Re: Functional Dependencies

From: Mikito Harakiri <Mikito_member_at_newsranger.com>
Date: Fri, 20 Apr 2001 01:46:17 GMT
Message-ID: <JxMD6.3765$D4.377075_at_www.newsranger.com>


In article <01c0c912$55b51920$2d00dd63_at_ts457030>, Shayne G. Wright says...
>
>Hi Selina
>
>In a FD (X -> Y), Y can only be a single value, which is functionally
>dependent on X
>
>ex.
>
>X Y
>-----
>X1 Y1
>X2 Y2
>X1 Y1
>
>In a MVD (X ->> Y), Y can be a fixed set of values
>
>X Y
>-------
>X1 {Y1, Y2}
>X2 {Y3, Y4}
>X1 {Y1, Y2}
>
>A FD is essentially a stronger form of MVD (with a single value
>requirement, as opposed to a set). So, all FDs can be considered MVDs, if
>the entire set is considered to be a single value, but a MVD is not
>necessarily a FD. That is, in all cases X -> Y |= X ->> Y, but X ->> Y \> X
>-> Y, because:
>
>(X ->> Y)
>
>X Y
>------
>X1 {Y1, Y2}, is equivalent to
>
>X Y
>-------
>X1 Y1
>X1 Y2, so obviously X \> Y
>
Those are called trivial multivalued dependencies. The confusing part is that trivial multivalued dependencies are useless (since it is always the case that X->>Y:-). With one more column Z added the question is "which set?". Multivalued dependencies are defined so that the set in question is the same no matter what value z in Z is chosen. Received on Fri Apr 20 2001 - 03:46:17 CEST

Original text of this message