[Top] [All Lists]

Re: [ontolog-forum] What is the role of an upper level ontology?

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: John F Sowa <sowa@xxxxxxxxxxx>
Date: Sun, 19 May 2013 13:42:09 -0400
Message-id: <51990EF1.10900@xxxxxxxxxxx>
William, John, Hans, Matthew, and Hassan,    (01)

My main point is that any upper-level ontology that claims to be
broadly applicable should avoid detailed axioms.  It should be
as neutral as possible with respect to any or all ontologies that
have proved to be useful for any practical purpose.    (02)

For complex reasoning, more detailed axioms are needed.  They can be
stored in a library of reusable modules that can be combined to form
more specialized ontologies or microtheories as needed.    (03)

> But, like anything else in science.  Fallible in application,
> in other words, we might measure wrong, and fallible as assertions.
> In other words, the identity conditions stated may turn out not
> to be quite right, or even altogether wrong.    (04)

Yes.  Fallible theories that are known to be "not quite right"
can still be very useful.  Example:  Newtonian mechanics.  The
upper levels should contain the terms Mass, Energy, and Momentum,
but the detailed axioms should be contained in specialized modules.    (05)

Note that the more detailed modules are likely to have different
identity conditions for terms with the same spelling.  For a
lexical hierarchy like WordNet, it's important to relate all
terms with the same spelling, but it's also important to have
conventions for distinguishing different senses or microsenses.    (06)

> Could you provide examples of a nontransitive part relation or
> maybe a link about this?    (07)

For example, a lock is part of a door, and a door is part of a house.
But a lock is not normally considered part of a house.  Hydrogen is
part of water, and water may be part of a cooling system.  But
hydrogen is not considered part of the cooling system.    (08)

Ingvar Johansson analyzes and discusses these and other examples.
He observes that there is always some reason why the part relation
is not transitive, and it can be represented with a triadic part
relation.  See http://hem.passagen.se/ijohansson/function2.PDF    (09)

> The key point is that the context and associated frame(s) of reference
> used to specify identity should be made explicit (in-line or via indirection
> to some source specifying said context parameters). The NCOIC has developed
> a set of net centric principles, one of which, “entity primacy”, grapples
> directly with this identity issue.    (010)

> Excellent.    (011)

I agree that it's a good ontology for ncoic.org, and it's more general
than many specialized ontologies.  But it is less than universal.    (012)

> Unfortunately, many information system contexts treat the identity they
> assign to entities as the only operable identity. This is usually done
> for very pragmatic reasons, especially in “closed world” contexts.
> Increasingly, such narrow information system contexts are being exposed
> to other and often unexpected contexts (i.e., interoperability issues),
> usually over a network connection.    (013)

Yes.  But it is essential to accommodate legacy systems and anything
that might be developed in the future.  Ignoring them is not an option.    (014)

> This [intransitive part relations] is conflating two things:
> a) Membership of a club to a sports federation, and of a player to a sports 
> b) Whole-part.
> I would argue that it is the membership relation that is not transitive,
> but that it is quite reasonable to create a mereological sum of the players
> of the clubs that are members of a sports club and of a sports federation,
> and that this is transitive.    (015)

That is true.  But some philosophers hope to use mereology as a basis
for defining many similar relations.  Membership is one of them.    (016)

> I was referring to: (1) team is-part-of club and (2) club is-part-of 
> - although (3) not every such clubs' team is-part-of federation...
> One could come up with another condition for a club to be part of a federation
> - say if it has at least one team that qualifies a being part of the 
> (whatever the condition may be).    (017)

This example and proposed solution are similar to the examples that IJ
cited, quoted, and analyzed.  I would also recommend other papers that
analyze these and related issues:  http://hem.passagen.se/ijohansson/    (018)

For example, see http://hem.passagen.se/ijohansson/information7.pdf
> there are good reasons to distinguish between four kinds of is_a relations:
> genus-subsumption, determinable-subsumption, specification, and 
> They behave differently in relation to definitions and so-called inheritance
> requirements.    (019)

This article has a good analysis of many issues that have been debated
in Ontolog Forum.  Among them are the treatment of time, single vs.
multiple inheritance, and realist speech vs. conceptual speech.    (020)

Johansson has not given up all hope of developing a good upper level
ontology, but he has found serious flaws (or at least limitations
and exceptions) in the current proposals.    (021)

My recommendation is to consider a universal upper ontology as
a goal for further research.  For practical applications, any
or all ontologies that have been useful must be accommodated.    (022)

John    (023)

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    (024)

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