[topicmapmail] Making ontologies : RDF vs TM

[Murray Altheim]

Dr. Agnes Mühlmeyer-Mentzel wrote:
> Hi,
> 
> thank you for answering, Murray. 
> 
> I agree with you on what you have written about ontologies and
> communication.
> 
> I have some questions to specific parts of your message. As I'm a
> beginner in thinking about ontologies I have a lot of questions.  
> 
> Murray Altheim wrote:
> 
>>Dr. Agnes Mühlmeyer-Mentzel wrote:
[...]
>>>>But rather than
>>>>discuss what ontology *is* (since I'm on very thin ice anyway),
>>>>perhaps we should first ask what the problem is that ontology is
>>>>suggested as a solution for.
> 
> In my opinion, ontologies (in the scope of internet) are expected to
> support communication (in the space of internet). 

Certainly, if we are confining ourselves to web-based or web-targeted
ontologies. But of course the field of Knowledge Representation spans
a much wider area and has a much longer history. When it emerged from
a combination of philosophy, cognitive science, computer science, and
other fields is a matter of some speculation, but it's at least been
around within the field of computing since the 1960's beginnings of
Artificial Intelligence.

But perhaps I'm making your reply more complicated than you intended
it. Another way I can see this is as the distinction between the
concept of "ontology" within philosophy and "ontology" within computing,
as they are certainly related but also two different beasts.

>>>>[E.g., will creating a large structure
>>>>of fixed categories and then placing everything into those categories
>>>>work?
>>>
>>>Hasn't this actually turned into a problem in its own right?]
> 
> Might you put the problem you are thinking on in concrete terms?  

Perhaps by way of example. Early libraries tried to create a whole
bunch of categories, then assign each book to a category. It was
rather quickly discovered that these fixed categories didn't work
very well for the real world, that almost nothing is *about* a
single subject, that the whole idea of "about" is very context-based.
So when in computing, say as in RDF, we see "rdf:about", or in Topic
Maps we see "subjectIndicatorRef", we should be very suspicious, or
at least very careful. Nothing has or is just one subject. The
statement of subject-hood is contextual.

This is what I mean when the whole idea of identity has become a
problem in its own right. If you Google on "nicola guarino identity"
you'll find a number of important documents [1][2] on the subject,
written by one of the few people in the field I trust on this topic.
And then I'm still not sure he's got it right. It's a really tough
epistemological problem. I don't know that it's ideally solvable,
perhaps only pragmatically (and I also mean Pragmatically) solvable.
The problem is again, what to do about context? It's recursive.

>>>May be, building categories and ontologies don't solve a problem in an
>>>optimal manner.
>>>The question is, do I have a choice?
>>
>>You certainly have a choice both in the philosophical/epistemological
>>approach you take, and in the actual tools/implementation. Certain
>>epistemologies (or lack thereof) lead one down cul de sacs, certain
>>tools have built-in deficiencies (both epistemologically and practically,
>>perhaps because these are linked).
> 
> I'm not a philosopher. What form of philosophical approach you are
> thinking on? Is it possible to have an approach without any terminology,
> without categories, without definitions?   

I'm still working on this one, so I'm not claiming any answer here. But
my feeling is that we shouldn't be looking for a set of fixed categories
but rather a particular *approach* to modeling.

I say this because it's patently obvious from a simple look at reality
(rather than say, mathematics) that fixed categories simply don't exist
except in the most simplistic of views, and definitions are exactly
like categories, i.e., they too are contextual, many-layered, recursive.

So the only way I can see to deal with this is to simply stop categorizing,
stop defining, and begin to look at processes of categorizing, processes
of defining. It's the processes that we need to focus on. I think this
is the same thing neuroscience is coming to in trying to understand the
human brain: that the brain is not composed of a bunch of Topics, it is
composed of a bunch of Associations. That Topics don't exist, except as
the confluence of a whole lot of Associations. (I use Topic Map termino-
logy here because I am also emphasizing that I think we can still use
Topic Map technology to do information or "knowledge" modeling; we just
need to alter our approach a bit when talking about subject identity.)

