Re: Is relational theory incomplete?

"Tom Hester" <tom_at_metadata.com> wrote in message news:a033$3fafe73a$45033832$30156_at_msgid.meganewsservers.com...
> Yes, I am very familiar with the catalogs. Compared to other, more
> semantically rich data models

A catalog is a catalog and not a data model. To make any sense of the above, I must assume you postulate a data model that is semantically richer than the relational model. Name one data model that is as semantically rich as the relational model--let alone richer.

>, the catalogs represent a very sparse amount

If a dbms fails to support domains or if a dbms does not describe domains using values in relations, the dbms suffers from significant relational infidelity. It strikes me as perverse and misleading to criticize the relational model for that product's lack of relational fidelity. Your assertion that the relational model prevents adequate metadata is plainly wrong and exhibits profound ignorance of the fundamentals of data management.

> You apparently misread my original reply. I infact said that Gödel did not
> invalidate the relational model.

Yes, I know that. However, I doubt anyone reading your reply would gain any useful information from it. The relational model is as incomplete as every other consistent formalism, which is to say some information external to the formalism exists. This is equivalent to saying that some part of the conceptual model extends beyond the logical model--nothing more and nothing less.

> You identify intension with conceptual.

As relates to computing, the conceptual level of discourse encompasses all information as understood by human beings where the logical level of discourse encompasses only data represented appropriately for processing or for communicating. You do not have to take my word for this: You can verify it using the ISO/IEC Standard Vocabulary for Information Technology (ISO/IEC 2382-17).

I equate the information external to any consistent formalism with the information at the conceptual level that is not represented for communicating or for processing as I see the formalism as the appropriate representation for communicating or for processing. The relational model is a consistent formalism which necessarily means it is incomplete; however, the incompleteness is irrelevant to the issue whether the relational model is useful for data management.

Given a choice between incomplete and inconsistent, which is the choice Goedel proved one cannot avoid, I will choose incomplete any day for a formal data model, because external information will always exist in any case. Whereas inconsistency will render the logical data model unreliable and incomprehensible.

> That is a great leap and one that
> many analytical philosophers would not agree with. See Montague for
example,
> who proposes that it concepts are extensions.

I suggest the above argument would have greater relevance in comp.math.philosophy where perhaps they use a slightly different standard vocabulary.

> "Bob Badour" <bbadour_at_golden.net> wrote in message
> news:NvGdnbW12t0LQTKiRVn-jw_at_golden.net...
> > Tom,
> >
> > The catalog in an rdbms is entirely meta-data and is a rich source of
> > meta-data. Goedel's proof basically states that the conceptual level of
> > discourse is necessary, which does not in any way invalidate any part of
a
> > logical data model.
> >
> > The combination of conceptual, logical and physical is complete. The
> > relational model is the best known logical data model, and it
specifically
> > limits itself to the logical level of discourse for important reasons
like
> > separation of concerns.
> >
> > Regards,
> > Bob
> >
> > "Tom Hester" <tom_at_metadata.com> wrote in message
> > news:f488$3fafcb95$45033832$28183_at_msgid.meganewsservers.com...
> > > Gödel's proof is of the incompleteness of arithmetic, not relational
> > algebra
> > > (or calculus, or...). Essentially, Gödel demonstrated that a theory
of
> > > arithmetic must contain at least one intensional statement of the
form:
> > > 'this sentence is false'.
> > >
> > > Arithmetic had always been assumed to be purely extensional. Codd's
> > > relational theory was purely extensional. Remember all of the early
> > caveats
> > > on relational theory of the form 'first order function free'. Those
> > caveats
> > > were to insure that the theory was only extensional. To put it in
other
> > > words, it only dealt with sets of things in the real world. A
> relational
> > > model is a first order description of some subset of the real world.
> > >
> > > The cost of this purely extensional restriction is the almost total
lack
> > of
> > > metadata information available in a relational system.
> > >
> > > "mountain man" <hobbit_at_southern_seaweed.com.op> wrote in message
> > > news:xJzrb.4373$aT.2467_at_news-server.bigpond.net.au...
> > > > Did Date ever make reference to Godel's
> > > > incompleteness theorem? If so, how did
> > > > he handle it?
> > > >
> > > > How does relational theory come to terms
> > > > with Godel's incompleteness theorem?
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > Farmer Brown
> > > > Falls Creek
> > > > OZ
> > > > www.mountainman.com.au
> > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
> Received on Mon Nov 10 2003 - 21:36:10 CET