Re: What predicates the following relation represents

"mAsterdam" <mAsterdam_at_vrijdag.org> wrote in message news:406c66a0$0$572$e4fe514c_at_news.xs4all.nl...
> You originally asked:
> > Is SALES a legal relation? Is database really a repository of facts?
>
> If you, by "legal" mean "some rows I can put into some table",
> then the answer to your first question is obviously: sure.
> However, in order to become statements, they need meaning.
> The proper (canonical, or 'legal' if you want) way to do
> that is by formulating a predicate.
> Now the database can become a repository of facts.
>
> Reiterating:
> Q: What is missing?
> A: The predicate.

Every table has a signature, right? In our case I implied

table SALES (

    parts String,
    sold Integer
)

and the table content

insert into SALES values (nuts, 10)

This can be translated into predicate calculus. We have

Constants: 0,1,2,3,4,5,...
String Literals (also constants): 'bolts', 'nuts',... Predicate constants such as <, >=, etc
User-defined predicate constants, such as SALES(x, y) where x in PARTS and y in SOLD.
And, finally, we have axioms like this
SALES('nuts', 10)

This naive translation could be criticised, of course, but I fail to see a missing "predicate".

> > How do you expand the example in order to resolve the ambiguity. I
prefer
> > formal solution, not just vague call for "more semantics".
>
> Your qualification is wrong, and put in an annoying way - but I'll get
> over the latter.
>
> I asked you to give the specific facts your example is
> supposed to state. Granted that is a call for semantics.
> But not just any.
> It is a specific call for only the semantics necessary to answer
> your question.

I still don't understand what do you mean by "semantics". More predicates? Axioms? The only "reasonable" semantics formalization I'm aware of is Model Theory (see below). I suggest, however, to stop here as our exchange becomes more and more cranky. Admittedly there is my share in it.

> I am so sorry to bore you all if everybody else here
> knows which model theory you are talking about.
> I really don't. So, again: which model theory?

Do you imply there is more than one?

http://en.wikipedia.org/wiki/Model_theory http://mathworld.wolfram.com/ModelTheory.html Received on Thu Apr 01 2004 - 21:52:55 CEST