Two notes resent from the ontology-summit list. (01)
-------- Original Message --------
Subject: Re: [ontology-summit] Invitation to a brainstorming call for
the 2011 Ontology Summit
Date: Wed, 15 Dec 2010 09:22:50 -0500
From: John F. Sowa (02)
Ian, (03)
I apologize for using the word 'hate': (04)
> I'm even more surprised that someone who apparently likes FOL
> "hates" these particular fragments. (05)
I have never "hated" description logics. And as you know, the ICCS
conferences include a number of DL aficionados who have invited you,
Franz Baader, Boris Motik, and other leaders in the DL community. (06)
And I have the highest regard for research on computational complexity,
decidability, and related issues -- including ways of addressing those
issues in practical computation. (07)
> In fact, you could think of OWL as being a design pattern, which
> you apparently like, as opposed to a fragment, which you don't like. (08)
I have always thought of DLs as a design pattern, but that is not
the message that comes across in any of the OWL documentation.
Instead, many people who know nothing about decidability parrot
the words "Oh, you can't use FOL because it's undecidable." (09)
I'm not blaming you for that misunderstanding, but that's how many
people interpret all the hype and confusion that surrounds the
Semantic Web. That view is reinforced by the layer cake, in which
the only reasoning tool available is OWL. There is a box labeled
RIF, but that box is not yet supported by any usable tools. (010)
The impression that OWL is the Only Web Logic is further enhanced
by the directions that OWL 2 has taken to include a rule language
and to spawn a variety of subsets, such as SKOS. There is nothing
wrong with those subsets, but they reinforce the impression that
OWL is necessary and sufficient for reasoning on the Web. (011)
> Coming back to the OWL -v- FOL question, I think that much of
> the "problem" arises from fundamental differences in how we view
> the design and use of ontologies. (012)
We both understand the issues very, very well. My position is
summarized in the title of a recent paper: "Two paradigms are
better than one, and multiple paradigms are even better." (013)
Among those paradigms are induction, deduction, and abduction
with many different subsets and supersets of FOL and methods
of learning, analogy, statistics, data mining, etc., etc., etc. (014)
I also mentioned UML because it comes closer to being a multi-
paradigm collection of diagrams plus FOL represented as a mosaic,
rather than a layer cake. I would be very happy to see OWL and
many other AI tools presented as components in such a mosaic. (015)
But the "decision makers", who never see anything more detailed
than slides with pretty pictures, don't have a clue about them.
And in public lectures, the OWL proponents omit or gloss over
the qualifications, even for students who are willing and able
to learn the details. Your research publications address the
details, but the decision makers dispense funding on the basis
of pretty pictures, and the students go where the funding goes. (016)
> Hopefully you can try to see OWL in a similar light -- it is
> raising the profile of ontologies, encouraging the use of
> (a fragment of) FOL as an ontology language, and providing you
> with a ready source of "customers" ripe for "upgrading". (017)
I agree that OWL is raising the profile of ontologies. And I was
one of the "early adopters" of the word 'ontology' in my 1983 book.
But that word has been misused in so many confusing and misleading
ways that I almost regret I ever used it. (018)
In any case, I do have a high regard for the research that you and
your colleagues have been doing. But my major quarrel is with the
way OWL has been advertised and sold to people who don't understand
the much wider range of issues. (019)
John (020)
-------- Original Message --------
Subject: Re: [ontology-summit] Invitation to a brainstorming call for
the 2011 Ontology Summit
Date: Wed, 15 Dec 2010 11:15:28 -0500
From: John F. Sowa (021)
Dear Matthew, Leo, Jack, and Cecil, (022)
LO:
> I recall in the late 1980s, papers in the relational database
> community that began to address semantic data models. Even these
> were not ontologies. The database folks did not address only real
> world semantics, but included many system and database level
> constructs -- as data models still do. (023)
MW:
> I quite agree that all this work was very significant, which largely
> grew out of what was called in the '70s Artificial Intelligence.
> And I'm quite happy that this is described as ontology. But what I
> do not think is justified is a terminology land grab that says that
> this is all that can be called ontology in a computational sense. (024)
I agree with Matthew and Leo. There's a difference between an ontology
and a data model, but that point was recognized in the 1970s, with the
ANSI/SPARC three-schema model. The conceptual schema was described
in different ways by different people, but many of them defined it
in a way that is indistinguishable from modern ontologies. (025)
For anybody who wants a "time stamp" on ideas, I recommend the following
collection of position papers that were presented in 1980: (026)
http://www.informatik.uni-trier.de/~ley/db/conf/sigmod/pingree80.html
Workshop on Data Abstraction, Databases and Conceptual Modelling (027)
This workshop brought together a few dozen people from the DB, AI,
and programming language communities. Pat Hayes and I are two of the
AI people who still participate in Ontolog Forum. (028)
JR:
> In one case, early 1970's, Simula educed a new way of representing
> intelligence spacecraft in a computational device. Then current
> practice was to write programs representing the spacecraft operations
> and limits. (029)
Yes. Simula 67 was the world's first object-oriented programming
language, and in some respects it was more advanced than some current
versions. Some of the talks at that 1980 conference related those
techniques to AI and DB -- in ways that are called ontology today. (030)
COL:
> Can you explain your statement "You can translate any OWL ontology
> to UML, but not vice-versa."
>
> The OMG Ontology Definition Metamodel specification seems to contradict
> your statement. Table 16.12 of the specification lists the OWL features
> with no equivalent UML feature as :
>
> Thing, global properties, autonomous individual' allValuesFrom,
> someValuesFrom, SymmetricProperty, TransitiveProperty, Classes as
> instances, disjointWith, complementOf (031)
I was the one who stated the original quotation above. The passage
from the OMG document didn't make a clear distinction between logic
and ontology. (032)
UML consists of a variety of diagrams types, each of which represents
one subset of logic together with some built-in ontology. But UML
supplements the diagrams with the Object-Constraint Language (OCL),
which can express full first-order logic. (033)
Class diagrams are the most widely used diagram type, and they
represent a very large part of what people represent in OWL, but
I agree that OWL can represent more detail about some important
aspects. (034)
But the other UML diagrams represent many aspects that OWL cannot
represent, and OCL is a superset of what OWL can express. (035)
In any case, the greatest strength of the UML diagrams is their
highly readable way of representing multiple aspects of a system
in a way that connects with design and development methodologies.
I prefer to use the word 'mosaic' for the UML languages and to
treat OWL as one component that could be added to the UML mosaic. (036)
But I would emphasize that OWL should be treated as just one
component. The UML scope is much broader and more inclusive
than OWL. (037)
John (038)
_________________________________________________________________
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 (039)
|