Re: Alternative method to normilize tabels ...
Date: 23 Oct 2002 00:19:50 +0200
Message-ID: <3db5cf06$1_at_news.uia.ac.be>
In article <coft9.173540$8o4.28014_at_afrodite.telenet-ops.be>,
joske_at_home <vandenbergeDOTj_at_village.uunet.be> wrote:
>
>On the website ... http://www.sum-it.nl/cursus/dbdesign/ you find a
>practical design method to normalize your data (up to 3NF).
>
>As a newbie, this method seems a little different then the usual
>normalization-steps. Is the method desribed there a valid method (up to the
>3NF)? They don't state you need to remove your repetitive data ...
It's roughly the standard method, but it simplifies a bit too much here and there. For example, it assumes that there is always only one key, which is of course nonsense. In the description of 3NF they forget to mention that that you only have to consider transitive dependencies that end in non-key attributes. If you take their definition you are actually normalizing to the next normal form, i.e., Boyce-Codd normal form. The abbreviations for the normal forms is a bit unconventional and UNF is usually defined as some kind nested data structure, but the term is not really fixed yet.
- Jan Hidders