Re: One-To-One Relationships
Date: Fri, 30 Nov 2007 13:36:36 -0800 (PST)
On 30 nov, 17:52, JOG <j..._at_cs.nott.ac.uk> wrote:
> On Nov 30, 4:19 pm, Jan Hidders <hidd..._at_gmail.com> wrote:
> > On 29 nov, 23:16, JOG <j..._at_cs.nott.ac.uk> wrote:
> > > On Nov 28, 8:43 pm, rp..._at_pcwin518.campus.tue.nl (rpost) wrote:
> > > > paul c wrote:
> > > > [...]
> > > > >Regarding ER, here are some quotes from Codd's book (available for free
> > > > >at acm.org). The sarcasm of the second one made me laugh.
> > > > The criticisms you quote may be amusing, and they have merit, but
> > > > they ultimately miss the point. The distinction between entities and
> > > > relationships: entities have identity (they can be referred to; attributes
> > > > can have entity-valued domains), while relationships do not (they are
> > > > completely identified by their, possibly entity-valued, attributes).
> > > I realise that others have attempted to point out your mistakes, but I
> > > just wanted to echo their sentiment.
> > How about providing some real arguments, in stead of just sentiments?
> > Haven't seen much of those here and I think Reinier deserves them.
> Feel free to offer some of your own for him instead of simply
> > > There is absolutely no difference
> > > between an entity and a relationship. E/R modelling has /itself/
> > > conceded this, translating relationships into "associative entities".
> > Ah, yes, and as we all know, if two concepts have overlap then they
> > are actually the same. *sigh* The distinction between entities /
> > relationships
> Great, if this distinction is so clear feel free to elucidate the
> final word on the matter.
You want me to explain the difference between the notion of "thing" and "fact"? Every fact is also a thing, but not everything is a fact. More precisely, a fact is an instantiation of a predicate where all roles have been filled in by things and which is true. That kind of thing?
> > domain objects / predicates is pretty well-established
> > in linguistics, philosophy and logic. First-order logic, you may have
> > heard of it, separates them even strictly.
> Puh-lease Jan, this sort of thing is beneath you. There is no need to
> be so condescending.
You do realize that your response to Reinier also had that ingredient, don't you? Although others were far more worse. Reinier has given these matters more thought than you may realize.
> This is a usenet forum not ACM transactions, so
> there is always going to be loose debate. And anyhow, the question is
> not of the use of entities in FOL, but of a preffered role for
> entities in a logical data model.
I don't think so. The ER model places entities and relationships on equal footing. On the contrary, I would claim that the RM is unbalanced and focuses too much on the relationships.
> > Are you now going to claim that it is no good for reasoning?
> Who ever claimed that?
I didn't say you did, I asked whether you were going to. It seems to follow from what you claim. If the distinction is not useful and in fact bad for data modeling then this should somehow become apparent when you start working with such data models. After all, if it doesn't then it becomes hard to maintain that there is actually a problem. And reasoning about data would seem a typical intensive use where you would start noticing that your data model is not what it should be.
> I am sure you don't believe that we should just accede that Jan has
> spoken and everything is already decided.
No, absolutely not. If anything I'm here to stimulate debate and fight they the idea that Codd and Date have spoken and everything is already decided. So accepting my opinion on authority is the last thing I would want. But I do reserve the right to sometimes be a bit contrary and try to tease people into a discussion. With some that works, others may get all defensive. :-)
> Ought we just forgo all the /
> ongoing/ issues in identity research, because FOL has its own
Hm? FOL doesn't say much about identity. It just assumes you know how to tell which objects in the domain are the same or not.
> Maybe you should have spoken to Geach before he wasted all
> his time on relative identity? After all Liebniz had already defined
> identity centuries ago right?
Not sure how that is related, but since you brought it up let me say that the ER model is a far better setting for discussions about such identity research than the relational model. The inheritance of identity. The notion of relative identity. The changing of identity with different roles or with time. All can be elegantly formalized and discussed in the context of ER models.
I was very annoyed by yours. Call it even?
- Jan Hidders