Re: BCNF and relational algebra

"Perception" <perceptionist2000_at_yahoo.com> wrote in message news:c1b1pk$inq$1_at_titan.btinternet.com...
> Hi all,
>
> I have the following relations (expressed as functional dependencies)
which
> I consider to be in BCNF (please correct me if I'm wrong):
>
> Patient Identifier -> Patient Name, Ward
> Date
> Doctor -> Clinic
>
> Now the problem here is that I am unable to deduce how to make any real
> query on these set of data because there are no common attributes between
> the relations. Is it even possible for relations under BCNF to have no
> common attributes? Note that the original relation was identical expect it
> had the additional functional dependency: Patient Identifier, Date,
> Doctor -> Patient Name, Ward, Clinic (i.e. the attributes on the LHS were
> the primary key). I decomposed the relation as I have shown above given
that
> I knew each Patient Identifier was unique, a patient would never change
> wards but could attend several clinics, on each occasion a patient would
see
> one doctor, doctor names were unique and that doctors were only associated
> with a single clinic. Now I believe that I have made a reasonable
> interpretation of this information both in my choice of primary key in the
> original relation (i.e. Patient Identifier, Date and Doctor) and in my
> decompositions into BCNF (every determinant is a candidate key).
>
> Thus, if I wanted to list, say, all patient identifiers and names for all
> patients attending a certain clinic and having not been seen by a certain
> doctor how on earth would this be possible in relational algebra given
that
> I cannot join over any common attribute?? So either I have made a mistake
in
> my decomposition or there are some techniques that I am not aware of (that
> can link the FDs in some way).
>
> In either case I would appreciate any help or clarification!
>
> Thanks in anticipation.

What do you suppose the unary relation of dates means? Received on Sun Feb 22 2004 - 21:15:01 CET