of tuples defined in some 
>>> way (e.g., average of the values of a property of all elements of 
>>> the set of all tuples having some other property, etc).
>>
>> Yep. But are these people wanting to use OWL for anything? UML maybe, 
>> though I've not seen *UMLers* step up and request this stuff. You're 
>> experience might vary. If there were a clear UML application (ICOM? 
>> :)) that we were targeting as a kind of killer app for owl, I would 
>> find that example more compelling. But then I'd want a plan for 
>> sweeping the UML folks off their feet.
>
> Aggregation is ubiquitous in databases and SQL.

We aren't disagreeing.

> It is not present in standard conceptual modelling languages (e.g., 
> EntityRelationship E/R data model) for historical reasons (they came 
> before SQL). It has a minor presence in UML (the association with a 
> diamond at one end) - but UML was not conceived as a database 
> conceptual modelling language. There is *plenty* of proposals for 
> extension of the E/R data model (and for UML) to model conceptually 
> aggregations, and all the conceptual modelling tools I know about have 
> some (non-standard) hook to model aggregation.

With no clear winner. We're not talking overall utility or 
significance, but what has the right balance between utility, ease of 
marketing, and easy implementability. For *1.1*, treading where there 
is non-standardness elsewhere seems a little off target. Plus, there is 
no proposal on the table.

(This seems critical to OWL 2.0 as is a more elaborate form of 
metamodeling, but it doesn't seem to be a nigh trivial win.)

>  DL people (me, Uli, et al) did study extensions of DLs with 
> aggregations, and I came up with an extension of E/R (and UML) which 
> (a) is implemented in ICOM (the DL based conceptual modelling tool), 
> and (b) deserved a chapter in a Datawarehouse design book :-).

I think ICOM is a wonderful tool, but what I *don't* see is that it, or 
things like it, are widely used. My *personal* strategy would be to 
revive ICOM and try to build a user base for it (or get one of the 
vendors to sell it).

> I also mentioned aggregation since I'd say it is the typical use case 
> for meta-modelling.
> If OWL starts to catch up to model databases and more generally 
> information integration (which we really hope!), then aggregation will 
> become ubiquitous as well.

I agree.

>  I wonder whether people in the SWBP WG came up with such a modelling 
> requisite.

Good question.

>> Or more simply, I don't know what the dominant set of users for 
>> robust metamodeling are thus have a hard time assessing the utility 
>> of standardizing various proposals. Annotations (with full punning) 
>> are ubiquitous and highly desired and (mostly) easy. So I find them 
>> reasonable for 1.1.
>
> The reason why they are ubiquitous in OWL is because they are already 
> as a partial feature in OWL.

Let me totally concede that....let me *stipulate* that (though I've 
never said or thought otherwise). How is this a point against me in 
this dialectic? All I can see is that you could claim that aggregates 
would be equally ubiquitous if only they were present even if in a 
partial form. Let's stipulate *that*, that doesn't establish 1) that it 
will be easy to articulate and decide between proposals, 2) implement 
in all the major reasoners and editors (remember, we're hoping to have 
all of OWL 1.1 well
on the way to ubiquitously implemented by *May*!).

Cheers,
Bijan.

From franconi at inf.unibz.it  Mon Jan  9 09:07:14 2006
From: franconi at inf.unibz.it (Enrico Franconi)
Date: Mon Jan  9 09:07:24 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <3d31cb13c6b49edf648da1726c5d3522@isr.umd.edu>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<428d71011b3c23eb1eddf878897035f2@isr.umd.edu>
	<97FC81BA-D4E5-4414-825E-65B74C6DCAB3@inf.unibz.it>
	<7a15980d679896c43d10c36f80a1516f@isr.umd.edu>
	<8460614E-8F29-45C5-B9CB-2A6E4BC5700E@inf.unibz.it>
	<3d31cb13c6b49edf648da1726c5d3522@isr.umd.edu>
Message-ID: <2530439F-5CCA-4323-9BD7-A114F39FAB4B@inf.unibz.it>

On 9 Jan 2006, at 14:12, Bijan Parsia wrote:
>> It is not present in standard conceptual modelling languages  
>> (e.g., EntityRelationship E/R data model) for historical reasons  
>> (they came before SQL). It has a minor presence in UML (the  
>> association with a diamond at one end) - but UML was not conceived  
>> as a database conceptual modelling language. There is *plenty* of  
>> proposals for extension of the E/R data model (and for UML) to  
>> model conceptually aggregations, and all the conceptual modelling  
>> tools I know about have some (non-standard) hook to model  
>> aggregation.
>
> With no clear winner. We're not talking overall utility or  
> significance, but what has the right balance between utility, ease  
> of marketing, and easy implementability. For *1.1*, treading where  
> there is non-standardness elsewhere seems a little off target.  
> Plus, there is no proposal on the table.
>
> (This seems critical to OWL 2.0 as is a more elaborate form of  
> metamodeling, but it doesn't seem to be a nigh trivial win.)

I'd like to emphasise my point of view:

1) aggregation as meta-modelling facility is ubiquitous and very  
relevant for conceptual modelling.

2) Nonetheless, I believe that OWL-DL is not at the stage of devising  
a *standard* supporting properly meta-modelling for aggregation,  
since there has been little discussion about that. My favourite  
proposal would stem (of course!) from [1] (top hit in <http:// 
scholar.google.com/scholar?q=franconi%20sattler>, with 34 citations),  
already implemented in ICOM.

3) If an extension to OWL-DL supporting aggregation will ever see the  
light, I'm sure that there will be a widespread use of it.

> I think ICOM is a wonderful tool, but what I *don't* see is that  
> it, or things like it, are widely used. My *personal* strategy  
> would be to revive ICOM and try to build a user base for it (or get  
> one of the vendors to sell it).

We are coming out with a wonderful new release (how many times have I  
already said that?:-))
Compatible with OWL etc. But, we killed the aggregation part - lack  
of time :-(

>>  I wonder whether people in the SWBP WG came up with such a  
>> modelling requisite.
>
> Good question.

Ehi: somebody from SWBP out there?

cheers
--e.

[1] Enrico Franconi and Ulrike Sattler (1999). A Data Warehouse  
Conceptual Data Model for Multidimensional Aggregation. Workshop on  
Design and Management of Data Warehouses (DMDW'99), Heidelberg,  
Germany, June 1999.

From bparsia at isr.umd.edu  Mon Jan  9 09:18:02 2006
From: bparsia at isr.umd.edu (Bijan Parsia)
Date: Mon Jan  9 09:18:14 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <2530439F-5CCA-4323-9BD7-A114F39FAB4B@inf.unibz.it>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<428d71011b3c23eb1eddf878897035f2@isr.umd.edu>
	<97FC81BA-D4E5-4414-825E-65B74C6DCAB3@inf.unibz.it>
	<7a15980d679896c43d10c36f80a1516f@isr.umd.edu>
	<8460614E-8F29-45C5-B9CB-2A6E4BC5700E@inf.unibz.it>
	<3d31cb13c6b49edf648da1726c5d3522@isr.umd.edu>
	<2530439F-5CCA-4323-9BD7-A114F39FAB4B@inf.unibz.it>
Message-ID: <6486c9d6212497cd4300dc900d90ce46@isr.umd.edu>

On Jan 9, 2006, at 9:07 AM, Enrico Franconi wrote:

> On 9 Jan 2006, at 14:12, Bijan Parsia wrote:
>>> It is not present in standard conceptual modelling languages (e.g., 
>>> EntityRelationship E/R data model) for historical reasons (they came 
>>> before SQL). It has a minor presence in UML (the association with a 
>>> diamond at one end) - but UML was not conceived as a database 
>>> conceptual modelling language. There is *plenty* of proposals for 
>>> extension of the E/R data model (and for UML) to model conceptually 
>>> aggregations, and all the conceptual modelling tools I know about 
>>> have some (non-standard) hook to model aggregation.
>>
>> With no clear winner. We're not talking overall utility or 
>> significance, but what has the right balance between utility, ease of 
>> marketing, and easy implementability. For *1.1*, treading where there 
>> is non-standardness elsewhere seems a little off target. Plus, there 
>> is no proposal on the table.
>>
>> (This seems critical to OWL 2.0 as is a more elaborate form of 
>> metamodeling, but it doesn't seem to be a nigh trivial win.)
>
> I'd like to emphasise my point of view:
>
> 1) aggregation as meta-modelling facility is ubiquitous and very 
> relevant for conceptual modelling.
>
> 2) Nonetheless, I believe that OWL-DL is not at the stage of devising 
> a *standard* supporting properly meta-modelling for aggregation, since 
> there has been little discussion about that. My favourite proposal 
> would stem (of course!) from [1] (top hit in 
> <http://scholar.google.com/scholar?q=franconi%20sattler>, with 34 
> citations), already implemented in ICOM.
>
> 3) If an extension to OWL-DL supporting aggregation will ever see the 
> light, I'm sure that there will be a widespread use of it.

Sounds like with completely agree.

>> I think ICOM is a wonderful tool, but what I *don't* see is that it, 
>> or things like it, are widely used. My *personal* strategy would be 
>> to revive ICOM and try to build a user base for it (or get one of the 
>> vendors to sell it).
>
> We are coming out with a wonderful new release (how many times have I 
> already said that?:-))

Heh. Cool. I'd love to see it.

> Compatible with OWL etc. But, we killed the aggregation part - lack of 
> time :-(

That seems to be dispositive. (True for SPARQL too.)

>>>  I wonder whether people in the SWBP WG came up with such a 
>>> modelling requisite.
>>
>> Good question.
>
> Ehi: somebody from SWBP out there?

Going over their home page, I don't see anything. It might be worth 
putting together an XG on aggregation and metamodeling for the semantic 
web. I.e., something with a bit more structure charged with coming up 
with a consensus set of proposals.

Some of this will perhaps be addressed in the RIF (or punted there, 
again).

Cheers,
Bijan.

From jpan at csd.abdn.ac.uk  Mon Jan  9 09:45:08 2006
From: jpan at csd.abdn.ac.uk (Jeff Z. Pan)
Date: Mon Jan  9 09:46:40 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
Message-ID: <02fd01c6152b$73db3280$d3d7858b@Newton>


"Enrico Franconi" <franconi@inf.unibz.it> wrote: 

> On 8 Jan 2006, at 21:05, Jeff Z. Pan wrote:
>> After reading Alan's following email and the proposed OWL 1.1  
>> syntax [1], it seems to me that punning is not a convincing choice  
>> for metamodeling in OWL 1.1. ( For those who are not familiar with  
>> punning - punning means that a name, like Person, can be used as  
>> both an individual and a class and a property.)
> 
> (...)
> 
>> 3.The semantics of punning is not quite intuitive. This can be  
>> shown in the following example. In the following  OWL 1.1 [1]  
>> ontology, Cat and Kitty are used as both classes and individuals.  
>> Although Cat and Kitty are the same individual and Ted is a Cat,  
>> the ontology
>>
>> Class (Cat partial)
>> Class (Kitty partial)
>> SameIndividual (Cat Kitty)
>> Individual (Ted Cat)
>>
>> does not entail that Ted is also a Kitty. This distinguishes the  
>> punning semantics from many other semantics, such as the OWL FA  
>> semantics [2], Hilog semantics [3] and RDF semantics [4].
> 
> I'm not sure that the mentioned alternative to punning are really  
> better:
> 
> - OWL-FA semantics does not exist yet: the last time I saw it was  
> severely bugged; it would break wrt normative RDF; it is not  
> compatible with punning.

Do you mean "being not compatible with punning" is a severe bug of [2]? 


> - Hilog captures more expected inferences than punning but (a) Peter  
> didn't say that it still leave some expected interesting inferences  
> out (see Boris' paper for an example), 

Could you be more specific on what kinds of expected interesting inferences are missing in the Hilog semantics?

>and (b) it requires a re- 
> implementation of the OWL reasoners, as Peter noticed.

> - The only approach that would capture exactly meta-modelling with  
> OWL-DL is undecidable (see Boris' paper) - this fact was hidden by  
> Peter...

Which (only) approach do you mean? As far as I understand, Boris' paper [3] propose two approaches: the context (punning) approach and the Hilog approach. 
 
> As we proved in [5], normative RDF semantics is completely equivalent  
> to punning semantics, so there is no break wrt normative RDF. In [5],  
> we give a complete account of punning semantics with OWL-DL and SPARQL.

OK - I will give it a check.

