Re: atomic

On Nov 7, 1:44 am, paul c <toledobythe..._at_ooyah.ac> wrote:
> paul c wrote:
> > David BL wrote:
> > ...
> >> Is this what you mean (in Prolog):
>
> >> r(Name,Car,Colour) :- Name owns Car with Colour.
>
> >> r1'(Name,Car) :- r(Name,Car,Colour).
>
> >> or in natural language
>
> >> r1'(Name,Car) :- there exists Colour such that r(Name,Car,Colour).
> > ...
>
> > I think so.
>
> By the same token,
>
> r1'(Colour) :- there exists Name, Car such that r(Name, Car)

I assume you mean

    r1'(Colour) :- there exists Name, Car such that r(Name, Car, Colour)

> et al, ie., applies to all combinations of projections.
>
> I think of these as symmetrical REFERENCE'ing statements and they are
> implicit from the choice of attributes of a relation. For me, this
> means that the orthodox view is that the form of optional data must
> involve more than one relation. So if set-valued attributes are
> eligible in the orthodox framework, I think I'd want to be able to say
> similar sentences about them.

Are you saying you want to make statements about sets without those statements being reinterpreted as instead applying to the elements of the sets?

>
> The snippet above doesn't mention the types of Name, Car, Colour. I
> think there is a difference between saying the type of Car is set of
> elements and the type of Car is a subset of those elements and if I
> understand correctly, David is talking about a Car attribute which
> allows subsets as values (as I have been too).

Are you alluding to the distinction between

    r1(Name) :- Name is a valid name
    r2(Name) :- It is known there is a person called Name.

r1 is defining the domain type. Let's assume that the domain type is any finite string. Then r1 has an infinite extent and essentially information-less because it can be algorithmically compressed.

By contrast r2's extent is finite and far less compressible than r1, and so only r2's extent deserves to be explicitly recorded in a DB.

In the case of r1 the bad news is that we can't physically store the infinite extent. The good news is that we don't need to.

I'm not sure how the distinction between information-less-type and information-full-type relates to the discussion about how to interpret multi-valued attributes.

I've read your post a few times, but I'm afraid I don't understand the point you're making. Received on Wed Nov 07 2007 - 03:59:34 CET