Re: One-To-One Relationships

On 30 nov, 19:33, JOG <j..._at_cs.nott.ac.uk> wrote:
> On Nov 30, 6:03 pm, Tegiri Nenashi <TegiriNena..._at_gmail.com> wrote:
>
> > On Nov 30, 8:19 am, Jan Hidders <hidd..._at_gmail.com> wrote:
> > The distinction between entities /
>
> > > relationships, domain objects / predicates is pretty well-established
> > > in linguistics, philosophy and logic.
>
> > That certainly means you can define them formally in database terms,
> > right?
> > Here is one such attempt:http://arxiv.org/PS_cache/arxiv/pdf/0710/0710.2083v1.pdf
> > It defines an entity as a relation (aka table:-) with a single
> > noncomposite key, and relationship as a table with composite key. Does
> > this definition pretty much exhausts the entity-relationship theory?
>
> I like the insight that both 'entities' and 'relationships' are
> subtypes of a parent concept, that is simply a set of attributes and
> values. I'd like to see a formalization of that which doesn't rely on
> relational theory and the concept of keys however, even though I
> imagine there would be a direct correspondence.
OK I will play devil's advocate and try to believe that entities could be an alternative expression of entities...

Quite frankly, I understand the intent but not the need to differentiate relations and entities...

I believe somehow that because Relations rely on set theory and relational algebra, it simply has a longer mathematical history and de facto more abstract tools concepts to work with than Entities that seem recent. I do believe that the keys was simply Codd's way to express *identifiablity* in a way that his IBM audience could be receptive to the rest of the model...I do not perceive keys as a sufficient reason to require differentiation between the two models... Received on Sat Dec 01 2007 - 20:27:13 CET