Cheers,
Jeff

[1] http://www-db.research.bell-labs.com/user/pfps/owl/syntax.html
[2] http://www.mindswap.org/2005/OWLWorkshop/sub15.pdf
[3] http://dip.semanticweb.org/documents/Boris-Motik-On-the-Properties-of-Metamodeling-in-OWL.pdf
[4] http://www.w3.org/TR/rdf-mt/
> [5] Jos de Bruijn, Enrico Franconi, Sergio Tessaris (2005). Logical  
> Reconstruction of normative RDF. Proc. of the Workshosp on OWL  
> Experiences and Directions (OWLED 2005), Galway, Ireland, November 2005.

--
Dr. Jeff Z. Pan (http://www.csd.abdn.ac.uk/~jpan/)
Department of Computing Science, The University of Aberdeen


From franconi at inf.unibz.it  Mon Jan  9 10:00:49 2006
From: franconi at inf.unibz.it (Enrico Franconi)
Date: Mon Jan  9 10:00:59 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <02fd01c6152b$73db3280$d3d7858b@Newton>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<02fd01c6152b$73db3280$d3d7858b@Newton>
Message-ID: <1C210251-1BA2-4752-A501-7C465E71E1E0@inf.unibz.it>


On 9 Jan 2006, at 15:45, Jeff Z. Pan wrote:
>> - OWL-FA semantics does not exist yet: the last time I saw it was
>> severely bugged; it would break wrt normative RDF; it is not
>> compatible with punning.
>
> Do you mean "being not compatible with punning" is a severe bug of  
> [2]?

No, it's just an additional problem for me.

>> - The only approach that would capture exactly meta-modelling with
>> OWL-DL is undecidable (see Boris' paper) - this fact was hidden by
>> Peter...
>
> Which (only) approach do you mean? As far as I understand, Boris'  
> paper [3] propose two approaches: the context (punning) approach  
> and the Hilog approach.

No, the very first part of the paper is about undecidability of the  
"real" approach.

--e.

From franconi at inf.unibz.it  Mon Jan  9 09:07:14 2006
From: franconi at inf.unibz.it (Enrico Franconi)
Date: Mon Jan  9 17:00:04 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <3d31cb13c6b49edf648da1726c5d3522@isr.umd.edu>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<428d71011b3c23eb1eddf878897035f2@isr.umd.edu>
	<97FC81BA-D4E5-4414-825E-65B74C6DCAB3@inf.unibz.it>
	<7a15980d679896c43d10c36f80a1516f@isr.umd.edu>
	<8460614E-8F29-45C5-B9CB-2A6E4BC5700E@inf.unibz.it>
	<3d31cb13c6b49edf648da1726c5d3522@isr.umd.edu>
Message-ID: <2530439F-5CCA-4323-9BD7-A114F39FAB4B@inf.unibz.it>


On 9 Jan 2006, at 14:12, Bijan Parsia wrote:
>> It is not present in standard conceptual modelling languages  
>> (e.g., EntityRelationship E/R data model) for historical reasons  
>> (they came before SQL). It has a minor presence in UML (the  
>> association with a diamond at one end) - but UML was not conceived  
>> as a database conceptual modelling language. There is *plenty* of  
>> proposals for extension of the E/R data model (and for UML) to  
>> model conceptually aggregations, and all the conceptual modelling  
>> tools I know about have some (non-standard) hook to model  
>> aggregation.
>
> With no clear winner. We're not talking overall utility or  
> significance, but what has the right balance between utility, ease  
> of marketing, and easy implementability. For *1.1*, treading where  
> there is non-standardness elsewhere seems a little off target.  
> Plus, there is no proposal on the table.
>
> (This seems critical to OWL 2.0 as is a more elaborate form of  
> metamodeling, but it doesn't seem to be a nigh trivial win.)

I'd like to emphasise my point of view:

1) aggregation as meta-modelling facility is ubiquitous and very  
relevant for conceptual modelling.

2) Nonetheless, I believe that OWL-DL is not at the stage of devising  
a *standard* supporting properly meta-modelling for aggregation,  
since there has been little discussion about that. My favourite  
proposal would stem (of course!) from [1] (top hit in <http:// 
scholar.google.com/scholar?q=franconi%20sattler>, with 34 citations),  
already implemented in ICOM.

3) If an extension to OWL-DL supporting aggregation will ever see the  
light, I'm sure that there will be a widespread use of it.

> I think ICOM is a wonderful tool, but what I *don't* see is that  
> it, or things like it, are widely used. My *personal* strategy  
> would be to revive ICOM and try to build a user base for it (or get  
> one of the vendors to sell it).

We are coming out with a wonderful new release (how many times have I  
already said that?:-))
Compatible with OWL etc. But, we killed the aggregation part - lack  
of time :-(

>>  I wonder whether people in the SWBP WG came up with such a  
>> modelling requisite.
>
> Good question.

Ehi: somebody from SWBP out there?

cheers
--e.

[1] Enrico Franconi and Ulrike Sattler (1999). A Data Warehouse  
Conceptual Data Model for Multidimensional Aggregation. Workshop on  
Design and Management of Data Warehouses (DMDW'99), Heidelberg,  
Germany, June 1999.


From bparsia at isr.umd.edu  Mon Jan  9 09:18:02 2006
From: bparsia at isr.umd.edu (Bijan Parsia)
Date: Mon Jan  9 17:38:15 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <2530439F-5CCA-4323-9BD7-A114F39FAB4B@inf.unibz.it>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<428d71011b3c23eb1eddf878897035f2@isr.umd.edu>
	<97FC81BA-D4E5-4414-825E-65B74C6DCAB3@inf.unibz.it>
	<7a15980d679896c43d10c36f80a1516f@isr.umd.edu>
	<8460614E-8F29-45C5-B9CB-2A6E4BC5700E@inf.unibz.it>
	<3d31cb13c6b49edf648da1726c5d3522@isr.umd.edu>
	<2530439F-5CCA-4323-9BD7-A114F39FAB4B@inf.unibz.it>
Message-ID: <6486c9d6212497cd4300dc900d90ce46@isr.umd.edu>


On Jan 9, 2006, at 9:07 AM, Enrico Franconi wrote:

> On 9 Jan 2006, at 14:12, Bijan Parsia wrote:
>>> It is not present in standard conceptual modelling languages (e.g., 
>>> EntityRelationship E/R data model) for historical reasons (they came 
>>> before SQL). It has a minor presence in UML (the association with a 
>>> diamond at one end) - but UML was not conceived as a database 
>>> conceptual modelling language. There is *plenty* of proposals for 
>>> extension of the E/R data model (and for UML) to model conceptually 
>>> aggregations, and all the conceptual modelling tools I know about 
>>> have some (non-standard) hook to model aggregation.
>>
>> With no clear winner. We're not talking overall utility or 
>> significance, but what has the right balance between utility, ease of 
>> marketing, and easy implementability. For *1.1*, treading where there 
>> is non-standardness elsewhere seems a little off target. Plus, there 
>> is no proposal on the table.
>>
>> (This seems critical to OWL 2.0 as is a more elaborate form of 
>> metamodeling, but it doesn't seem to be a nigh trivial win.)
>
> I'd like to emphasise my point of view:
>
> 1) aggregation as meta-modelling facility is ubiquitous and very 
> relevant for conceptual modelling.
>
> 2) Nonetheless, I believe that OWL-DL is not at the stage of devising 
> a *standard* supporting properly meta-modelling for aggregation, since 
> there has been little discussion about that. My favourite proposal 
> would stem (of course!) from [1] (top hit in 
> <http://scholar.google.com/scholar?q=franconi%20sattler>, with 34 
> citations), already implemented in ICOM.
>
> 3) If an extension to OWL-DL supporting aggregation will ever see the 
> light, I'm sure that there will be a widespread use of it.

Sounds like with completely agree.

>> I think ICOM is a wonderful tool, but what I *don't* see is that it, 
>> or things like it, are widely used. My *personal* strategy would be 
>> to revive ICOM and try to build a user base for it (or get one of the 
>> vendors to sell it).
>
> We are coming out with a wonderful new release (how many times have I 
> already said that?:-))

Heh. Cool. I'd love to see it.

> Compatible with OWL etc. But, we killed the aggregation part - lack of 
> time :-(

That seems to be dispositive. (True for SPARQL too.)

>>>  I wonder whether people in the SWBP WG came up with such a 
>>> modelling requisite.
>>
>> Good question.
>
> Ehi: somebody from SWBP out there?

Going over their home page, I don't see anything. It might be worth 
putting together an XG on aggregation and metamodeling for the semantic 
web. I.e., something with a bit more structure charged with coming up 
with a consensus set of proposals.

Some of this will perhaps be addressed in the RIF (or punted there, 
again).

Cheers,
Bijan.


From jpan at csd.abdn.ac.uk  Mon Jan  9 09:45:08 2006
From: jpan at csd.abdn.ac.uk (Jeff Z. Pan)
Date: Mon Jan  9 19:38:25 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
Message-ID: <02fd01c6152b$73db3280$d3d7858b@Newton>



"Enrico Franconi" <franconi@inf.unibz.it> wrote: 

> On 8 Jan 2006, at 21:05, Jeff Z. Pan wrote:
>> After reading Alan's following email and the proposed OWL 1.1  
>> syntax [1], it seems to me that punning is not a convincing choice  
>> for metamodeling in OWL 1.1. ( For those who are not familiar with  
>> punning - punning means that a name, like Person, can be used as  
>> both an individual and a class and a property.)
> 
> (...)
> 
>> 3.The semantics of punning is not quite intuitive. This can be  
>> shown in the following example. In the following  OWL 1.1 [1]  
>> ontology, Cat and Kitty are used as both classes and individuals.  
>> Although Cat and Kitty are the same individual and Ted is a Cat,  
>> the ontology
>>
>> Class (Cat partial)
>> Class (Kitty partial)
>> SameIndividual (Cat Kitty)
>> Individual (Ted Cat)
>>
>> does not entail that Ted is also a Kitty. This distinguishes the  
>> punning semantics from many other semantics, such as the OWL FA  
>> semantics [2], Hilog semantics [3] and RDF semantics [4].
> 
> I'm not sure that the mentioned alternative to punning are really  
> better:
> 
> - OWL-FA semantics does not exist yet: the last time I saw it was  
> severely bugged; it would break wrt normative RDF; it is not  
> compatible with punning.

Do you mean "being not compatible with punning" is a severe bug of [2]? 


> - Hilog captures more expected inferences than punning but (a) Peter  
> didn't say that it still leave some expected interesting inferences  
> out (see Boris' paper for an example), 

Could you be more specific on what kinds of expected interesting inferences are missing in the Hilog semantics?

>and (b) it requires a re- 
> implementation of the OWL reasoners, as Peter noticed.

> - The only approach that would capture exactly meta-modelling with  
> OWL-DL is undecidable (see Boris' paper) - this fact was hidden by  
> Peter...

Which (only) approach do you mean? As far as I understand, Boris' paper [3] propose two approaches: the context (punning) approach and the Hilog approach. 
 
> As we proved in [5], normative RDF semantics is completely equivalent  
> to punning semantics, so there is no break wrt normative RDF. In [5],  
> we give a complete account of punning semantics with OWL-DL and SPARQL.

OK - I will give it a check.

Cheers,
Jeff

[1] http://www-db.research.bell-labs.com/user/pfps/owl/syntax.html
[2] http://www.mindswap.org/2005/OWLWorkshop/sub15.pdf
[3] http://dip.semanticweb.org/documents/Boris-Motik-On-the-Properties-of-Metamodeling-in-OWL.pdf
[4] http://www.w3.org/TR/rdf-mt/
> [5] Jos de Bruijn, Enrico Franconi, Sergio Tessaris (2005). Logical  
> Reconstruction of normative RDF. Proc. of the Workshosp on OWL  
> Experiences and Directions (OWLED 2005), Galway, Ireland, November 2005.

--
Dr. Jeff Z. Pan (http://www.csd.abdn.ac.uk/~jpan/)
Department of Computing Science, The University of Aberdeen



From jpan at csd.abdn.ac.uk  Tue Jan 10 04:26:32 2006
From: jpan at csd.abdn.ac.uk (jpan@csd.abdn.ac.uk)
Date: Tue Jan 10 04:27:03 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <20060109.030209.64527517.pfps@research.bell-labs.com>
References: <43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<20060109.030209.64527517.pfps@research.bell-labs.com>
Message-ID: <20060110092632.uoj32vc5hg004o40@www.csd.abdn.ac.uk>


"Peter F. Patel-Schneider" <pfps@research.bell-labs.com> wrote:

> From: "Jeff Z. Pan" <jpan@csd.abdn.ac.uk>
> Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1
> Date: Sun, 8 Jan 2006 20:05:20 -0000
>
>> Hi Peter, Alan and all,
>>
>> After reading Alan's following email and the proposed OWL 1.1 syntax [1], it
>> seems to me that punning is not a convincing choice for metamodeling in OWL
>> 1.1. ( For those who are not familiar with punning - punning means 
>> that a name,
>> like Person, can be used as both an individual and a class and a property.)
>
> Punning may not be the ideal way to proceed.  However, it is an easy 
> way to go,
> and appears to be at least somewhat useful.

