Re: Relation or attribute and why

x wrote:
> "dawn" <dawnwolthuis_at_gmail.com> wrote in message
> news:1147826073.046909.174590_at_u72g2000cwu.googlegroups.com...
> > Let's say that we have a noun present in a conceptual data model, but
> > not identified as a strong entitiy (recognizing not all conceptual
> > modeling techniques use such distinctions). What are the conditions
> > under which this noun will/should translate into an attribute in a
> > logical data model? What are the conditions under which this noun
> > will/should translate into a relation in the logical data model?
>
> It will always translate to an attribute or a set of attributes.

Hey, thanks x ;-) but are you sure? It might be dropped entirely when going from the CDM to the LDM, although less likely if the LDM includes derived sets and attributes too.

We could take the noun "name" from our CDM and model it as a derived attribute or user-defined function, perhaps associating it with a derived set so we have something like

select function('fullName') as name from Person; perhaps placing this in a create view statement so we can then do select name from PersonView;

If we have a noun A that has-a B in our CDM, then we often model A as a relation and if B doesn't have-a C, then we might model it as an attribute in set A. In the case of "A:name has-a B:firstName" we often model "name" as a derived attribute, perhaps leaving it out of the LDM altogether so that each application can paste together the parts to form a name.

Is there a guideline for this that we could state in terms of functional dependencies or otherwise that would lead us to a noun such as "name" becoming a derived attribute rather than a set?

I'm trying to gain clarity in how one goes from a conceptual data model to a logical data model (when employing the RDM) and why. What is the technique (is it a normalization technique, is it based on FDs?) that might eliminate "name" in going from the conceptual to the logical data model? What rules or rules of thumb might make either A or B a relation or an attribute? For example, Marshall mentioned the technique once known as 1NF in the case where B is multivalued. In that case, B becomes a set instead of an attribute. Are there other instances when B which has no C would become a set instead of an attribute? What rule are we employing when we drop B from the LDM or make it derived (A:date has-a B:month would be an example)?

Has-a is just one type of relationship in a conceptual data model, but maybe it is worth zeroing in on. Thanks. --dawn Received on Fri May 19 2006 - 00:10:42 CEST