Re: One-To-One Relationships

On Nov 30, 3:41 am, David BL <davi..._at_iinet.net.au> wrote:
> On Nov 30, 10:53 am, JOG <j..._at_cs.nott.ac.uk> wrote:
>
>
>
> > On Nov 30, 12:53 am, David BL <davi..._at_iinet.net.au> wrote:
>
> > > On Nov 30, 7:34 am, JOG <j..._at_cs.nott.ac.uk> wrote:
>
> > > > On Nov 28, 10:37 pm, paul c <toledobythe..._at_ooyah.ac> wrote:
>
> > > > > Bob Badour wrote:
> > > > > > rpost wrote:
> > > > > ...
> > > > > >> This is the exact problem Chen identified. In the relational model
> > > > > >> it is impossible to have entity-valued attributes, which, in practice,
> > > > > >> we have a huge amount of.
>
> > > > > > Entities are figments of our imaginations.
> > > > > > ...
>
> > > > > That's much better than my reply, looks like the essential point to me.
>
> > > > Entities are concepts that we impose on that which is there (mostly
> > > > shapes made out of atoms, but sometimes abstractions too).
>
> > > > Entities are 'real' in that they are patterns that we think up and
> > > > apply, and 'not real' in that if we were all dead, well there wouldn't
> > > > be any 'entities' would there. Just atoms again.
>
> > > Is this merely an argument against mathematical realism or something
> > > more specific?
>
> > I haven't argued anything, I have just defined entities. All seems
> > like common sense to me.
>
> Suppose I define a bijective mapping called "f" between the integers
> {1,2,3} and particular groups of atoms that I perceive right now on my
> desk as follows
>
> 1 <--> {set of atoms that looks like a pen}
> 2 <--> {set of atoms that looks like a mug}
> 3 <--> {set of atoms that looks like a keyboard}
>
> Does this mapping only exist in my mind, and would cease to exist if I
> died, or does the mapping have an independent existence?

If everyone died who agreed to use that mapping, well of course, then there is no pen, mug or keyboard constucts. Just a bunch of atoms with no-one left to impose descriptions on them.

> This is
> merely a philosophical question about mathematical realism. Or are
> you saying that the problem is in a precise definition of the groups
> of atoms?

The problem is only if one views the bunch of atoms as one single predefined  thing. We're a lot more flexible than that. When one uses entities in conversation one is incredibly adaptable with them and their scope. When one encodes them in a data base its the opposite, cementing a single, fixed and brittle viewpoint of the world. And that will end up at best with query bias as requirements change, and at worst a broken database.

>
> > > > Because we conjure them up 'entities' aren't neatly defined, can
> > > > overlap, change, and will differ from person to person. In fact its a
> > > > testament to the amazing flexibilty of our noggins that that we
> > > > manage to communicate at all.
>
> > > Do you consider all mathematics to be "conjured up" and therefore not
> > > neatly defined as well?
>
> > Non-sequitur. I have made no comments on mathematics nor on anything
> > being ill-defined.
>
> You made the statement
>
> "Because we conjure them up 'entities' aren't neatly defined"
>
> The use of the word "because" seems to suggest
>
> "conjured up" implies "not neatly defined".

Agh, you're right. Apologies if I came across overly defensive - it is very frustrating trying to describe what I see with clarity in my noggin into cogent sentences. I was trying to get at the point that, because entities don't exist as some a priori force of nature, that they often aren't nice and neatly partitioned from other possible entitiwa. Lots of nebulousness, overlap and change.

>
> > What I am suggesting is that given the contextual nature of how we
> > define any given entity, any model centered upon them is going to have
> > a very inflexible, brittle view of the world.
>
> How do you distinguish between a model centered upon entities versus a
> model that is not?

Off the top of my head I'd say a model centred on entities, uses them as first order citizens through which to encode data (e.g. objectorientation).  Conversely, other models only reference entities indirectly via statements about them (e.g. RM). The former has a single viewpoint of the partitioning of the universe of discourse, the latter does not, allowing one to extract any particular conceptual model from a neutral underlying logical model.

>
> > There are a load of atoms. Someone else looks at them and says hey its
> > a dead bit of wood, its a 'stump' entity. Someone else considers a
> > bigger picture and says its a snapshot of a 'tree' entity. Me, being
> > lazy, say hey I could sit on that, its a 'seat' entity, etc., ad
> > infinitum.
>
> > Whos right? Noone. Everyone. Who cares. It just depends on the
> > context, and with shared data, thats gonna vary. So lets encode the
> > underlying bloody information, and not whatever brittle view of the
> > world that first pops into our heads, ending up with a brittle,
> > creaking database.
>
> This appears to be a different argument to your previous one. I think
> you are now stating the underlying problem with entities is non-
> uniqueness, which I find more agreeable than your previous meta-
> physical statements concerning existence.

It appears to be a natural consequence of following the line of logic to me. Perhaps you can ellucidate though.

>
> Or do you consider the non-uniqueness problem to only be a symptom of
> your meta-physical standpoint?
>
> > > > And as such, anyone who tells you that you can build a permanent, all
> > > > encompassing model out of such utterly woolly things is not to be
> > > > trusted as far as you can throw them.
>
> > > > P.S I have also had similar experinces at Ikea :) I don't follow the
> > > > arrows anymore. They lie.
>
> > > If there is a valid argument against "entities" I would hope it can be
> > > stated more carefully than your or Bob's attempts.
>
> > Again I haven't argued against entities. However, I am of course
> > interested in what you deem invalid in my comments.
>
> Yes, you argued against models that are centered around entities.
> However I don't actually know what that means.
>
> Please don't take my post in the wrong way. I'm hoping the outcome
> will be a more accurate way of saying what's wrong with ER
> diagrams! :)

Me too ;) I don't really see anything wrong with E/R diagrams per se. The problem is the lack of understanding that they represent only a / single view/ of the universe of discourse. If one isn't aware of that, well you can end up getting stuck down a one way road when new requirements arrive. Received on Fri Nov 30 2007 - 15:53:54 CET