I agree that it is an easy way to go, although I don't see how useful it is.


>> 1. It is impossible to distinguish higher order statements from 
>> annotations of
>> symbols and
>> artefacts we are using to represent that domain, as pointed out in Alan's
>> email. The reason that they are not distinguishable is because 
>> annotations in
>> [1] are simply syntactic sugar of individual axioms.
>
> Yes,

OK.

> but what proposal does distinguish between higher-order statements and
> annotations of symbols?

Higher-order statements (axioms about meta-classes and meta-properties) and
annotations (in the sense of OWL DL) are two seperate things, I don't
understand why we cannot distinguish them.


>> 2. Datatype axioms, unlike other axioms in OWL 1.1 [1], cannot have
>> annotations. This seems pretty strange, at least to me. The reason is that
>> although individuals, object properties and classes can share names, classes
>> and datatypes cannot.
>
>> From the OWL 1.1 syntax document [1], recapitulating the OWL DL syntax:
>
> axiom ::= 'DatatypeProperty(' datavaluedPropertyID ['Deprecated'] { 
> annotation }                { 'super(' datavaluedPropertyID ')'} 
> ['Functional']
>                { 'domain(' description ')' } { 'range(' dataRange ')' } ')'
>
> This sure looks as if datatype axioms can have annotations.

What I meant was

axiom ::= 'Datatype(' datatypeID 'base(' datatypeID ')' { 
datatypeRestriction }
')'.

We don't have annotations here, don't we?

Cheers,
Jeff

--Dr. Jeff Z. Pan (http://www.csd.abdn.ac.uk/~jpan/)Department of Computing
Science, The University of Aberdeen


From pfps at research.bell-labs.com  Tue Jan 10 06:43:56 2006
From: pfps at research.bell-labs.com (Peter F. Patel-Schneider)
Date: Tue Jan 10 06:44:02 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <20060110092632.uoj32vc5hg004o40@www.csd.abdn.ac.uk>
References: <005101c6148e$df46aba0$d3d7858b@Newton>
	<20060109.030209.64527517.pfps@research.bell-labs.com>
	<20060110092632.uoj32vc5hg004o40@www.csd.abdn.ac.uk>
Message-ID: <20060110.064356.131777543.pfps@research.bell-labs.com>

From: jpan@csd.abdn.ac.uk
Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1
Date: Tue, 10 Jan 2006 09:26:32 +0000

> 
> "Peter F. Patel-Schneider" <pfps@research.bell-labs.com> wrote:
> 
> > From: "Jeff Z. Pan" <jpan@csd.abdn.ac.uk>
> > Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1
> > Date: Sun, 8 Jan 2006 20:05:20 -0000
> >
> >> Hi Peter, Alan and all,

[...]

> >> 1. It is impossible to distinguish higher order statements from 
> >> annotations of
> >> symbols and
> >> artefacts we are using to represent that domain, as pointed out in Alan's
> >> email. The reason that they are not distinguishable is because 
> >> annotations in
> >> [1] are simply syntactic sugar of individual axioms.
> >
> > Yes,
> 
> OK.
> 
> > but what proposal does distinguish between higher-order statements and
> > annotations of symbols?
> 
> Higher-order statements (axioms about meta-classes and meta-properties) and
> annotations (in the sense of OWL DL) are two seperate things, I don't
> understand why we cannot distinguish them.

Yes, it would be possible, but what proposal on the table does currently have
both of them, distinguished?

> >> 2. Datatype axioms, unlike other axioms in OWL 1.1 [1], cannot have
> >> annotations. This seems pretty strange, at least to me. The reason is that
> >> although individuals, object properties and classes can share names, classes
> >> and datatypes cannot.
> >
> >> From the OWL 1.1 syntax document [1], recapitulating the OWL DL syntax:
> >
> > axiom ::= 'DatatypeProperty(' datavaluedPropertyID ['Deprecated'] { 
> > annotation }                { 'super(' datavaluedPropertyID ')'} 
> > ['Functional']
> >                { 'domain(' description ')' } { 'range(' dataRange ')' } ')'
> >
> > This sure looks as if datatype axioms can have annotations.
> 
> What I meant was
> 
> axiom ::= 'Datatype(' datatypeID 'base(' datatypeID ')' { 
> datatypeRestriction }
> ')'.
> 
> We don't have annotations here, don't we?

That is a bug, which will be fixed immediately.

> Cheers,
> Jeff

peter
From Christine.Golbreich at univ-rennes1.fr  Wed Jan 11 06:27:24 2006
From: Christine.Golbreich at univ-rennes1.fr (Christine Golbreich)
Date: Wed Jan 11 06:27:30 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
Message-ID: <007b01c616a1$ff6a9a80$db9ffea9@LUZ>

I apologize for my delay in reacting about the OWL1.1 metamodelling and
annotation issue and for any misunderstanding or mistake. Sorry, but I
cannot follow and read all emails (not the last ones) nor all related
papers neither (I did not read OWL-FA). Having  no time to elaborate more,
please find a few provisary comments.

- First, I prefer to distinguish metadata (previously represented using
OWL annotation) from metamodelling.

Taking Motik's Eagle example (from C. Welty) we can distinguish:

1. Eagle a class of individuals represented in OWL by the class Eagle,
subClass of Bird
2. Metadata about the class Eagle would say that class Eagle was
created by "C. Welty" in 1994, ID  ... :
creator: C. Welty, year: 1994
3. The specy EagleSpecy, later renamed Eagle, an individual of the
'metaconcept' RedListSpecies, having a different intrepretation from the
class Eagle when breaking the connection between classes and individuals
that share a name

* Metadata
metadata were defined in "traditional" OWL DL using :

1?)  comments
 no axioms, no inference at all
2?) or OWL annotations
limitations related to OWL-DL as recalled by Peter, which in particular
prevents some axioms domain, range,  functional etc.

Such limitations are problematic. The main needs I met is a means to assert
the unicity of some metadata, in particular to assert that a concept
identifier is unique (e.g., UWDAID, UMLS-ID etc.) or unicity mapping for
ontology and terminology alignment. However, I am not sure the same holds
for axioms. It seems that comments can be enough for asserting  who wrote
these axioms, when, when last modified. I am not sure why it would be
important to have more, and for what inferences.

The OWL 1.1  proposal with punning + syntactic sugar
-------------------------------------------------------------
In OWL1.1 proposal: "A class or property axiom with an 'annotation' is
syntactic sugar for an extra Individual axiom for the name with just the
annotations".

If I understand it correctly (?)
- the term 'annotation'  here must now be understood as any (meta level)
property p which is assigned a value v:  p(C,v) where C is a class.
I am not sure I understood it correctely, but I thought annotationProperty
has disappeared (?)
- the punning + syntactic sugar solution is an elegant trick to remain in
FOL, assuming  in such a case the existence of an individual of same name C
(for example of  a class ConceptName), and consider values asserted to C
properties at class level, as assigned to a fictive (or external ?)
individual C (or of some  fictive ConceptName class) at instance level
instead.

Indeed, this solution is attractive when applied to metadata (corresponding
to the present OWL annotation properties), as it meets our needs to define
Preferred_name, Synonyms, UWDAID etc. (cf.  FMA OWL annotations) with domain
and range, functional, inverseFunctional axioms.

* Metamodelling
But I am not completely  conviced by the punning solution when applied for
metamodelling in general. Indeed, again if I understood it correctly, the
same facility can be used to provide values not only to what I called above
metadata but to any property.  Then, there are two questions that worry me.
1)  From my experience with the FMA for example, I am afraid this may lead
to a big confusion for users (missing understanding punning)  that may be
confused about a same entity that may be sometimes viewed as a class, and
sometimes as an individual of some metaconcept ( "a specy") because of a
single name with values assigned to that specific individual.
This may lead to a big confusion for example between the class Heart whose
individual are MyHeart, YourHeart etc.and the "Heart canonical entity"
instance of some metaconcept ListOfAnatomicalEntity etc. (I may elaborate
this further later). The same would occur for ontologies in genomics for
example for genes, protein, chemical entities etc. (and databases)
This may perhaps lead to a complete change in the way biomedical ontologies
are presentely designed and considered.
2) I worry also about inferences that might be missed in doing so, as soon
as some "meta-property" are assigned values for example a class SulcusTypeA
assigned with an a priori defined value averageDepth = 30, while another
class SulcusTypeB  with averageDepth = 5)
(I did not get why the focus is only on aggregation, the same seems to
hold )
Although such "real" metamodelling seems also relevant, it might perhaps be
preferable to distinguish it from metadata and look for solutions later ?

Would it be possible to have 2 steps. First in OWL1.1 adopt the
punning option but only for "metadata", restricting its use to properties
declared as annotatationProperty (although I am aware it does not provide
any guarentee that users may not declare other properties as annotation.
Perhaps introducing a class conceptName would make it more clear for users
(?), and to look for another solution for more general metamodelling in  OWL
2 ? Here, I leave this discussion to specialists of this mailing list !

My impression is that things seem quite subtile and need carfull
investigation, as they may have strong implications, not only from OWL/RDF
perspective but also from an ontological perspective for SW ontologies in
general, and apparentely also for data integration.

Again sorry if I am completely confused.


Christine

----- Original Message -----
From: "Alan Rector" <Alan.Rector@manchester.ac.uk>
To: "Denny Vrandecic" <dvr@aifb.uni-karlsruhe.de>; "Peter Patel-Schneider"
<pfps@research.bell-labs.com>
Cc: <owl@lists.mindswap.org>
Sent: Thursday, January 05, 2006 1:17 AM
Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1


Peter, Danny, All

Is part of the problem that we have no way to distinguish higher
order statements about the domain from annotations of the symbols and
artefacts we are using to represent that domain?   Or do we?

This seems to me a common problem.  When I make an annotation on some
bit of an symbolic artefact, I am giving information about the
symbols in the artefact, not the domain they represents. So if I
author a symbol "Galapagos Finch" and a symbol "Darwin"  and a symbol
"Rector", I want to be able to say that the Symbols "Galapagos Finch"
and "Darwin" in this ontology were created by "Rector" but I don't
want to reason about them in the same space. (By contrast if I want
to say that the class denoted by the symbol "Galapagos Finch" was
described by "Darwin", I am talking about the class "Galapagos Finch"
which is an instance of "owl:Class". Or as a more clear cut example,
I want to say that "Galapagos Finch" is an endagered species and be
able to ask queries or form the class of "All members of endangered
species". Again, I am making a higher order statement about the class
"Galapagos Finch" in the domain rather than about the symbolic
representation of that that class. )

For annotation of the symbolic representation as opposed to higher
order statements about the class represented, what you seem to need
is a) a means of referring to the symbolic expressions "Galapagos
Finch", "Darwin" etc. and the expressions using them rather than to
their referents.

Unfortunately, this is not - at least as I understand it - what RDF
Reification does.  As I read the Semantics document - please correct
me if I am wrong - the semantics of reification as described by:

".. <begin quote http://www.w3.org/TR/rdf-syntax-grammar.>..

<ex:a> <ex:b> <ex:c> .

and

_:xxx rdf:type rdf:Statement .
_:xxx rdf:subject <ex:a> .
_:xxx rdf:predicate <ex:b> .
_:xxx rdf:object <ex:c> .

.... end quote"

is to create a blank node that represents ANY triple of the form

  <ex:a> <ex:b> <ex:c>

rather than the specific triple in my representation artefact.

I want to say things about the specific symbolic triple in a specific
resource.  I don't know if this is possible by referring to the full
URI of the rdf statement or not, but I can't see how as rdf triples
do not appear to have URIs, even though their subject, property, and
object all do.  I think what is being suggested is that OWL axioms,
or at least some OWL axioms, should.

Obviously, without some kind of partitioning, allowing a statement to
refer to another statement would allow a statement to refer to
itself, directly or indirectly, and  would bring with it the
paradoxes of self reference and therefore undecidability,  at least
as soon it was used with an expressive  enough semantics to include
negation or falsehood.

