Re: ADR's Normalization question
Date: Mon, 16 Nov 2009 23:31:57 -0500
Message-ID: <UPidnZnQfrejup_WnZ2dnUVZ_hqdnZ2d_at_giganews.com>
"Sampo Syreeni" <decoy_at_iki.fi> wrote in message
news:c3f74b55-aa8e-4637-b600-12268417cbdd_at_w19g2000yqk.googlegroups.com...
>> The claim is due to Date's "otherwise quite independent" criterion. [...]
>
> Okay, this is new to me, and given your further exposition, would
> appear highly nonstandard to me. Can you point me towards a freely
> available paper in which Date nails himself to the cross with this
> interpretation?
I suggest you read Ronald Fagin's paper "Multivalued Dependencies and a New Normal Form for Relational Databases." Date's "otherwise quite independent" criterion exemplifies the kind of relationship defined in the paper. For example, an employee's salary is quite independent from (Fagin uses "orthogonal" to) an employee's set of children. In the same way the set of cities associated with a particular status is orthogonal to the set of classes that are associated with it. The appearance of being nonstandard is probably due to the fact that multivalued dependencies are not ordinarily considered in the normalization procedure unless the database scheme is already in BCNF. Received on Tue Nov 17 2009 - 05:31:57 CET