Re: A Normalization Question
Date: Wed, 30 Jun 2004 14:57:46 -0400
Message-ID: <2kggoeF24agdU1_at_uni-berlin.de>
"Neo" <neo55592_at_hotmail.com> wrote in message
news:4b45d3ad.0406300959.36b2b9d2_at_posting.google.com...
> > THINGS
> > thing_tin (pk),
> > name,
> > vendor_tin (fk),
> > etc (fk)
> >
> > VENDOR_ATTRIBUTES
> > vendor_tin (pk/fk), ...
> >
> > ETC_ATTRIBUTES
> > ...
>
> If one wanted to store two persons named brown whose color was also
> brown, how could one use the above schema and avoid redundancy?
That's like asking how one would store information about airline flights in the above scehema. You wouldn't. Anyway...
"Brown" is an instance of an attribute. Actually, two different attributes:
Attribute 1 = last_name
Attribute 2 = color of person
You could normalize to a higher NF to prevent storing the value of either attribute more than once. This is done is real estate databases to save space:
STREETS
street_id
street_name
HOUSES
house_number
street_id
etc
The idea here is that the street_id will _usually_ take less space on disk than the street name, so it is not necessary to record the street name for each house- just the street id. Received on Wed Jun 30 2004 - 20:57:46 CEST