Re: Objects and Relations
Date: 6 Feb 2007 06:04:26 -0800
Message-ID: <1170770666.806752.41280_at_v45g2000cwv.googlegroups.com>
On Feb 6, 12:54 pm, "David BL" <davi..._at_iinet.net.au> wrote:
> On Feb 6, 7:26 pm, "JOG" <j..._at_cs.nott.ac.uk> wrote:
>
> > On Feb 6, 5:26 am, "David BL" <davi..._at_iinet.net.au> wrote:
>
> > > On Feb 6, 12:29 am, "JOG" <j..._at_cs.nott.ac.uk> wrote:
>
> > > > On Feb 5, 1:25 pm, "David BL" <davi..._at_iinet.net.au> wrote:
>
> > Discussion of OID's snipped because it appears we are in fact largely
> > agreeing that all labels are attributes - and consequently that
> > identity should not depend on a pointer or memory address, which is
> > obviously not part of the real world data.
>
> Good!
>
> > > > > I agree that E/R modelling should be "consigned to being an
> > > > > organizational tool", but again for different reasons. Unlike you I
> > > > > see no problem with saying entities have identity.
>
> > > > I have never said that. What I have argued is that they are identified
> > > > by their attributes and not some imaginary hidden barcode.
>
> > > I agree with that but don't see how that explains why E/R diagrams can
> > > be problematic.
>
> > This was an argument against OID's not E/R diagrams.
>
> I hadn't realised you had a different argument against E/R diagrams.
>
>
>
> > The argument against E/R diagrams is trying to define what an entity /
> > is/. For example, if bob is married to shiela, is married_to a
> > relationship or an entity? And if it is an associative entity, why are
> > all relationships not then just entities? Or to put it a different way
> > (and the way much of modern philosophy views knowledge) should all
> > entities not just be viewed as n-ary relationships themselves? Once
> > you arrive at this premise, by whatever route, whether it be data
> > analysis, reductionism, etc, you are left with something that starts
> > to look incredibly like NIAM/ORM, which in turn has a natural
> > translation to relational modelling.
>
> > > The claim that entities are illusionary or abstract
> > > doesn't appeal to me.
>
> > Fair enough. You are incorrect, but that would be a philosophical
> > debate, and largely tangential to the practical issues at hand.
> > However it is worth noting that your standpoint is that of
> > essentialism, which carries little weight in modern epistimology.
>
> My tendency these days is to believe that the universe (actually
> multiverse) only exists in a Platonic sense and physical reality is a
> kind of trick played on us as self-aware substructures of the
> multiverse. That kind of kills the whole question of what is real
> versus what is abstract. It seems more convenient to define
> everything to be real (not abstract), whether they be humans or
> numbers.
>
> > So,
> > perhaps instead you could clarify what you mean by an 'entity' and
> > define exactly what it is, as a starter to convincing me that it is
> > not in fact an abstract notion. (and please try and avoid using the
> > word 'thing' in any such a definition).
>
> Perhaps a more pertinent question is to ask you to define the
> difference between real and abstract.
>
> It's impossible to define entity in terms of simpler notions but I can
> clarify my viewpoint.
>
> If I point at a bunch of atoms and say "That is a human named Fred"
> then I consider Fred to be conceptual, mathematically imprecise but
> nevertheless the concept itself is real and exists in a Platonic sense
> (to the extent that I can meaningfully make statements about that
> bunch of atoms as I conceive of it as being a human named Fred).
Well first pointing isn't allowed. One cannot continually follow Fred
round pointing at him all day, we need to identify to him in some
other way, but there I think we are agreed.
However consider that some people's view of Fred is that Fred is Fred
is Fred, throughout his whole lifetime. Other people/applications may
need to view Fred only as he is /now/ - after all he will be composed
of a completely different set of atoms in a few years anyway. Then
there must be parts of him that make up our view of him which are
purely abstract, his bravery, his humour, his IQ, his body of
publications, etc., how should they perdure? Already the notion of
where he starts and finishes as an 'entity' is seriously blurring.
The old washington's axe/theseus ship paradox is perhaps a better
example of the concept of a well defined entity being insufficient. I
am sure you know it - "In a museum somewhere there is the axe used by
washington. Ok, the handles been changed a few times. Oh and so has
the headstock, but hey... its still the same axe". Same 'axe entity',
completely different physical components. Something seems intuitively
wrong with this notion of an axe 'entity/object'.
But this problem is not really a paradox at all if you just discard the illusion that an entity exists anywhere outside a single individuals head. That's again why I believe one should always consider communicated propositions (which we have a shared common understanding of). Then you it becomes clear that the axe is the same axe if propositions discussing it use its name as the primary key. It is a different axe if the headstock/handle attributes are the compound key.
The purpose of this is not anal philosophising. Imo philosophy is absolutely pointless if it cannot be applied practically and usefully. And in this Codd's approach is genius, because this entityless philosophy underpins his original RM, and we've seen it actually / applies/ to solving real world data handling problems. Enough that it even seems possible to help me store information about employees _and_ solve the Theseus Ship paradox in terms of RM keys. And unlike OO, simply from building on a firm mathematical foundation of predicate logic. Who'd have thought.
> As
> a Platonist I don't like to say that the entity labelled Fred is
> abstract. Instead I say it is real because I define all (reasonable)
> mathematical notions to be real, whether they map back to atoms around
> me or not. I say something physically exists if there is a mapping
> back to atoms.
Fair enough, but I don't really care if your a platonist, a reductionist, a mereological nihilist or a senior gynaecologist (and trust me compared to practicitioners like Gene and Marshall, I'm positively wooly). A philosophy is only worth its salt here if it has practical positive impact.
> Entities are always real (ie exist) but they may or
> may not physically exist. I note that the definition of physical
> existence is dependent on the observer as a self-aware substructure in
> a multiverse.