Re: Question on 5NF

davide.mauri_at_gmail.com wrote:
> > > I found myself in a situation where I have the following rule (speaking
> > > using wikipedia sample's terms): the Condition that the Insurance
> > > provider can cover are *only* those conditions covered by the
> > > Psychiatrist which are affiliated with that Insurance Provider.
> >
> > That's actually not a static constraint, and that's what normalization
> > deals with, so your question is not really a normalization question. To
> > understand why it's not a static constraint try to formulate it as a
> > logical constraint on the relation. You get something like:
> >
> > "for every tuple (P,I,C) in the relation there is a tuple (P',I',C') in
> > the relation such that I=I' and C=C' "
> >
> > Which is a tautology. But if you formulate it as a dynamic constraint
> >
> > "you can add a tuple (P,I,C) in the relation if there is already a
> > tuple (P',I',C') in the relation such that I=I' and C=C' "
> >
> > then it makes sense. :-)
>
> Ok :-)
>
> > Depends on what FDs, MVDs and JDs you think hold. In particular you
> > haven't told us yet if the JD that might lead to a split of PIC
> > actually holds or not. Presuming that it does and there are no further
> > non-trivial FDs and MVDs then you are not in 5NF.
>
> But if there aren't non-trivial FDs and MVDs shouldn't I be in 5NF?

Then (meaning: if there are non-trivial JDs, but no non-trivial FDs and MVDs) you are in 4NF but not in 5NF.

• Jan Hidders