Ian Bailey wrote: (01)
> The SWeb, as far as
> I can deduce, is doing for data what HTML did for hypertext...i.e. allowing
> relationships to go outside the confines of the computer the data sits on. (02)
Now there's a definition. :-o (03)
> If you treat RDF and RDFS as simple modelling syntaxes (that's all they
> are), you can use them quite nicely to represent your ontology - even an
> extensional, 4D, higher-order ontology. It's less easy to do that with OWL,
> as that language tries to be a bit more clever (and as always with things
> that try to be too clever, shoots itself in the foot). It's just a syntax -
> the semantics should be in your ontology, not in the language you choose to
> publish it in. (04)
This is false, and it misses an important point. (05)
RDF introduces a minimal modeling vocabulary, which allows the
expression of essentially arbitrary relations and logical statements.
OWL/DL introduces a much richer vocabulary, but removes certain RDF
vocabulary elements that allow the expression of arbitrary logical
statements (e.g., "implies"). What that does is to limit the kinds of
statements you can make in OWL. (And those limitations guarantee that
certain reasoning algorithms will terminate in boundable time, which is
not true of RDF.) In each case, the base vocabulary has a (strongly)
specified semantics and that semantics enables the interpretation of
sentences in the language. It is not "just a syntax". (06)
The "ontology" is the knowledge model captured in some language. The
semantics in your ontology are those you specify in the axioms
(statements) you write about the terms you introduce, coupled with the
semantics of the terms of the base language that you use in the
sentences. There are no additional semantics "in" your ontology. (07)
There may be much more semantics in your knowledge model, and you may
try to share those semantics by writing lots of natural language text in
the margins, but those semantics are not captured "in" the ontology. (08)
So the language you choose, and the base vocabulary and grammar it gives
you, are very important to the knowledge that your ontology captures. (09)
> We use UML as the master for the IDEAS ontology. Again, if
> you use it carefully (in this case we have a UML profile that exactly
> replicates the ontic categories in the IDEAS Foundation), it's fine. (010)
That is, you have created a "UML profile" -- a new language that uses
UML drawing conventions -- in which to write your ontology. That is not
at all the same thing as using UML symbols with exactly the formal (and
intentionally ambiguous) definitions in the UML standard; it's a
different language. (011)
> Even
> more shocking, we use the UML to auto-generate an RDBMS implementation of
> IDEAS and an RDF Schema of the IDEAS ontology. These languages are just
> tools of the trade. Use them carefully and they'll do the job. (012)
And RDF models are commonly represented in triple stores, which has
nothing to do with either knowledge or semantics. (013)
> I'm not wedded to the semantic web. I don't think there is even a common
> consensus for what it is. However, don't dismiss RDF and RDFS for that
> reason. They're just a neat(ish) way to express your models in XML (be they
> data models, ontologies, taxonomies, etc.) that lends itself well to
> distributed data. (014)
Yes. RDF is a "neat(ish)" way to _express_ your models, because its
vocabulary and grammar give you that "expressiveness". (015)
> All implementation will involve compromise. If you don't think so, you've
> never worked on a commercial system. The trick is to implement in such a way
> that you do not restrict your ontology's expressiveness.
> ... a few (acceptable) work-arounds can get you a long way. (016)
I'm trying to parse "restrict your ontology's expressiveness".
An ontology is partly a vocabulary for use with an existing basic
vocabulary and grammar to construct assertions. And the other part is a
set of such assertions, which may be taken as "definitive" of the
vocabulary terms. It is the combination of the underlying grammar (e.g.
RDF) and your vocabulary that creates the expressiveness. (017)
The trick is to do all that in a way that captures the knowledge you
intend to share. A few circumlocutions may obfuscate the concept for
humans, but they can still correctly capture the intended knowledge.
Other "work-arounds" may actually add nonsense or lose knowledge. (018)
> The moral of this story is that you should concentrate on building a good
> ontology rather than whining about the languages that can be used to do it.
> It's a poor workman who blames the tool. (019)
But you can't make a silk purse from a sow's ear.* A good workman
chooses the proper tools for the job. There are many knowledge
engineering languages out there -- choose one that allows you to capture
the knowledge you mean and enables you to perform the reasoning you need
to do. (020)
-Ed (021)
*I am indebted to a mutual acquaintance of Ian's and mine for the
Russian version of this proverb (Iz govna ne sdyelat' konfyetku) =
you can't make candy from manure. (022)
--
Edward J. Barkmeyer Email: edbark@xxxxxxxx
National Institute of Standards & Technology
Manufacturing Systems Integration Division
100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528
Gaithersburg, MD 20899-8263 FAX: +1 301-975-4694 (023)
"The opinions expressed above do not reflect consensus of NIST,
and have not been reviewed by any Government authority." (024)
_________________________________________________________________
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 (025)
|