The problem is that almost the entirety of western culture is built
upon thousands of years of thinking about things as subjects, as
categories, as identifiable "types", when I believe reality is telling
us otherwise. If you look into the Lao Tse, written around 300 BCE, you
find an admonishment to avoid such hard and fast rules, such easy
categorizations. We want life to be simple, and the western approach
was to try to come up with a Complete Set of Categories. The eastern
approach was to look at How Things Flow. The difference between the
CCC and HTW approaches to life only began to meld together in the
western intellectual community (if we ignore the wealth of the Ottomans)
in the late 19th to early 20th century with the translation of some
of the eastern classics by intellectual adventurers like Herbert Giles,
James Legge, Arthur Waley and Richard Wilhelm. By the 1950's it had
thoroughly infected western thought, such that I think the great
flourish of the arts and sciences in the latter half of the last
century owes a great deal to this melding. It transformed us. The
industrial revolution brought us "progress", but the East brought
us to a new way of understanding, which is a much greater form of
progress.

>>I've long advocated looking to fields that have demonstrable expertise
>>in categories and taxonomies, such as library sciences. I note that
>>few if any computer-based ontology systems have a built-in classification
>>system, such that trying to navigate through them is usually difficult.
> 
> I'm interested in your experiences, even if they were disappointing.
> What makes it difficult? 

Well, as an example, try browsing the Cyc ontology [3]. You're only looking
at a very small of the upper and middle ontology, and already it becomes
a complete mess (e.g., [4]). The lower ontology is not even listed, and
that's where things get really complicated, just like real life. Not that
the Cyc model is necessarily bad, it just becomes impossible to navigate.

Compare Cyc with a library card catalog system, which is dealing with
many orders of magnitude more objects to categorize. Librarians are our
information experts, not computer scientists. To be fair, Cyc and a
library are doing two different things, and in fact Cyc is taking on
that big task I mentioned above of trying to map the relationships
between things, which the library is not tackling.

>>>Building categories and ontologies is an effort to transform knowledge
>>>into linguistic forms. Linguistic forms may be weak, but they are our
>>>most important forms for interacting and sometimes the only ones.
> 
>>You must note that many scientists, of both the "hard" (e.g., physics)
>>and "soft" (e.g., ethics) varieties, though particularly the former,
>>believe that what they're discovering is a closer approximation to
>>some platonic ideal, some better representation of reality than that
>>of the last generation of scientists. They don't look at this as a
>>form of communication or inter-communication.
> 
> I agree on that, but only partly. 
> Are categorization and making ontologies the problem or 
> is the misusage of categorization and ontologies the problem?

My snide answer is "yes" (though this joke may not translate well
for non-native speakers of English). As I said above, I think the
problem is in the approach. Now, if one's approach is misguided,
any usage will be a misusage.

> Or in addition to it: is the underlying problem the misusage of a
> powerful appointment some scientists hold?
> I think it's very important, to differentiate at this point.

I'm not sure if I understand that question exactly. Though if I
do understand you correctly, yes, this is the very aspect of
science that Arthur Fine was talking about, about pulling the
"hard" science down off of its pedestal and understanding it as
being of the same kind of inquiry as any other.

>>>The problems of ontologies are the problems of language. We have to
>>>solve them by communicating and interacting.
>>
>>Yes, and one of the points I will be making fairly strong in my own
>>thesis is that ontologies are a form of communication, they are a way
>>of stating: "this is what *I* believe." They are not, and I repeat,
>>not expressions of universal, platonic reality. This is a mistake I
>>see commonly repeated. I had dinner with Doug Lenat, inventor of the
>>Cyc ontology, and was pleased to hear him say that Cyc was not such
>>an expression, that it was, in effect, "Doug's Ontology" (now, more
>>correctly, "Cycorp's Ontology").
>>
>>I think that it's important to consider the development of computer-
>>based ontologies as a means of communicating ideas about our
>>understanding of reality, of a specific domain. And that when we
>>allow others to see that ontology, they understand it as something
>>we are *saying*, not an expression of the universal. Now, ontologies
>>can and are being developed by cooperative groups, consortia, etc.,
>>and these are considered agreements between people.
> 
> I agree on that. 
> 
> As it seems there is a competitive situation between those two
> approaches, the idea of an universal ontologie and the idea of a
> communicated ontologie. I am unacquainted with the idea of an universal
> ontologie, but I must admit not to know much about it. 

