Re: more closed-world chatter

On 5 mai, 17:28, paul c <toledobythe..._at_oohay.ac> wrote:
> Cimode wrote:
> > ...
> > It seems to me that the spirit of your question leads to the computing
> > model that actually allows the representation of sets. What I mean is
> > that the arbitration of whether exceptions should or shoud not occur
> > is tightly bound to the chronological availability of the information
> > to the db engine of set disjointness/non disjointness. As a
> > consequence, I believe the arbitration is more related to the
> > computing model that would allow the implementation of a TRDBMS,
> > rather than the abstract aspect of RM.
> > ...
>
> Philosophically, I think I buy that. I don't know how useful it is but
> I like such extreme questions if only because they help me find a useful
> logical "centre".
Not that I do not appreciate a little philosophy but I am afraid these complex issues must me adressed if we want to imagine a decent computing model without falling into simplyism. I am getting to the point where I believe that RM really needs a stronger computing model to back it up. I sometime wander if these issues were dealt in transrelational works.

> I like D&D's (or maybe it should "D|D's") premise that goes something
> like "domains/types are the things we talk about and relations are the
> things we say about the things we are talking about" even though some US
> president said it first ("let me say this about that"). With finite
> machines, this may seem like wild and woolly sci-fi, but what if fixed
> domains aren't the starting point?
I am not hundred percent sure *fixed* is the right term for that. (always hard to put a term on something unknown)

I would not have rather called them *relative domain representation*. They are relative because the adressing scheme of the intervals defining two sets of values may be infered one from another as much much as their characteristics.

For instance suppose we have the 2 following sets of Integers and decimals:

R1= {1, 2, 3}
R2={4.0}

Core question is: what data layer representation may allow the system to infer that both are disjoint *declaratively*. So far I came up with the following assumptions:

suppose a storage of the values, that:
--> uses the highest required precision for representing a numerical value independently of its belonging to a specific domain. --> represents *all* numerical values under the above format

We could therefore imagine a basic data layer of numerical defined as

D:{1.0, 2.0, 3.0, 4.0} (which is equivalent logically as a UNION between the two sets BUT represented wih the highest degree of precision - in this case decimals)

Logically the sets are represented by the intervals

R1: 1.0 -> 3.0
R2: 4.0-> 4.0
D:  1.0-> 4.0

Therefore to answer the core question(I apologize in advance for the notation used), one could imagine the following information required:

R1: D - 4.0
R2: D - R1 = D-(1->3)

--> therefore R1 /\ R2 can be expressed as (D - 4.0) /\ (D - (1->3))

>From that equation and a scheme to allow to infer (for example a
mathematical adressing scheme), it is not unreasonnable to believe that disjointness could be expressed truly in a declarative manner. Some overhead is of course required but I do believe it to be feasible. So far I already have some interesting results on the above example using vectorial mathematical tools to create the scheme but I am not happy with the mathematical tool . But I guaranteee this is far from being scifi.(just wish I had more time).

> p
Received on Sat May 05 2007 - 18:39:35 CEST