Re: Dawn doesn't like 1NF

"Laconic2" <laconic2_at_comcast.net> wrote in message news:<RcGdncR7qd4Z7fjcRVn-pQ_at_comcast.com>...
> "Dan" <guntermann_at_verizon.com> wrote in message
> news:1Qd9d.12265$na.4447_at_trnddc04...
>
> > I'm looking at it from a perspective of being even more fundamental --
> > systems of logic and theory. I am interested in being able to ask a
> > question and being able to retrieve a consistent and complete answer, and
> > perhaps even deduce new information from facts, whether held in a DBMS or
> > not.
>
> Two points with regard to this:
>
> First, "Inference Engines". Inference Engines, or whatever the correct
> name is, are at the very least expert systems, and perhaps even
> artificially intelligent systems, that not only know a lot of facts, can
> make deductions from those facts.

I'm not talking about "expert" systems though. We can derive and infer plenty from a plain old relational system. What is a functional dependency? -- An inference rule really (A->B). What is a foreign key? A transitive inference rule (A->B->C).

>
> I'm not sure, but I think one such "inference engine" was BASE/BALL, an
> early precursor to the DBMS engines that came along later. But it isn't
> just a database. It's really an inference engine that understands a
> database.

Granted, there are inference engines and languages, such as Prolog, that do much much more than relational systems. For me, predicate calculus is sufficient because it is constructed within the confines of 1st order logic.
>
> Here's a link to the "Database Hall of Fame" where Base/Ball, IMS, and
> Nelson/Pick are all credited:
>
> http://ourworld.compuserve.com/homepages/Ken_North/db_hall.htm
>
> Second, when Codd introduced the relational data model in 1970, he didn't
> claim to thave invented it. He made reference to its prior use in inference
> engines, although I think he used a different term. What was novel in the
> 1970 paper was the application of the relational data model to the problems
> of data storing and data sharing that databases and their management systems
> were dealing with.

I have actually read Codd's paper dozens of times, and I agree. But it is by no accident that he relies on relational calculus as a criterion for the definition of relationally complete languages, in contrast to relational algebra.  

[snip]

> I may be misconstruing your remarks here. But I think the confusion between
> a "set with only one element" and the element itself is fairly pervasive.
>
I posted a previous reply with these sentiments.

[snip]

>
> Equally inforutnate, in my view, is Dawn's use of sets and elements as
> being interchangeable in her demonstration that functions are just as
> expressive as relations.

I agree. But if she get's the degree of rigor in her reasoning to address this issue, then she might have something.

I suspect that what she's done is mathematically
> illegitimate, although I'm no mathematician. I'm reasonably sure that it
> plays havoc with any kind of inference engine you might be interested in.

I'm no mathematician either.

Regards,

Dan Received on Fri Oct 08 2004 - 01:37:04 CEST