Re: Why spurious tuples with fifth normal form?

Jan Hidders wrote:
>
> To give a small example: let for the relation R(A,B,C,D,E) the candidate
> keys be AB and C, and assume that the JD { ABC, ABE, CE } holds. Is it
> implied?:

Oops. The JD should be (see below) JD {ABC, AB*D*, CE}

> step 1. H = { ABC, ABD, CE }
> step 2. (#1) H = { ABCD, CE } (because ABC and ABC share the c.k. AB)
> step 2. (#2) H = { ABCDE } (because ABCDE and CE share the c.k. C)
> step 3. no more two distinct components with common cand. key
> step 4. The JD is implied by the cand. keys because ABCDE is in H
>
> So even though this join dependency holds, you do not have to split.

• Jan hidders