The simplest possible form of partitioning would seem to be to
declare that some properties are annotations and do not affect
inference - what I understood annotations did.  In fact, what I want
to do, almost always, is query the actual representation with
something like database (datalog) semantics, e.g. "Find all the
axioms in ontology X that were imported from ontology A"; "Find all
the statements in ontologies X,Y and Z asserted to be authored by
'Rector'" .  "Find statements in ontology X for which there is no
assertion of authorship", etc.

This seems to follow naturally from Ian Horrocks' note on Databases
and Ontologies in which he says "DL KB doesn?t define a single model,
it is a set of constraints that define a set of possible models...In
contrast, DBs (and frame/rule KR systems) make assumptions such that
DB/KB defines a single model"  This is a case where I am clearly
interested in one possible model - the symbols and axioms actually in
my implementation of the OWL KB and how they got there, how they fit
together, etc.

There are a host of issues about copies and versions etc.,  but I
can't see trying to solve them until the fundamental principles are
clear. To start with at a minimum, I need to be able to send an
artefact along with the description of that artefact in terms of
provenance, usage, and editing annotations needed for somebody else
(who shares the same annotation vocabulary) to use it in the same
way, know where it came from, decide if they trust it, etc. (Note I
trust artefacts not the things the artefacts represent.).

As far as I can see,  The use of the same symbol as a class and
instance in OWL Full is a way of doing higher order representation,
although this is not explicitly clear in anything that I can find.
In which case the problems of fine grained annotation apply equally
to OWL full and OWL-DL.

In summary, my contention is that there are three distinct kinds of
statements about a) all instances of a class, b) the class itself,
and c) the symbolic representation of the class. Further that we need
to be able to refer to all constructs of the representation that can
be expressed individually including individual triples, axioms,  (and
the use of axioms to apply a restriction to a class) as well as
classes and individuals.   My concern is that b) and c)  are often
conflated. I would refer to b) as "higher order representation" and
c) as" annotation". This conflation is reflected in the common use of
the word "metadata" indiscrimately to refer to either or both.

I find it hard to see how we can resolve the issues around annotation
definitively without a clear distinction between these two notions.
If this is true, then there is a long term problem for which a best a
'band aid' can be provided in OWL 1.1.

Or am I just completely confused?

Regards

Alan




On 15 Dec 2005, at 13:37, Denny Vrandecic wrote:

> Hi Bijan,
>
> thanks, very good and important questions. I'll try to answer them,
> but I am happy if others would join in resolving them as well.
>
> First, let's ask a question in return: are axioms resources? Do we
> want to be *able* to speak about axioms? Is there anything
> interesting to say about axioms? If you say yes to any of these,
> they need an URI. It appears so totally wrong to me, that we
> (supposedly) can talk about anything and that we ask anyone to
> provide URIs for their stuff, so we can pull it into the Semantic
> Web - but we do not offer URIs for our own statements. I'm against
> a "You shall not annotate axioms"-law. Does anyone disagree on
> that? - and if so, why?
>
>> When is the URI being treated as an axiom?
>
> URIs of individuals, classes, properties... and axioms are all
> disjoint from each other, and thus, just as URIs of individuals,
> classes, properties... etc. axiom URIs need to be either declared
> or it must be inferrable from the context that this is an axiom URI.
>
>> Does the mere presence of an axiom URI in a KB put the axiom
>> itself there?
>
> I don't know. This would mean that the axiom must be known to the
> reasoner. And if it is known, isn't it already there? Hmm. Maybe
> not. I'd prefer a "no, it doesn't", but I am not sure about the
> implications of that.
>
>> Can I make any interesting constructions using the Axiom uri?
>
> I sure hope so, but what do you have in mind?
>
>> Should axiom equivalence entail URI equivalence?
>
> Yes. Because then I can merge ontologies and the annotations of
> axioms merge fittingly.
>
>> (Presumably not, eh, but what about through trivial rewritings,
>> i.e., different rdf serializations?)
>
> Huh, why not?
>
>> Are all axioms named by default, or must one coin a name to talk
>> about them?
>
> The latter.
>
>> Are annotations on axioms restricted to axioms with names/URIs?
>
> Can you make annotations on unnamed descriptions in OWL DL?
>
>> How do we avoid the pain that RDF Reification brings (and it's not
>> just the rdf:subject/predicate, etc. the Statement triple itself
>> and the introduction of the URI means CONSIDERABLE pain....note
>> the fact that no store, to my knowledge, that introduces
>> "contexts" does so by naming all the triples).
>
> I wish I had a convincing answer for that. What about "only
> annotated axioms actually need an URI"? Or "this is just enabling
> you to give an axiom an URI, it is not forcing you"? Or "hmm, I
> think we can hide the pain most of the time, by nicely extending
> the abstract syntax or the OWL XML presentation syntax. It will
> only remain really painful and ugly in the XML/RDF serialization"?
>
> As said, none of these answers really convince me, but maybe they
> are a starter for a solution. Or someone else has one. Maybe there
> is a way to give URIs to axioms without RDF reification, this would
> be great, and I'm just not seeing it.
>
> The question is, is avoiding this pain worth the restriction of
> disallowing to annotate axioms?
>
> Cheers,
> denny
> _______________________________________________
> OWL mailing list
> OWL@lists.mindswap.org
> http://lists.mindswap.org/mailman/listinfo/owl

-----------------------
Alan Rector
Professor of Medical Informatics
Department of Computer Science
University of Manchester
Manchester M13 9PL, UK
TEL +44 (0) 161 275 6188/6149
FAX +44 (0) 161 275 6204
www.cs.man.ac.uk/mig
www.clinical-esciences.org
www.co-ode.org

_______________________________________________
OWL mailing list
OWL@lists.mindswap.org
http://lists.mindswap.org/mailman/listinfo/owl


From matthew.pocock at ncl.ac.uk  Wed Jan 11 07:04:44 2006
From: matthew.pocock at ncl.ac.uk (Matthew Pocock)
Date: Wed Jan 11 07:05:41 2006
Subject: [OWL] role value maps
Message-ID: <200601111204.44810.matthew.pocock@ncl.ac.uk>

Hi,

I've been reading the DL handbook in a vein attempt to catch up on what's 
going on. The last bit of chapter two discusses role-value-maps. These let 
you say interesting things about how the values of properties are correlated. 
I've no idea how much work it is to add these concept constructors to the 
existing reasoners, or even what restrictions would need to be placed upon 
roles that can be composed (non-transient, functional, ...?) to keep things 
decidable.

Is this on the table for owl 1.1, or even for owl 2?

Let's use a -> b for role composition (i a -> b j == i a m b j) and a < b for 
role chains (i a m => j b m).

  Person and (hasChild -> hasFriend < knows)

This describes persons who have children who have friends who are known by the 
person. Apparently if the properties in the property chain are functional, 
then these DLs are decidable.

The ability to capture these sorts of constraints would be extremely useful. 
For example, we can model things like:

* Information in pedegrees, like consanguinuity (in-breeding - children of 
cousins or siblings)

  Consanguinous == hasMother -> hasParent = hasFather -> hasParent

* In meta-modelling, templated types and roles.

  no example a we don't have a meta-modelling syntax

* A biological sample is treated with flourescent dies that are active at 
specific wave-lengths. Images of the sample are captured at specific 
wavelengths. We want to check that the image is captured at atleast those 
needed by the flourescent dies. I guess for this end-user application we need 
a mixture of role value maps and the k operator.

  SensibleImage
    Image
    capturedFrom -> treatedWith -> flourescesAt < scannedAt

  sample treatedWith die1
  die1 flourescesAt 360
  image capturedFrom sample
  image scannedAt 360
  image scannedAt 420

* Lots of the things you do in SQL through joins

Matthew
From Ulrike.Sattler at manchester.ac.uk  Wed Jan 11 07:19:45 2006
From: Ulrike.Sattler at manchester.ac.uk (Uli Sattler)
Date: Wed Jan 11 07:22:15 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <1C210251-1BA2-4752-A501-7C465E71E1E0@inf.unibz.it>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<02fd01c6152b$73db3280$d3d7858b@Newton>
	<1C210251-1BA2-4752-A501-7C465E71E1E0@inf.unibz.it>
Message-ID: <67B139DA-6298-45B4-A742-685D4CABD283@cs.man.ac.uk>


On 9 Jan 2006, at 15:00, Enrico Franconi wrote:

>
> On 9 Jan 2006, at 15:45, Jeff Z. Pan wrote:
>

[snippp]

>>> - The only approach that would capture exactly meta-modelling with
>>> OWL-DL is undecidable (see Boris' paper) - this fact was hidden by
>>> Peter...
>>
>> Which (only) approach do you mean? As far as I understand, Boris'  
>> paper [3] propose two approaches: the context (punning) approach  
>> and the Hilog approach.
>
> No, the very first part of the paper is about undecidability of the  
> "real" approach.
>

I slightly disagree -- it is about undecidability caused by the
mixing of logical and meta-logical symbols, not meta-modelling
per se.

Cheers, Uli


> --e.
>
> _______________________________________________
> OWL mailing list
> OWL@lists.mindswap.org
> http://lists.mindswap.org/mailman/listinfo/owl

From franconi at inf.unibz.it  Thu Jan 12 03:09:40 2006
From: franconi at inf.unibz.it (Enrico Franconi)
Date: Thu Jan 12 03:09:41 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <67B139DA-6298-45B4-A742-685D4CABD283@cs.man.ac.uk>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<02fd01c6152b$73db3280$d3d7858b@Newton>
	<1C210251-1BA2-4752-A501-7C465E71E1E0@inf.unibz.it>
	<67B139DA-6298-45B4-A742-685D4CABD283@cs.man.ac.uk>
Message-ID: <9421E748-9443-4FA9-9EF6-C1CC90C0D32F@inf.unibz.it>


On 11 Jan 2006, at 13:19, Uli Sattler wrote:

>
> On 9 Jan 2006, at 15:00, Enrico Franconi wrote:
>
>>
>> On 9 Jan 2006, at 15:45, Jeff Z. Pan wrote:
>>
>
> [snippp]
>
>>>> - The only approach that would capture exactly meta-modelling with
>>>> OWL-DL is undecidable (see Boris' paper) - this fact was hidden by
>>>> Peter...
>>>
>>> Which (only) approach do you mean? As far as I understand, Boris'  
>>> paper [3] propose two approaches: the context (punning) approach  
>>> and the Hilog approach.
>>
>> No, the very first part of the paper is about undecidability of  
>> the "real" approach.
>>
>
> I slightly disagree -- it is about undecidability caused by the
> mixing of logical and meta-logical symbols, not meta-modelling
> per se.

Oh yes, you are right.
--e.
From pfps at research.bell-labs.com  Thu Jan 12 08:14:25 2006
From: pfps at research.bell-labs.com (Peter F. Patel-Schneider)
Date: Thu Jan 12 08:14:35 2006
Subject: [OWL] [1.1] minor modification of OWL 1.1 syntax document
Message-ID: <20060112.081425.38064928.pfps@research.bell-labs.com>

The OWL 1.1 syntax document, available at
	http://www-db.research.bell-labs.com/user/pfps/owl/syntax.html
has been changed (very) slightly.  The wording has been changed in a few places
to make the intent clearer.  Annotations have been added to the syntax for
datatype axioms to fix a minor bug in the syntax.

peter
From pfps at research.bell-labs.com  Thu Jan 12 10:30:17 2006
From: pfps at research.bell-labs.com (Peter F. Patel-Schneider)
Date: Thu Jan 12 10:30:20 2006
Subject: [OWL] role value maps
In-Reply-To: <200601111204.44810.matthew.pocock@ncl.ac.uk>
References: <200601111204.44810.matthew.pocock@ncl.ac.uk>
Message-ID: <20060112.103017.04992899.pfps@research.bell-labs.com>

Some kinds of role composition (role chain) inclusions are not hard to handle,
and, indeed, have been included in OWL 1.1.  (See the syntax document for more
details.)  Other kinds are more difficult, even causing undecidability.  There
are several DL papers on this topic.  (Unfortunately, I'm travelling just now,
and it is not easy for me to dig up the best references.)

peter


From: Matthew Pocock <matthew.pocock@ncl.ac.uk>
Subject: [OWL] role value maps
Date: Wed, 11 Jan 2006 12:04:44 +0000

