Re: A Normalization Question

Neo wrote:
>>>With respect to dbs, normalization is the process of eliminating or
>>>replacing duplicate things with a reference to the original thing
>>>being represented.
>>
>>Not things, *facts*! Normalization is about preventing that a certain
>>fact is being represented in more than one way.

>  
> Give me an accurate definition of fact and I will show 'brown' can
> also be a fact.

You have to show that in the tuple ("brown", "brown", "brown") the string "brown" each time represents the same fact.

>>That's why having several copies of the string "Brown" is not redundancy.
>
> "Having several copies" (ie 'brown', 'brown', 'brown') is redundant.

I'm afraid that repeating the claim does not make it more true.

> It requires a limited data model to ignore this fact.

It's true in any data model we know. It's true in the flat relational model, it's true in the nested relational model, it's true in the format data model, it's true in object-oriented data models, it's true in object-relational data models, it's true in ER data models, in object-role data models, in the functional data model, in the XML data model, in the semistructured data model, and it's also true in the MV data model (to the extent that this is actually a well-defined notion).

• Jan Hidders