If you can locate a copy of John Sowa's "Knowledge Representation",
it begins on the front cover with the 13th century's Tree or Porphyry.
Sowa's a funny chap, as while he has an amazingly encyclopedic
understanding of the field, with a grasp of its history that is
remarkable, he does (to me at least) seem very bound to the ideas
of the Greeks. I think he tends to declaim when I say this, but in
reading his two books I can't help but see him as a proponent of a
well-categorized world. And while I can't find the quote, he wrote
into one of the lists we are both on that he thought that the
problems of KR are solvable with just a better understanding, that
we just hadn't cracked it yet.

My feeling is that it isn't crackable, that we'll never have any
kind of universal ontology because there are almost no shared
universals amongst all people. People don't like to hear that,
and the field of KR ignores the problem and just moves on with
finding a solution. Baffles me how you can claim to be representing
something you can't really define.

> But beside of that competitive situation I'm convinced that there is
> plenty of work to achieve a communication with a better understanding. 

Yes, and I think that a greater focus on the aspects of this as a
form of communication will lead to much greater understanding,
moving away from the mechanistic to the humanistic. This is to
my understanding what people like Richard Rorty, Robert Brandom,
and Jurgen Habermas are all about. Rorty seems to be screaming
it in books like "Philosophy and the Mirror of Nature."

> To come back to the specific subject: Ontologies and Knowledge in
> science:
>   
> For a long period of time science has been subdivided into many many
> disciplines. Every discipline has developed its categorization and
> terminology and use everyday words in a specific way.
> Nowadays the scientific community begin to stress the inter-disciplinary
> approach and the possibilities of the internet are forcing this
> approach. As there are a lot of terms that are used in different ways 
> the interdisciplinary approach can only become fruitful, when scientists
> are willing to communicate about this. That is an arduous challenge.

Absolutely. And any technology that allows people to intercommunicate
between these various (and I might add: overlapping) communities, whilst
maintaining the necessary context and understanding with regard to
context-of-discussion, domain-of-discussion, time-of-discussion, etc.
we may see the beginnings of fully context-preserving communication. I
think Topic Maps are a technology capable of doing this.

So throw out that dictionary and begin to use one that maps all known
*usages* of a term rather than a fixed definition. A nightmare? No,
the releasing from a straightjacket, the pulling off of blinders, at
the same time as the words are still shown within their contextual
anchorings, with all the concomitant richness, muddiness and flavour.

> Here an example:
> 
> We have had a interdisciplinary panel discussion about "emotion".
> Sixteen scientists of six disciplines were discussing (at the end: 
> combatting). At the end a woman of the audience asked: "Are you sure,
> you are speaking about the same subject?"  

Heh. I would be hard pressed to understand under what circumstances
sixteen scientists from six disciplines could possibly be intelligently
discussing emotion, without emotion. But yes, this perfectly illustrates
what we've been talking about.

[Sorry about the delay in getting back to you. I've been particularly
behind schedule on things here, and I also was dealing with a family
illness that took my attention away from everything else. One heart
bypass surgery later, things are happily improving daily...]

Murray

[1] http://www.sigmod.org/sigmod/dblp/db/indices/a-tree/g/Guarino:Nicola.html
[2] Guarino, Nicola and Chris Welty. In press. Identity and Subsumption
     http://www.cs.vassar.edu/faculty/welty/papers/
[3] Cyc Reference
     http://www.cyc.com/cycdoc/vocab/vocab-toc.html
[4] objectOfPossessionTransfer
    http://www.cyc.com/cycdoc/vocab/financial-vocab.html#objectOfPossessionTransfer
......................................................................
Murray Altheim                    http://kmi.open.ac.uk/people/murray/
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK               .

   A schoolteacher at a Bush event was on her way to the bathroom when
   she was stopped by a volunteer and told she wasn't welcome. The
   volunteer pointed to her T-shirt and said it was 'obscene'. She and
   her two friends (also wearing the same shirts) were escorted out by
   police officers and threatened with arrest if they did not comply.

           The T-Shirts read: 'Protect Our Civil Liberties'"

   Shutting Them Up
   http://dneiwert.blogspot.com/2004_10_17_dneiwert_archive.html#109830472470609571

   Dismantling Democracy
   http://dneiwert.blogspot.com/2004_10_17_dneiwert_archive.html#109830622308006215