> Hi,
> 
> I've been reading the DL handbook in a vein attempt to catch up on what's 
> going on. The last bit of chapter two discusses role-value-maps. These let 
> you say interesting things about how the values of properties are correlated. 
> I've no idea how much work it is to add these concept constructors to the 
> existing reasoners, or even what restrictions would need to be placed upon 
> roles that can be composed (non-transient, functional, ...?) to keep things 
> decidable.
> 
> Is this on the table for owl 1.1, or even for owl 2?
> 
> Let's use a -> b for role composition (i a -> b j == i a m b j) and a < b for 
> role chains (i a m => j b m).
> 
>   Person and (hasChild -> hasFriend < knows)
> 
> This describes persons who have children who have friends who are known by the 
> person. Apparently if the properties in the property chain are functional, 
> then these DLs are decidable.
> 
> The ability to capture these sorts of constraints would be extremely useful. 
> For example, we can model things like:
> 
> * Information in pedegrees, like consanguinuity (in-breeding - children of 
> cousins or siblings)
> 
>   Consanguinous == hasMother -> hasParent = hasFather -> hasParent
> 
> * In meta-modelling, templated types and roles.
> 
>   no example a we don't have a meta-modelling syntax
> 
> * A biological sample is treated with flourescent dies that are active at 
> specific wave-lengths. Images of the sample are captured at specific 
> wavelengths. We want to check that the image is captured at atleast those 
> needed by the flourescent dies. I guess for this end-user application we need 
> a mixture of role value maps and the k operator.
> 
>   SensibleImage
>     Image
>     capturedFrom -> treatedWith -> flourescesAt < scannedAt
> 
>   sample treatedWith die1
>   die1 flourescesAt 360
>   image capturedFrom sample
>   image scannedAt 360
>   image scannedAt 420
> 
> * Lots of the things you do in SQL through joins
> 
> Matthew
> _______________________________________________
> OWL mailing list
> OWL@lists.mindswap.org
> http://lists.mindswap.org/mailman/listinfo/owl
From shikha_mah at rediffmail.com  Fri Jan 13 04:23:20 2006
From: shikha_mah at rediffmail.com (Shikha Maheshwari )
Date: Fri Jan 13 04:23:37 2006
Subject: [OWL] a query regarding OWL
Message-ID: <20060113092320.24484.qmail@webmail52.rediffmail.com>

 ?
If I want to represent the semantic relationship between two concepts then how would it be done in OWL?

for ex. C1 Relation C2
      "Educational_Institute"  facilitates_to  "Student"

Can it be done <owl:ObjectProperty/>? As:

<owl:class rdf ID="Educational_Institute">
    <owl:ObjectProperty rdf ID: ="Facilitates_to">
      .... 

But what would be complete code for it?
Please do help me.

Thanks and Regards,
Shikha Maheshwari
Researcher
Ministry of IT
Delhi, India.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mindswap.org/pipermail/owl/attachments/20060113/cbd10c43/attachment.html
From jpan at csd.abdn.ac.uk  Fri Jan 13 05:17:08 2006
From: jpan at csd.abdn.ac.uk (jpan@csd.abdn.ac.uk)
Date: Fri Jan 13 05:17:30 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <1C210251-1BA2-4752-A501-7C465E71E1E0@inf.unibz.it>
References: <20051215073809.DLL67381@isrmail.isr.umd.edu><43A1717E.5060207@aifb.uni-karlsruhe.de>
	<46043C99-9548-44F4-AE15-9F96F831DDBD@manchester.ac.uk>
	<005101c6148e$df46aba0$d3d7858b@Newton>
	<6F46CEDD-A4F2-4749-AFDB-9877D5262599@inf.unibz.it>
	<02fd01c6152b$73db3280$d3d7858b@Newton>
	<1C210251-1BA2-4752-A501-7C465E71E1E0@inf.unibz.it>
Message-ID: <20060113101708.m4hnjupo63cc44s4@www.csd.abdn.ac.uk>


"Enrico Franconi" <franconi@inf.unibz.it> wrote:

>
> On 9 Jan 2006, at 15:45, Jeff Z. Pan wrote:
>>> - OWL-FA semantics does not exist yet: the last time I saw it was
>>> severely bugged; it would break wrt normative RDF; it is not
>>> compatible with punning.
>>
>> Do you mean "being not compatible with punning" is a severe bug of  [2]?
>
> No, it's just an additional problem for me.

It would be helpful if you could also clarify the other problems (that you
claimed) of OWL FA [2], if any.


>> - Hilog captures more expected inferences than punning but (a) Peter
didn't say that it still leave some expected interesting inferences
out (see Boris' paper for an example).

I am also expecting your support for the above claim.

Cheers,

Jeff.
--


1] http://www-db.research.bell-labs.com/user/pfps/owl/syntax.html
[2] http://www.mindswap.org/2005/OWLWorkshop/sub15.pdf
[3]
http://dip.semanticweb.org/documents/Boris-Motik-On-the-Properties-of-Metamodeling-in-OWL.pdf
[4] http://www.w3.org/TR/rdf-mt/
> [5] Jos de Bruijn, Enrico Franconi, Sergio Tessaris (2005). Logical  
> Reconstruction of normative RDF. Proc. of the Workshosp on OWL  
> Experiences and Directions (OWLED 2005), Galway, Ireland, November 
> 2005.


From gstoil at image.ece.ntua.gr  Fri Jan 13 05:18:24 2006
From: gstoil at image.ece.ntua.gr (George Stoilos)
Date: Fri Jan 13 05:18:28 2006
Subject: [OWL] role value maps
In-Reply-To: <200601111204.44810.matthew.pocock@ncl.ac.uk>
Message-ID: <001101c6182a$b0364ec0$760b6693@image.ece.ntua.gr>



> -----Original Message-----
> From: owl-bounces@lists.mindswap.org
> [mailto:owl-bounces@lists.mindswap.org]On Behalf Of Matthew Pocock
> Sent: Wednesday, January 11, 2006 2:05 PM
> To: owl@lists.mindswap.org
> Subject: [OWL] role value maps
>
>
> Hi,
>
> I've been reading the DL handbook in a vein attempt to catch
> up on what's
> going on. The last bit of chapter two discusses
> role-value-maps. These let
> you say interesting things about how the values of properties
> are correlated.
> I've no idea how much work it is to add these concept
> constructors to the
> existing reasoners, or even what restrictions would need to
> be placed upon
> roles that can be composed (non-transient, functional, ...?)
> to keep things
> decidable.
>
> Is this on the table for owl 1.1, or even for owl 2?
>
> Let's use a -> b for role composition (i a -> b j == i a m b
> j) and a < b for
> role chains (i a m => j b m).
>
>   Person and (hasChild -> hasFriend < knows)
>
> This describes persons who have children who have friends who
> are known by the
> person. Apparently if the properties in the property chain
> are functional,
> then these DLs are decidable.

But if you restrict these properties to be functional then you have not
modeled "persons who know the friends of their children",
but persons who know the single friend of their only child. So there is a
great restriction to expressivity (obviously) in order to regain
decidability. Additionally, I think that there might be other standard DL
ways to have the same effect as the only-child only-friend chain.

-gstoil

>
> The ability to capture these sorts of constraints would be
> extremely useful.
> For example, we can model things like:
>
> * Information in pedegrees, like consanguinuity (in-breeding
> - children of
> cousins or siblings)
>
>   Consanguinous == hasMother -> hasParent = hasFather -> hasParent
>
> * In meta-modelling, templated types and roles.
>
>   no example a we don't have a meta-modelling syntax
>
> * A biological sample is treated with flourescent dies that
> are active at
> specific wave-lengths. Images of the sample are captured at specific
> wavelengths. We want to check that the image is captured at
> atleast those
> needed by the flourescent dies. I guess for this end-user
> application we need
> a mixture of role value maps and the k operator.
>
>   SensibleImage
>     Image
>     capturedFrom -> treatedWith -> flourescesAt < scannedAt
>
>   sample treatedWith die1
>   die1 flourescesAt 360
>   image capturedFrom sample
>   image scannedAt 360
>   image scannedAt 420
>
> * Lots of the things you do in SQL through joins
>
> Matthew
> _______________________________________________
> OWL mailing list
> OWL@lists.mindswap.org
> http://lists.mindswap.org/mailman/listinfo/owl
>

From jpan at csd.abdn.ac.uk  Fri Jan 13 05:59:57 2006
From: jpan at csd.abdn.ac.uk (Jeff Z. Pan)
Date: Fri Jan 13 06:00:12 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <20060110.064356.131777543.pfps@research.bell-labs.com>
References: <005101c6148e$df46aba0$d3d7858b@Newton>
	<20060109.030209.64527517.pfps@research.bell-labs.com>
	<20060110092632.uoj32vc5hg004o40@www.csd.abdn.ac.uk>
	<20060110.064356.131777543.pfps@research.bell-labs.com>
Message-ID: <20060113105957.ad1so8w0dx0lcw0g@www.csd.abdn.ac.uk>

"Peter F. Patel-Schneider" <pfps@research.bell-labs.com> wrote:

> From: jpan@csd.abdn.ac.uk
> Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1
> Date: Tue, 10 Jan 2006 09:26:32 +0000
>
>>
>> "Peter F. Patel-Schneider" <pfps@research.bell-labs.com> wrote:
>>
>> > From: "Jeff Z. Pan" <jpan@csd.abdn.ac.uk>
>> > Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1
>> > Date: Sun, 8 Jan 2006 20:05:20 -0000
>> >
>> >> Hi Peter, Alan and all,
>
> [...]
>
>> >> 1. It is impossible to distinguish higher order statements from 
>> >> annotations of
>> >> symbols and
>> >> artefacts we are using to represent that domain, as pointed out in Alan's
>> >> email. The reason that they are not distinguishable is because >> 
>> annotations in
>> >> [1] are simply syntactic sugar of individual axioms.
>> >
>> > Yes,
>>
>> OK.
>>
>> > but what proposal does distinguish between higher-order statements and
>> > annotations of symbols?
>>
>> Higher-order statements (axioms about meta-classes and meta-properties) and
>> annotations (in the sense of OWL DL) are two seperate things, I don't
>> understand why we cannot distinguish them.
>
> Yes, it would be possible, but what proposal on the table does currently have
> both of them, distinguished?

At least OWL FA does.


>> >> 2. Datatype axioms, unlike other axioms in OWL 1.1 [1], cannot have
>> >> annotations. This seems pretty strange, at least to me. The 
>> reason is that
>> >> although individuals, object properties and classes can share 
>> names, classes
>> >> and datatypes cannot.
>> >
>> >> From the OWL 1.1 syntax document [1], recapitulating the OWL DL syntax:
>> >
>> > axiom ::= 'DatatypeProperty(' datavaluedPropertyID ['Deprecated'] 
>> { > annotation }                { 'super(' datavaluedPropertyID ')'} 
>> > ['Functional']
>> >                { 'domain(' description ')' } { 'range(' dataRange 
>> ')' } ')'
>> >
>> > This sure looks as if datatype axioms can have annotations.
>>
>> What I meant was
>>
>> axiom ::= 'Datatype(' datatypeID 'base(' datatypeID ')' { 
>> datatypeRestriction }
>> ')'.
>>
>> We don't have annotations here, don't we?
>
> That is a bug, which will be fixed immediately.

> The OWL 1.1 syntax document, available at
> http://www-db.research.bell-labs.com/user/pfps/owl/syntax.html
> has been changed (very) slightly.  The wording has been changed in a 
> few places
> to make the intent clearer.  Annotations have been added to the syntax for
> datatype axioms to fix a minor bug in the syntax.

I am not sure if the following sentences (from [1]) arecorrect:  "With this
change, non-annotation properties can be placed on any name.  The property
applies to the use of the name as an individual. As a simple syntactic sugar,
non-annotation properties can be part of certain class and property axioms. "
Can we put non-annotation properties on datatype names?


As suggested in earlier emails, the main concern that I have is what 
the uses of
the punning semantics are. We appreciate that it is easy, but if it has 
no use,
why do we bother to adopt it in OWL 1.1?

Cheers,

Jeff.
--


[1] http://www-db.research.bell-labs.com/user/pfps/owl/syntax.html
[2] http://www.mindswap.org/2005/OWLWorkshop/sub15.pdf
[3]
http://dip.semanticweb.org/documents/Boris-Motik-On-the-Properties-of-Metamodeling-in-OWL.pdf
[4] http://www.w3.org/TR/rdf-mt/
> [5] Jos de Bruijn, Enrico Franconi, Sergio Tessaris (2005). Logical  
> Reconstruction of normative RDF. Proc. of the Workshosp on OWL  
> Experiences and Directions (OWLED 2005), Galway, Ireland, November 
> 2005.

From shikha_mah at rediffmail.com  Fri Jan 13 06:23:46 2006
From: shikha_mah at rediffmail.com (Shikha Maheshwari )
Date: Fri Jan 13 06:24:03 2006
Subject: [OWL]query regarding OWL
Message-ID: <20060113112346.5760.qmail@webmail47.rediffmail.com>

 ?
If I want to represent the semantic relationship between two concepts then how would it be done in OWL?

for ex. C1 Relation C2
      "Educational_Institute"  facilitates_to  "Student"

Can it be done <owl:ObjectProperty/>? As:

<owl:class rdf ID="Educational_Institute">
    <owl:ObjectProperty rdf ID: ="Facilitates_to">
      .... 

