Re: A Normalization Question

>From: "D Guntermann" guntermann_at_hotmail.com
>Date: 2004-07-09 2:16 PM Eastern Daylight Time
>Message-id: <I0LK4C.J6_at_news.boeing.com>
>
>
>"VHarris001" <vharris001_at_aol.com> wrote in message
>news:20040709081455.29796.00001156_at_mb-m18.aol.com...
>> Normalizing the data this finely might be necessary in some cases.
>
>What "data" are you normalizing? As a logical application data model, I
>could see this serve as the definition for alphabets and languages which
>would correspond to functional dependencies, but these are modeled quite
>nicely already in other forms and methods. Do you other applications where
>this would be useful in mind? An example, perhaps?
>
>>
>> V Harris
>>
>
>- Dan

One data model I always had trouble with was 'family' relationships and related detail.

Suppose we want to keep all contacts in a database. And for some of them we also want to track their relatives, friends, dates for special events, and other personal information.

As you know, there are many types and levels of relationships, e.g., biological, legal, religious, social, etc.

And while some of the relations are factual, some are only speculative. For example, some parentage facts can only be represented as a probability (e.g., it was either Mr. Brown or some guy named Tom, or . . .).

The table design seemed too complicated to normalize. That is when I began to think that rather than make a new table for each new, unique relationship I encountered, it might be easier to put all relationships in one table and describe them as I encountered more new, unique relationships.

Also, rather than have the contact database be user-centric, it might be nice to have the output centered on the individual represented by the record being displayed. IOW, grandfather, father, son are three unique individuals. When we highlight father, then grandfather should appear as father, and father should appear as self. If we highlight grandfather, grandfather should appear as self, father should appear as son, and son should appear as grandson.

Can we accurately model the familial dataset with RDBMS?

V Harris Received on Sat Jul 10 2004 - 00:21:07 CEST