Re: Relation or attribute and why
Date: Wed, 24 May 2006 12:41:22 GMT
Message-ID: <SvYcg.1606$GN4.1403_at_trndny07>
"dawn" <dawnwolthuis_at_gmail.com> wrote in message
news:1148432908.908347.174660_at_j55g2000cwa.googlegroups.com...
> Gene Wirchenko wrote:
> > On 23 May 2006 15:44:15 -0700, "dawn" <dawnwolthuis_at_gmail.com> wrote:
> >
> > >Gene Wirchenko wrote:
> > >> On 23 May 2006 13:44:29 -0700, "dawn" <dawnwolthuis_at_gmail.com> wrote:
> > >>
> > >> [snip]
> > >>
> > >> >In my example, the conceptual data model includes: name, firstName,
> > >> >lastName with relationships such that name has-a firstName and name
> > >> >has-a lastName.
> > >>
> > >> That is part of the logical model. The conceptual model is the
> > >> business statement.
> > >
> > >The identification of name, lastName, firstName and the relationships
> > >among these terms is not part of the business scope and definition, the
> > >conceptual data model? What would the CDM include related to these
> > >terms? I put everything that is conceptual and not directed to a
> > >particular implementation model (such as the RM) in the CDM. Where do
> > >you draw the (possibly fuzzy) line? --dawn
> >
> > Possibly fuzzy, but nonetheless.
> >
> > Conceptual might fit on a napkin: "We want an E-commerce Web site
> > for selling our products. It has to be able to handle North America,
> > but we plan to go international, so have the capability to add other
> > languages, etc. easily without rewriting large chunks. We want
> > someone able to complete an order quickly. Do not forget good
> > security. OtherCorp recently had a bad situation, and they are taking
> > a kicking. Now is our chance, if we do it right."
>
> Those are high level requirements, but not a conceptual data model.
> Conceptual, yes; CDM, no. You need at least an ORM, ERD, or a cleaner
> list of propositions related to these requirements, I would think.
>
> >
> > Logical gets into the details, but not the implementation.
>
> A typical use of the term would allow for a conceptual data model to go
> from high level to detailed (in the end). [Although if a decision is
> made earlier that the requirements have been captured sufficiently to
> do flesh it out using common sense in a transition to an LDM, I can
> live with some agility.]
>
> I once considered the LDM to be implementation-independent, but after
> reading other definitions the logical data model appears to be data
> model dependent to most who use the term. In that case, you would not
> have the same logical data model for an RM implementation as an MV
> implementation, for example.
>
Agreed.
The LDM is data model dependent. It's also a design model, not an analysis model. Elements of the LDM pertain to the solution domain, not the problem domain. People who insist that there is only one viable data model tend to nerge the LDM with either the PDM or the CDM. Most of the RM "catholics" in c.d.t. merge the LDM with the CDM, because they think that selecting a data model is a choice with only one answer.
A product like "Data Architect", back in 1999, merged the LDM with the PDM. The CDM was like ER, with more features, but still an analysis model. To be fair, DA only offered PDMs for SQL databases, so their choice made sense, in their context. Received on Wed May 24 2006 - 14:41:22 CEST