But what would be complete code for it?
Please do help me.

Thanks and Regards,
Shikha Maheshwari
Researcher
Ministry of IT
Delhi, India. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mindswap.org/pipermail/owl/attachments/20060113/4d51c458/attachment.html
From pfps at research.bell-labs.com  Fri Jan 13 06:25:07 2006
From: pfps at research.bell-labs.com (Peter F. Patel-Schneider)
Date: Fri Jan 13 06:25:10 2006
Subject: [OWL] annotations and meta-modelling in OWL 1.1
In-Reply-To: <20060113105957.ad1so8w0dx0lcw0g@www.csd.abdn.ac.uk>
References: <20060110092632.uoj32vc5hg004o40@www.csd.abdn.ac.uk>
	<20060110.064356.131777543.pfps@research.bell-labs.com>
	<20060113105957.ad1so8w0dx0lcw0g@www.csd.abdn.ac.uk>
Message-ID: <20060113.062507.80330354.pfps@research.bell-labs.com>

From: "Jeff Z. Pan" <jpan@csd.abdn.ac.uk>
Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1
Date: Fri, 13 Jan 2006 10:59:57 +0000

[...]

> >> Higher-order statements (axioms about meta-classes and meta-properties) and
> >> annotations (in the sense of OWL DL) are two seperate things, I don't
> >> understand why we cannot distinguish them.
> >
> > Yes, it would be possible, but what proposal on the table does currently have
> > both of them, distinguished?
> 
> At least OWL FA does.

It does?  Does OWL FA really distinguish between statements about names of
things (annotations) and statements about their meaning (including higher-order
statements)?

[...]

> I am not sure if the following sentences (from [1]) arecorrect:  "With this
> change, non-annotation properties can be placed on any name.  The property
> applies to the use of the name as an individual. As a simple syntactic sugar,
> non-annotation properties can be part of certain class and property axioms. "
> Can we put non-annotation properties on datatype names?

Yes, as the syntactic category of annotations in OWL 1.1 includes stating
values for non-annotation properties.

> As suggested in earlier emails, the main concern that I have is what 
> the uses of
> the punning semantics are. We appreciate that it is easy, but if it has 
> no use,
> why do we bother to adopt it in OWL 1.1?

I'm not convinced that it has no use.

> Cheers,
> 
> Jeff.

peter

From shikha_mah at rediffmail.com  Fri Jan 13 02:08:41 2006
From: shikha_mah at rediffmail.com (Shikha Maheshwari )
Date: Fri Jan 13 06:32:31 2006
Subject: [OWL] a query regarding OWL
Message-ID: <20060113070841.5264.qmail@webmail46.rediffmail.com>

 ?

If I want to represent the semantic relationship between two concepts then how would it be done in OWL?

for ex. C1 Relation C2
       "Educational_Institute" ?facilitates_to  "Student"

Can it be done <owl:ObjectProperty/>? As:

<owl:class rdf ID="Educational_Institute">
    <owl:ObjectProperty rdf ID: ="Facilitates_to">
      .... 

But what would be complete code for it?
Please do help me.

Thanks and Regards,
Shikha Maheshwari
Researcher
Ministry of IT
Delhi, India.

Keep Smiling & Take things in a POSITIVE way.

Shikha Maheshwari
M.Tech (CS)
Banasthali Vidyapith
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mindswap.org/pipermail/owl/attachments/20060113/8a2881de/attachment.html
From pfps at research.bell-labs.com  Fri Jan 13 06:38:13 2006
From: pfps at research.bell-labs.com (Peter F. Patel-Schneider)
Date: Fri Jan 13 06:38:17 2006
Subject: [OWL] a query regarding OWL
In-Reply-To: <20060113070841.5264.qmail@webmail46.rediffmail.com>
References: <20060113070841.5264.qmail@webmail46.rediffmail.com>
Message-ID: <20060113.063813.94343898.pfps@research.bell-labs.com>

In OWL 1.1, this is simple, provided that you really need it.  

Class(Student partial)
Class(Educational_Institute partial value(facilitates_to Student))

peter

PS: It is probably better to write examples in the functional syntax, not in
RDF/XML.





From: "Shikha Maheshwari " <shikha_mah@rediffmail.com>
Subject: [OWL] a query regarding OWL
Date: 13 Jan 2006 07:08:41 -0000

> If I want to represent the semantic relationship between two concepts
> then how would it be done in OWL?
> 
> for ex. C1 Relation C2
>       "Educational_Institute"  facilitates_to  "Student"
> 
> Can it be done <owl:ObjectProperty/>? As:
> 
> <owl:class rdf ID="Educational_Institute">
>     <owl:ObjectProperty rdf ID: ="Facilitates_to">
>       .... 
> 
> But what would be complete code for it?
> Please do help me.
> 
> Thanks and Regards,
> Shikha Maheshwari
> Researcher
> Ministry of IT
> Delhi, India. 
> 
> Keep Smiling & Take things in a POSITIVE way. 
> 
> Shikha Maheshwari 
> M.Tech (CS) 
> Banasthali Vidyapith 
> 
> 
>  
> <http://adworks.rediff.com/cgi-bin/AdWorks/sigclick.cgi/www.rediff.com/s
> ignature-home.htm/1507191490@Middle5?PARTNER=3>  
From bernardo at mindswap.org  Fri Jan 13 08:05:52 2006
From: bernardo at mindswap.org (Bernardo Cuenca Grau)
Date: Fri Jan 13 08:05:54 2006
Subject: [OWL] [ALL][OWL 1.1] OWL 1.1 Model-Theoretic Semantics 
Message-ID: <Pine.LNX.4.58.0601130803300.10121@gimli.mindlab.umd.edu>


 The formal Semantics draft for OWL 1.1 is now available at:

 http://www.cs.man.ac.uk/~bcg/OWLSemantics.pdf


 Cheers,

 Bernardo
From jpan at csd.abdn.ac.uk  Fri Jan 13 10:34:20 2006
From: jpan at csd.abdn.ac.uk (Jeff Z. Pan)
Date: Fri Jan 13 10:34:38 2006
Subject: [OWL] [ALL][OWL 1.1] OWL 1.1 Model-Theoretic Semantics
Message-ID: <20060113153420.x7xfnkc09jwwgo8g@www.csd.abdn.ac.uk>

Bernardo,

> The formal Semantics draft for OWL 1.1 is now available at:
>
> http://www.cs.man.ac.uk/~bcg/OWLSemantics.pdf

Thanks for the work. A quick comment about datatypes: according to the OWL 1.1
syntax [1], OWL 1.1 supports datatype expressions, which are not covered by
concrete domains but by datatype groups [2].

Cheers,

Jeff.
--


[1] http://www-db.research.bell-labs.com/user/pfps/owl/syntax.html#2.3
[2] OWL-Eu: Adding Customised Datatypes into OWL, an online version is 
available
at http://www.websemanticsjournal.org/ps/pub/2005-24



From Christine.Golbreich at univ-rennes1.fr  Sat Jan 14 02:34:53 2006
From: Christine.Golbreich at univ-rennes1.fr (Christine Golbreich)
Date: Sat Jan 14 02:34:58 2006
Subject: [OWL]query regarding OWL
References: <20060113112346.5760.qmail@webmail47.rediffmail.com>
Message-ID: <03c801c618dd$0340df60$db9ffea9@LUZ>

Shikha,

Could you precise what you mean exactly?

- relationships between classes (concepts) are usually represented by owl:ObjectProperty (or owl:DatatypeProperty) in OWL.
is your question about getting the complete corresponding code ?
Then, you can use any editor you like for example Prot?g? OWL 
Create the class, then add object property with domain and range 
(see Matthew's excellent guide http://www.co-ode.org/resources/tutorials/ProtegeOWLTutorial.pdf) for all details)
Then you can go to the Tab Code or edit your OWL file to get complete code
(you have the Prot?g? OWL mailing list and archieve for any questions)

- if object property does not seem appropriate to your application, it is another story.
Then could you better explain why ?

Christine

  ----- Original Message ----- 
  From: Shikha Maheshwari 
  To: owl@lists.mindswap.org 
  Sent: Friday, January 13, 2006 12:23 PM
  Subject: [OWL]query regarding OWL


   
  If I want to represent the semantic relationship between two concepts then how would it be done in OWL?

  for ex. C1 Relation C2
        "Educational_Institute"  facilitates_to  "Student"

  Can it be done <owl:ObjectProperty/>? As:

  <owl:class rdf ID="Educational_Institute">
      <owl:ObjectProperty rdf ID: ="Facilitates_to">
        .... 

  But what would be complete code for it?
  Please do help me.

  Thanks and Regards,
  Shikha Maheshwari
  Researcher
  Ministry of IT
  Delhi, India. 




   


------------------------------------------------------------------------------


  _______________________________________________
  OWL mailing list
  OWL@lists.mindswap.org
  http://lists.mindswap.org/mailman/listinfo/owl
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mindswap.org/pipermail/owl/attachments/20060114/74b3d4fe/attachment.html
From alanruttenberg at gmail.com  Sun Jan 15 21:30:03 2006
From: alanruttenberg at gmail.com (Alan Ruttenberg)
Date: Sun Jan 15 21:30:10 2006
Subject: [OWL] Qualified/restricted properties
Message-ID: <5EB416EC-E9BC-4F40-917F-59DF98179BD3@gmail.com>

In a qualified cardinality restrictions we say something like

Property X can have at most n fillers of Class C

Another way to think about this is that there is an anonymous property

_P/C == The fillers of Property X which are of class C

which now has a simple cardinality restriction.

I'm wondering whether, under the hood, so to speak, these two are  
equivalent, and if so, whether  it is worth exposing the latter in OWL.

I can think of one case that has come up recently in BioPAX. We have  
a property XREF, which can be filled with instances of xrefs, a class  
with 3 subclasses, one of which is called a unificationXref. Suppose  
we would like to express that when there is a unificiationXref as the  
value of an XREF, it will uniquely determine the instance on which  
the relationship is asserted. It might be nice to be able to say:

ObjectProperty(_XREF/unificationXref InverseFunctional)

Right now I would model this as a second property UNIFICATION-XREF, a  
subproperty of XREF, and make sure that whenever I wanted to assert a  
unificationXref filler, I used the UNIFICATION-XREF property, rather  
than the XREF property.

If we want to make the decision late in modeling, the second choice  
requires more work than if we had these qualified/restricted  
properties available to work with.

Maybe this is already available and I'm not seeing it (please teach).  
Or maybe it introduces a reasoning problem which I am not qualified  
to discern (please say so).

Hope this makes sense. Let me know if I need to clarify.

Alan


From bparsia at isr.umd.edu  Sun Jan 15 22:46:31 2006
From: bparsia at isr.umd.edu (Bijan Parsia)
Date: Sun Jan 15 22:46:30 2006
Subject: [OWL] Qualified/restricted properties
In-Reply-To: <5EB416EC-E9BC-4F40-917F-59DF98179BD3@gmail.com>
References: <5EB416EC-E9BC-4F40-917F-59DF98179BD3@gmail.com>
Message-ID: <0caa10fce4443cdc2d2ebdbf8972cc61@isr.umd.edu>

On Jan 15, 2006, at 9:30 PM, Alan Ruttenberg wrote:

> In a qualified cardinality restrictions we say something like
>
> Property X can have at most n fillers of Class C
>
> Another way to think about this is that there is an anonymous property
>
> _P/C == The fillers of Property X which are of class C

Not quite since the restriction allows for there to be other fillers of 
X. So you capture part, but not all.

> which now has a simple cardinality restriction.

See this discussion:
	http://www.cs.vu.nl/~guus/public/qcr.html

> I'm wondering whether, under the hood, so to speak, these two are 
> equivalent,

Well, you can't exactly model qualified restrictions in OWL (the 
subproperty trick runs into trouble with max cardinalities, consider:

1)	E = max p 2 C & max p 2 D

If you model it as:

2)	E = max pC 2 & max pD 2.
	pC sub p. pC range C.
	pD sub p. pD range D.

In 1), you can only have 2 ps to a C. In 2) you can have an arbitrary 
number of ps to a C, not what you wanted.).

> and if so, whether  it is worth exposing the latter in OWL.

I don't think so, but hey, on to the use case.

