Re: Objects and Relations

On Feb 21, 6:05 pm, "Cimode" <cim..._at_hotmail.com> wrote:
> On Feb 21, 3:06 am, "David BL" <davi..._at_iinet.net.au> wrote:
>
>
>
>
>
> > On Feb 21, 12:03 am, "Walt" <wami..._at_verizon.net> wrote:
>
> > > In casual conversation, ER modelers will tend to use the simple word
> > > "entity" in place of either "entity set" or "entity instance", leaving the
> > > listener to disambiguate by means of the context in whuch the word appears.
>
> > > By analogy, in casual conversation, object oriented programmers will use the
> > > simple word "object" in place of either "object class" or "object instance",
> > > similarly leaving disambiguation up to the listener.
>
> > > The problem comes when the listener is not familiar with the underlying mode
> > > of thinking. In that case, the listener will sometimes disambiguate
> > > incorrectly. That is why introductory tutorials on object oriented
> > > programming tend to spell out "object class" or "object instance", at least
> > > until the reader can be presumed to have gotten accustomed to the mode of
> > > thinking. I'm sure you will have noticed this, if you've gone back and read
> > > some introductory material after gaining proficiency.
>
> > > Similarly, introductory material on ER modeling should spell out when we
> > > are talking about the "set of all vehicles" and when we are talking about "a
> > > particular vehicle". Some such material does this.
>
> > > Discourse in this newgroup tends to be a little more formal than casual
> > > conversation, but far less formal than introductory tutorials.
>
> > > Hope this helps.
>
> > I prefer to only use "object" to mean instance. Why would one say
> > object when one means class? ie given that we have different words
> > for these different concepts, let's use them!
>
> > A word like "book" can mean an instance as well as a type, depending
> > on the context. It could be argued that it really means a type, and
> > its use for naming an instance is a curious feature of how we
> > communicate because we don't generally want to go to the trouble to
> > give things around us explicit names.
>
> > I agree that "entity" can have both meanings as well. However, I was
> > taking the statement "entities are illusionary" to mean that all
> > instances of entities are illusionary, rather than only entity types.
>
> > It is interesting that in UML we have class diagrams and object
> > diagrams. An ERD is analogous to a class diagram. That could
> > explain Jim's tendency to regard "entity" as being a type.
>
> David. RM uses a more elementary, antecedent and efficient
> terminology into naming entities, objects, classes, or other OO
> related concepts. Once you understand that in an RM perspective a
> class is the representation of some relation and that in RM a relation
> systematically defines a data type. To understand thoroughly such
> difference, you need to understand better RM concepts.
>
> Up to you either to educate yourself

It seems clear that relations give you the effect of a very rich taxonomy, if that's what you mean. I agree that the best way by far to classify things is through RM.

I guess you could say that RA can be regarded as a means to perform (extensional) type analysis. For example you can define a type as the set of married humans with green cars, and with minimal effort RA lets you calculate the extension of that type. Is that terminology acceptable?

I intend to have a read of the material you suggested. Received on Wed Feb 21 2007 - 13:27:26 CET