[Top] [All Lists]

Re: [ontolog-forum] Foundation ontology, CYC, and Mapping

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Christopher Menzel <cmenzel@xxxxxxxx>
Date: Tue, 16 Mar 2010 13:30:02 -0500
Message-id: <7623CC27-316C-4A1F-963A-08B0B251A4E2@xxxxxxxx>
On Mar 16, 2010, at 7:15 AM, doug foxvog wrote:
Chris Menzel wrote:
>> On Mar 15, 2010, at 12:05 PM, Cory Casanave wrote:
>>> One comment, re: Adding a contradiction to any consistent theory causes
>>> it to degenerate to the absurd theory.
>>> One pragmatic reality of architectural work is that theories do not
>>> agree. Particularly in the realm of business related modeling there are
>>> clearly different interpretations and opinions about the same thing.  So
>>> we have seemingly conflicting theories but we agree that the theories
>>> describe the same entities that exist or may exist. Such conflicting
>>> theories are normal and even desirable in our discourse.
>> Absolutely.
>>> One capability that a knowledge technology should be able to provide is
>>> the identification of these conflicts and the source of assertions and
>>> conflicts.
>> Yes, that is one thing that one hopes to get out of (among other things)
>> the use of automated reasoning tools.
>>> One thing that this requires is the ability to assert that 2 symbols
>>> represent the same entities or concepts, yet may make very different
>>> (and perhaps conflicting) assertions about these entities. In the OWL
>>> world, sameAs is frequently used for this, but sameAs seems to strong a
>>> commitment.
>> There is nothing that I can see in the semantics of "owl:sameAs" that
>> would enable it to be used to indicate that two different symbols "may
>> make very different (and perhaps conflicting) assertions" about about
>> their denotations.
> What is missing here is the idea of contexts.  Statements made using terms 
>equated by "owl:sameAs" are usually made by different asserters in different 
>contexts. If the different contexts make conflicting assertions, it is the 
>contexts which conflict. Such conflicts can also occur with different users of 
>the same term, or even a single individual who makes statements using the same 
>terms in different contexts. E.g., in one knowledge base might state that 
>dbp:BarakObama is an Illinois senator, while another might state that he is 
>president of the USA.    (01)

Well, sure, Doug, some formal way of representing contexts is necessary for 
solving the problem of conflicting assertions -- as has been discussed here 
many times of the years.  Perhaps you were simply making that point again, but 
just to be clear, my post was only intended to address the claim that there is 
something about the semantics of owl:sameAs that enables it to be used to 
indicate that two different symbols "may make very different (and perhaps 
conflicting) assertions" about about their denotations.  There just isn't; the 
semantics is straight up identity.  Your suggested solution in fact makes my 
point.  Some *additional* mechanisms or practices over and above the bare 
semantics of OWL are necessary for avoiding the conflicts that can exist 
between ontologies arising from different contexts.    (02)

>> From the OWL reference document:  "owl:sameAs is used
>> to state that two URI references refer to the same individual."  If, in
>> fact, two URI references are used to say things that cannot be true of one
>> and the same thing, then declaring one to be "owl:sameAs" the other will
>> simply lead to logical inconsistency, i.e., absurdity, full stop.  The
>> only way to avoid it is to withdraw the assertion or alter what is said
>> about the individual in question by means of those URI references.
>>> EquivelentClass is closer,...
>> I don't see how.  "<A owl:equivalentClass B>" means nothing more or less
>> than that the classes A and B (or, if you like, the classes referred to by
>> the names "A" and "B") have the same members.
> Actually, that they have the same members in the same context.      (03)

Actually, no.  There are no contexts in the semantics of OWL.  Of course, OWL 
is *used* in different contexts to build different knowledge bases and, hence, 
e.g., the same name can end up being used for different classes and different 
names can be used for the same class.  Again, that is just the problem: 
mechanisms and practices have to be introduced to deal with the inconsistencies 
that can arise if such knowledge bases are simply merged.  (A simple initial 
solution that can be implemented without any extensions to OWL, of course, is 
just to prefix every name in each database with a unique identifier for that 
database.)    (04)

>> Again, there is nothing
>> there that could be exploited to indicate that "very different (and
>> perhaps conflicting) assertions" are being made using those names.
> Again, this is only true when restricted to the same context.    (05)

Again, there is no formal notion of context in OWL.    (06)

>>> In summary, when using ontologies to represent architectures we should
>>> embrace conflict ...
>> I think there is an important ambiguity here.  One should certainly
>> "embrace conflict" in the sense that one should acknowledge the fact that
>> there different outlooks and opinions and that dialog between communities
>> with different outlooks and opinions can lead to rich and fruitful
>> results.  One should certainly *not* embrace conflict in the sense of
>> allowing logical inconsistency within an ontology.
> Note that reasoning need not use but a single ontology.  Reasoning can use a 
>vocabulary ontology to define the classes and relations which it uses, a 
>theory ontology to define rules about those classes and relations, one data 
>ontology to specify instances of those classes being considered, and a second 
>data ontology specifying the context, including instances of the relations 
>applied to the class instances and spatio-temporal restrictions on the 
>statements.    (07)

I have no idea how this is relevant.  But let me simply rephrase the point: 
With regard to any of the various ontologies one is reasoning upon 
("vocabulary" ontology, "theory" ontology", "data" ontology 1, "data" ontology 
2, etc), one should certainly *not* embrace conflict in the sense of allowing 
logical inconsistency within that ontology.    (08)

>>> and not consider conflicting theories as absurd.
>> Are you suggesting someone has? No one has suggested that, if T1 and T2are 
>*jointly* absurd (i.e., if their union is logically inconsistent),then T1 and 
>T2 are *individually* absurd. The only that would be absurdwould be to merge 
>incompatible theories without resolving the conflict(which, as you note, might 
>require dialog between the advocates of thetheories).
> Agreed.
>>> Conflict is an indication of differing opinions and interpretations, both 
>normal and expected conditions of dialog.
>> Of course.
> It is also an indication of different contexts.    (09)

I thought I said that. ;-)    (010)

-chris    (011)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (012)

<Prev in Thread] Current Thread [Next in Thread>