> I can think of one case that has come up recently in BioPAX. We have a 
> property XREF, which can be filled with instances of xrefs, a class 
> with 3 subclasses, one of which is called a unificationXref. Suppose 
> we would like to express that when there is a unificiationXref as the 
> value of an XREF, it will uniquely determine the instance on which the 
> relationship is asserted.

Er...do you mean that unificationXref is a subclass of max 1 
inverseOf(XREF)?

> It might be nice to be able to say:
>
> ObjectProperty(_XREF/unificationXref InverseFunctional)

I don't think it would be nice :)

> Right now I would model this as a second property UNIFICATION-XREF, a 
> subproperty of XREF, and make sure that whenever I wanted to assert a 
> unificationXref filler, I used the UNIFICATION-XREF property, rather 
> than the XREF property.

I think that having to make a names inverseOf is a wart in OWL (instead 
of being able to use an inverse operator),b ut it seems nicer than 
this.

> If we want to make the decision late in modeling, the second choice 
> requires more work than if we had these qualified/restricted 
> properties available to work with.

I don't see why you want this in addition to QCRs. Actually, in your 
use case, I don't see what QCRs have to do with it. It is a bit late 
thought :)

> Maybe this is already available and I'm not seeing it (please teach). 
> Or maybe it introduces a reasoning problem which I am not qualified to 
> discern (please say so).
>
> Hope this makes sense. Let me know if I need to clarify.

Please!

Cheers,
Bijan.

From alanruttenberg at gmail.com  Sun Jan 15 23:35:27 2006
From: alanruttenberg at gmail.com (Alan Ruttenberg)
Date: Sun Jan 15 23:35:31 2006
Subject: [OWL] Qualified/restricted properties
In-Reply-To: <0caa10fce4443cdc2d2ebdbf8972cc61@isr.umd.edu>
References: <5EB416EC-E9BC-4F40-917F-59DF98179BD3@gmail.com>
	<0caa10fce4443cdc2d2ebdbf8972cc61@isr.umd.edu>
Message-ID: <AF1B389E-6199-4DF7-94C4-0C2891FC3D20@gmail.com>


On Jan 15, 2006, at 10:46 PM, Bijan Parsia wrote:

> On Jan 15, 2006, at 9:30 PM, Alan Ruttenberg wrote:
>
>> In a qualified cardinality restrictions we say something like
>>
>> Property X can have at most n fillers of Class C
>>
>> Another way to think about this is that there is an anonymous  
>> property
>>
>> _P/C == The fillers of Property X which are of class C
>
> Not quite since the restriction allows for there to be other  
> fillers of X. So you capture part, but not all.

I think what I meant to say is "there is an anonymous subproperty of  
X: _X/C == ..."
Does this address your comment?

> Well, you can't exactly model qualified restrictions in OWL (the  
> subproperty trick runs into trouble with max cardinalities, consider:
>
> 1)	E = max p 2 C & max p 2 D
>
> If you model it as:
>
> 2)	E = max pC 2 & max pD 2.
> 	pC sub p. pC range C.
> 	pD sub p. pD range D.
>
> In 1), you can only have 2 ps to a C. In 2) you can have an  
> arbitrary number of ps to a C, not what you wanted.).

I think, possibly, that this addresses the case in OWL 1.0. I should  
have said that I am referring to a baseline of OWL 1.1 which will  
have real QCRs.

>
>> and if so, whether  it is worth exposing the latter in OWL.
>
> I don't think so, but hey, on to the use case.
>
>> I can think of one case that has come up recently in BioPAX. We  
>> have a property XREF, which can be filled with instances of xrefs,  
>> a class with 3 subclasses, one of which is called a  
>> unificationXref. Suppose we would like to express that when there  
>> is a unificiationXref as the value of an XREF, it will uniquely  
>> determine the instance on which the relationship is asserted.
>
> Er...do you mean that unificationXref is a subclass of max 1  
> inverseOf(XREF)?

I don't think so, but it's possible I'm confused. I think I mean that  
if i1,i2 are some instances in the domain of XREF and x1 is a  
unificationXref, and (i1  XREF x1) then if (x1 inverseOf(XREF) i2)  
then sameAs(i1,i2)

>
>> It might be nice to be able to say:
>>
>> ObjectProperty(_XREF/unificationXref InverseFunctional)
>
> I don't think it would be nice :)
>
>> Right now I would model this as a second property UNIFICATION- 
>> XREF, a subproperty of XREF, and make sure that whenever I wanted  
>> to assert a unificationXref filler, I used the UNIFICATION-XREF  
>> property, rather than the XREF property.
>
> I think that having to make a names inverseOf is a wart in OWL  
> (instead of being able to use an inverse operator),b ut it seems  
> nicer than this.
>
>> If we want to make the decision late in modeling, the second  
>> choice requires more work than if we had these qualified/ 
>> restricted properties available to work with.
>
> I don't see why you want this in addition to QCRs. Actually, in  
> your use case, I don't see what QCRs have to do with it. It is a  
> bit late thought :)
>
>> Maybe this is already available and I'm not seeing it (please  
>> teach). Or maybe it introduces a reasoning problem which I am not  
>> qualified to discern (please say so).
>>
>> Hope this makes sense. Let me know if I need to clarify.
>
> Please!
>
> Cheers,
> Bijan.
>

From bparsia at isr.umd.edu  Sun Jan 15 23:49:37 2006
From: bparsia at isr.umd.edu (Bijan Parsia)
Date: Sun Jan 15 23:49:37 2006
Subject: [OWL] Qualified/restricted properties
In-Reply-To: <AF1B389E-6199-4DF7-94C4-0C2891FC3D20@gmail.com>
References: <5EB416EC-E9BC-4F40-917F-59DF98179BD3@gmail.com>
	<0caa10fce4443cdc2d2ebdbf8972cc61@isr.umd.edu>
	<AF1B389E-6199-4DF7-94C4-0C2891FC3D20@gmail.com>
Message-ID: <23b89ac3dc1415b2bf4390f27eee8f0d@isr.umd.edu>

On Jan 15, 2006, at 11:35 PM, Alan Ruttenberg wrote:

> On Jan 15, 2006, at 10:46 PM, Bijan Parsia wrote:
>
>> On Jan 15, 2006, at 9:30 PM, Alan Ruttenberg wrote:
>>
>>> In a qualified cardinality restrictions we say something like
>>>
>>> Property X can have at most n fillers of Class C
>>>
>>> Another way to think about this is that there is an anonymous 
>>> property
>>>
>>> _P/C == The fillers of Property X which are of class C
>>
>> Not quite since the restriction allows for there to be other fillers 
>> of X. So you capture part, but not all.
>
> I think what I meant to say is "there is an anonymous subproperty of 
> X: _X/C == ..."
> Does this address your comment?

Perhaps. But I don't get it. Thinking semantically for a moment, 
suppose you have a two place relation X. You notice that for every 
value of the first argument, there are at most n values for the second 
argument that are also members of the one place relation C. Now, it is 
true that there is a subrelation of X such that all of its second 
argument are also members of C. But, so? There's nothing special about 
the number of second place Cs in X that fits in here. This is true for 
any relation. (Of course, this is *in a specific* interpretation.)

>> Well, you can't exactly model qualified restrictions in OWL (the 
>> subproperty trick runs into trouble with max cardinalities, consider:
>>
>> 1)	E = max p 2 C & max p 2 D
>>
>> If you model it as:
>>
>> 2)	E = max pC 2 & max pD 2.
>> 	pC sub p. pC range C.
>> 	pD sub p. pD range D.
>>
>> In 1), you can only have 2 ps to a C. In 2) you can have an arbitrary 
>> number of ps to a C, not what you wanted.).
>
> I think, possibly, that this addresses the case in OWL 1.0. I should 
> have said that I am referring to a baseline of OWL 1.1 which will have 
> real QCRs.

I was trying to show why you can't simply encode QCRs into OWL 1.0. 
Which is the only use for subproperties in this, anonymous or not, that 
I can see. I'm missing something, clearly.

[snip]
>>> I can think of one case that has come up recently in BioPAX. We have 
>>> a property XREF, which can be filled with instances of xrefs, a 
>>> class with 3 subclasses, one of which is called a unificationXref. 
>>> Suppose we would like to express that when there is a 
>>> unificiationXref as the value of an XREF, it will uniquely determine 
>>> the instance on which the relationship is asserted.
>>
>> Er...do you mean that unificationXref is a subclass of max 1 
>> inverseOf(XREF)?
>
> I don't think so, but it's possible I'm confused. I think I mean that 
> if i1,i2 are some instances in the domain of XREF and x1 is a 
> unificationXref, and (i1  XREF x1) then if (x1 inverseOf(XREF) i2) 
> then sameAs(i1,i2)
[snip]
Ok. So suppose:
	unificationXref is a subclass of max 1 inverseOf(XREF)

Now consider:
	i1 XREF x1.
	i2 XREF x1.
	x1 type unificiationXref

These together entail sameAs(i1, i2).

So, it seems like we mean the same thing. No need to QCRs for this, 
just inverses. So I don't see the connection.

Cheers,
Bijan.

From alanruttenberg at gmail.com  Mon Jan 16 00:06:13 2006
From: alanruttenberg at gmail.com (Alan Ruttenberg)
Date: Mon Jan 16 00:06:17 2006
Subject: [OWL] Qualified/restricted properties  (NOT!)
In-Reply-To: <23b89ac3dc1415b2bf4390f27eee8f0d@isr.umd.edu>
References: <5EB416EC-E9BC-4F40-917F-59DF98179BD3@gmail.com>
	<0caa10fce4443cdc2d2ebdbf8972cc61@isr.umd.edu>
	<AF1B389E-6199-4DF7-94C4-0C2891FC3D20@gmail.com>
	<23b89ac3dc1415b2bf4390f27eee8f0d@isr.umd.edu>
Message-ID: <94C5054B-C280-4734-A657-0120B7937DBC@gmail.com>

Looks like we do. Thanks for teaching me something!
-Alan

On Jan 15, 2006, at 11:49 PM, Bijan Parsia wrote:

> Ok. So suppose:
> 	unificationXref is a subclass of max 1 inverseOf(XREF)
>
> Now consider:
> 	i1 XREF x1.
> 	i2 XREF x1.
> 	x1 type unificiationXref
>
> These together entail sameAs(i1, i2).
>
> So, it seems like we mean the same thing. No need to QCRs for this,  
> just inverses. So I don't see the connection.
>
> Cheers,
> Bijan.

From Christine.Golbreich at univ-rennes1.fr  Tue Jan 17 04:38:20 2006
From: Christine.Golbreich at univ-rennes1.fr (Christine Golbreich)
Date: Tue Jan 17 04:38:27 2006
Subject: [OWL] meta
Message-ID: <097d01c61b49$c1447710$db9ffea9@LUZ>

Hi

As  there was no answer to my earlier email of January 11, I wonder
whether I was not clear or whether I am confused. So I try to reformulate.

Is it possible to have some clarifications?

The current OWL1.1. proposal says in 2.4:
"with this change, *non-annotation* property can be placed on any names".
Does this change allow to place OWL 1.1 *non-annotation* i.e. OWL1.1
ObjectProperty or DataTypeProperty  on any names ?
Or is there in OWL 1.1 a limitation on property names that prevents
ObjectProperty or DataTypeProperty  to be used as AnnotationProperty  and
allows only properties that have been declared as AnnotationProperty to be
placed on any names with values?

"the extended version of annotation (to incorparate non-annotation
values and types)"  may be confusing ...
Does  'non-annotation' values refers to OWL or OWL 1.1  non annotation ?
Does extended version of 'annotation' refers to OWL 1.1 ?
Does it mean that any object, data property, annotation, as soon as it is
placed on a name with value becomes an OWL 1.1 annotation ?

As punning (without restrictions on the property that can be placed at
names), may lead to some big mixture of class and metaclass notions among
users and also to important change, at least for existing biomedical
ontologies, I wonder whether it would not be preferable for the
moment to have only a very limited form of metamodelling in OWL 1.1  that
enables to overcome 'some' limitations of the OWL annotations (allow axioms
such as functional, inversfunctional, domain, range, but only for
AnnotationProperty placed on a name), and to delay more powerful
metamodelling to later.

As asked in my previous email : would it be possible to have 2 steps?

- at a first step with OWL1.1  to allow only properties asserted as
AnnotationProperty (and not *non-annotation* properties) to be placed on any
names, while keeping the punning + syntactic sugar so as to overcome OWL
previous limitations for annotation (i.e. OWL 1.1 AnnotationProperty applies
to the use of the name as an individual) ?
- at a second step, design more powerful metareasoning, but with solutions
preventing some of the possible defaults resulting from punning + syntactic
sugar when applied with no restriction

would it make sense ?

Again, sorry for any possible confusion.

Christine



----- Original Message -----
From: "Christine Golbreich" <Christine.Golbreich@univ-rennes1.fr>
To: <owl@lists.mindswap.org>
Sent: Wednesday, January 11, 2006 12:27 PM
Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1

...

Would it be possible to have 2 steps. First in OWL1.1 adopt the
punning option but only for "metadata", restricting its use to properties
declared as annotatationProperty (although I am aware it does not provide
any guarentee that users may not declare other properties as annotation.
Perhaps introducing a class conceptName would make it more clear for users
(?), and to look for another solution for more general metamodelling next in
OWL.
2 ?


From Alan.Rector at manchester.ac.uk  Wed Jan 18 09:22:21 2006
From: Alan.Rector at manchester.ac.uk (Alan Rector)
Date: Wed Jan 18 09:19:15 2006
Subject: [OWL] [1.1] minor modification of OWL 1.1 syntax document
In-Reply-To: <20060112.081425.38064928.pfps@research.bell-labs.com>
References: <20060112.081425.38064928.pfps@research.bell-labs.com>
Message-ID: <E5EAB146-BBC8-442E-B65E-7BC68D0EE03A@manchester.ac.uk>

All

Many thanks to everyone, especially Peter, for so much hard work.  As  
I come back from first total immersion in other things and then  
pleasant holiday, I am trying to catch up.

I am unclear about several things in this proposed syntax as of 12  
Jan and would like to check them back against some of the original  
requirements and against my queries concerning the representation of  
meta information about the arftefact and higher order information  
about the entities represented by the artefacts.

i)	The status of annotation properties, comments and puns and whether  
puns in effect replace annotation properties.

i.a) As I understand it, annotation properties remain, although many  
of their functions might be  overtaken by puns. I take this to be the  
effect of changing rdfs:label and rdfs:comment from annotation  
properties to datatype properties.

i.b) It is unclear to me, given i.a) , what function the annotation  
properties retain.

i.c) The notion of a "comment" has been introduced, separate from  
rdfs:comment, which is treated simply as white space.  I presume that  
parsers would still be expected to note or make available comments in  
some form.  I presume that comments differ from values of  
rdfs:comment in that they do not apply explicitly to any construct  
but are just general explanatory content.  They would therefore not  
be appropriate for annotating axioms - or would they.  The ability to  
annotate individual axioms was one of the prime requirements for  
managing ontology evolution and modularisation which I hoped would be  
achieved by OWL 1.1 - little did I know how difficult it would be.   
Or is the assumption that the referents of comments will be implicit  
by their position? If so, this seems to me worriesome, although  
perhaps a necessary compromise.

i.d) I am worried by the sentence (end 2.4): "A class, datatype, or  
propeprty axiom with an annotation is really just syntactic sugar for  
an extra individual axiom for the name with just the annotation".   
This would seem to confirm that there is still no way to put comments  
on subclassOf or type() axioms.

i.e) One of the primary requirements for richer metadata was that  
annotation properties should have domains, ranges, and subproperties  
in order to organise them more effectively.  I am unclear as to  
whether this requirement is overtaken because the use of punning  
makes annotation properties vestigial or simply left out.


2) I agree that making invalid combinations of data types  
unsatisfiable rather than an error is in the spirit of other OWL  
constructs, but I think it is likely to make the language harder to  
use.  Certainly, I would want any tool to scan for such constructs  
and return warnings/errors rather than being classified as  
unsatisfiable and thereby having the potential to propagate  
unsatisfiability throughout the ontology.   Is there a clear  
statement of which datatypes are compatible someplace in the XML  
literature?


Regards

Alan





On 12 Jan 2006, at 13:14, Peter F. Patel-Schneider wrote:

> The OWL 1.1 syntax document, available at
> 	http://www-db.research.bell-labs.com/user/pfps/owl/syntax.html
> has been changed (very) slightly.  The wording has been changed in  
> a few places
> to make the intent clearer.  Annotations have been added to the  
> syntax for
> datatype axioms to fix a minor bug in the syntax.
>
> peter
> _______________________________________________
> OWL mailing list
> OWL@lists.mindswap.org
> http://lists.mindswap.org/mailman/listinfo/owl

-----------------------
Alan Rector
Professor of Medical Informatics
Department of Computer Science
University of Manchester
Manchester M13 9PL, UK
TEL +44 (0) 161 275 6188/6149
FAX +44 (0) 161 275 6204
www.cs.man.ac.uk/mig
www.clinical-esciences.org
www.co-ode.org

From pfps at research.bell-labs.com  Wed Jan 18 09:55:14 2006
From: pfps at research.bell-labs.com (Peter F. Patel-Schneider)
Date: Wed Jan 18 09:55:21 2006
Subject: [OWL] [1.1] minor modification of OWL 1.1 syntax document
In-Reply-To: <E5EAB146-BBC8-442E-B65E-7BC68D0EE03A@manchester.ac.uk>
References: <20060112.081425.38064928.pfps@research.bell-labs.com>
	<E5EAB146-BBC8-442E-B65E-7BC68D0EE03A@manchester.ac.uk>
Message-ID: <20060118.095514.111628189.pfps@research.bell-labs.com>

From: Alan Rector <Alan.Rector@manchester.ac.uk>
Subject: Re: [OWL] [1.1] minor modification of OWL 1.1 syntax document
Date: Wed, 18 Jan 2006 14:22:21 +0000

> All
> 
> Many thanks to everyone, especially Peter, for so much hard work.  As  
> I come back from first total immersion in other things and then  
> pleasant holiday, I am trying to catch up.
> 
> I am unclear about several things in this proposed syntax as of 12  
> Jan and would like to check them back against some of the original  
> requirements and against my queries concerning the representation of  
> meta information about the arftefact and higher order information  
> about the entities represented by the artefacts.
> 
> i)	The status of annotation properties, comments and puns and whether  
> puns in effect replace annotation properties.
> 
> i.a) As I understand it, annotation properties remain, although many  
> of their functions might be  overtaken by puns. I take this to be the  
> effect of changing rdfs:label and rdfs:comment from annotation  
> properties to datatype properties.
> 
> i.b) It is unclear to me, given i.a) , what function the annotation  
> properties retain.

Historical only.  

> i.c) The notion of a "comment" has been introduced, separate from  
> rdfs:comment, which is treated simply as white space.  I presume that  
> parsers would still be expected to note or make available comments in  
> some form.  I presume that comments differ from values of  
> rdfs:comment in that they do not apply explicitly to any construct  
> but are just general explanatory content.  They would therefore not  
> be appropriate for annotating axioms - or would they.  

It depends on what you want here.   If you don't need annotations on axioms to
have semantic import, then comments do the job just right.

> The ability to  
> annotate individual axioms was one of the prime requirements for  
> managing ontology evolution and modularisation which I hoped would be  
> achieved by OWL 1.1 - little did I know how difficult it would be.   
> Or is the assumption that the referents of comments will be implicit  
> by their position? If so, this seems to me worriesome, although  
> perhaps a necessary compromise.

Well, how else would one "associate" comments with pieces of syntax?

If you want a heavy-weight means of associating something that has semantic
import to pieces of syntax, then you are playing in the middle of a minefield -
you have to exercise great care or something is likely to blow up and injure
you.

> i.d) I am worried by the sentence (end 2.4): "A class, datatype, or  
> propeprty axiom with an annotation is really just syntactic sugar for  
> an extra individual axiom for the name with just the annotation".   
> This would seem to confirm that there is still no way to put comments  
> on subclassOf or type() axioms.

Well annotations don't do the trick, and were never intended to do the trick.
Comments are all that you have for this purpose.

> i.e) One of the primary requirements for richer metadata was that  
> annotation properties should have domains, ranges, and subproperties  
> in order to organise them more effectively.  I am unclear as to  
> whether this requirement is overtaken because the use of punning  
> makes annotation properties vestigial or simply left out.

If you want "annotations" with domains, ranges, etc., then use the punning
facilities.

> 2) I agree that making invalid combinations of data types  
> unsatisfiable rather than an error is in the spirit of other OWL  
> constructs, but I think it is likely to make the language harder to  
> use.  

How so?  The lack of "errors" gives OWL a nice consistency, in my view.

> Certainly, I would want any tool to scan for such constructs  
> and return warnings/errors rather than being classified as  
> unsatisfiable and thereby having the potential to propagate  
> unsatisfiability throughout the ontology.   

Agreed.  UI tools should do some sanity checking, including some checking for
inconsistent data usage.

> Is there a clear  
> statement of which datatypes are compatible someplace in the XML  
> literature?

No.  See the SWBP working group mail logs for lots of discussion on this topic.

> Regards
> 
> Alan

peter
From pfps at research.bell-labs.com  Tue Jan 31 11:24:13 2006
From: pfps at research.bell-labs.com (Peter F. Patel-Schneider)
Date: Tue Jan 31 11:24:18 2006
Subject: [OWL] meta
In-Reply-To: <097d01c61b49$c1447710$db9ffea9@LUZ>
References: <097d01c61b49$c1447710$db9ffea9@LUZ>
Message-ID: <20060131.112413.84723130.pfps@research.bell-labs.com>

[I'm not sure whether I responded to this before, but just in case not ....]

From: "Christine Golbreich" <Christine.Golbreich@univ-rennes1.fr>
Subject: [OWL] meta
Date: Tue, 17 Jan 2006 10:38:20 +0100

> Hi
> 
> As  there was no answer to my earlier email of January 11, I wonder
> whether I was not clear or whether I am confused. So I try to reformulate.
> 
> Is it possible to have some clarifications?
> 
> The current OWL1.1. proposal says in 2.4:
> "with this change, *non-annotation* property can be placed on any names".
> Does this change allow to place OWL 1.1 *non-annotation* i.e. OWL1.1
> ObjectProperty or DataTypeProperty  on any names ?

Yes, this is the intent.

> Or is there in OWL 1.1 a limitation on property names that prevents
> ObjectProperty or DataTypeProperty  to be used as AnnotationProperty  and
> allows only properties that have been declared as AnnotationProperty to be
> placed on any names with values?

No.  

> "the extended version of annotation (to incorparate non-annotation
> values and types)"  may be confusing ...
> Does  'non-annotation' values refers to OWL or OWL 1.1  non annotation ?
> Does extended version of 'annotation' refers to OWL 1.1 ?
> Does it mean that any object, data property, annotation, as soon as it is
> placed on a name with value becomes an OWL 1.1 annotation ?

All the documents refer to OWL 1.1.

> As punning (without restrictions on the property that can be placed at
> names), may lead to some big mixture of class and metaclass notions among
> users and also to important change, at least for existing biomedical
> ontologies, I wonder whether it would not be preferable for the
> moment to have only a very limited form of metamodelling in OWL 1.1  that
> enables to overcome 'some' limitations of the OWL annotations (allow axioms
> such as functional, inversfunctional, domain, range, but only for
> AnnotationProperty placed on a name), and to delay more powerful
> metamodelling to later.

Hmm.  Punning is actually quite weak - essentially it liberates the syntax but
has no significant semantic consequences.

> As asked in my previous email : would it be possible to have 2 steps?
> 
> - at a first step with OWL1.1  to allow only properties asserted as
> AnnotationProperty (and not *non-annotation* properties) to be placed on any
> names, while keeping the punning + syntactic sugar so as to overcome OWL
> previous limitations for annotation (i.e. OWL 1.1 AnnotationProperty applies
> to the use of the name as an individual) ?

One could so restrict oneself.

> - at a second step, design more powerful metareasoning, but with solutions
> preventing some of the possible defaults resulting from punning + syntactic
> sugar when applied with no restriction

> would it make sense ?

My view of the transition is that punning is the first step.  Any semantically
significant change would be the next step.

> Again, sorry for any possible confusion.
> 
> Christine

peter



> ----- Original Message -----
> From: "Christine Golbreich" <Christine.Golbreich@univ-rennes1.fr>
> To: <owl@lists.mindswap.org>
> Sent: Wednesday, January 11, 2006 12:27 PM
> Subject: Re: [OWL] annotations and meta-modelling in OWL 1.1
> 
> ...
> 
> Would it be possible to have 2 steps. First in OWL1.1 adopt the
> punning option but only for "metadata", restricting its use to properties
> declared as annotatationProperty (although I am aware it does not provide
> any guarentee that users may not declare other properties as annotation.
> Perhaps introducing a class conceptName would make it more clear for users
> (?), and to look for another solution for more general metamodelling next in
> OWL.
> 2 ?