Archives of the TeradataForum

Message Posted: Wed, 23 Apr 2003 @ 18:52:44 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

The passing of Dr. Codd, whom I had the privilege of meeting many years ago, got me to thinking about this whole normal form business. There is a story out there that Codd coined the phrase "normalizing relations" in 1972 because he didn't have a good name for the process of squeezing redundancy out of schemas. Codd never used the word "table," rather, he called a set of tuples a relation. As the myth goes, he felt that if Nixon could normalize relations with China, then he could normalize relations too.

I always thought this was a charming story, but I never really believed it. Codd was a mathematician, and he would have been quite familiar with the phrase "normal form" which is used in a variety of different ways in mathematics. In general, it refers to a way of representing objects so that, although each may have many different names, every possible name corresponds to exactly one object. For example, the term "normal form" is used in linear algebra to describe matrices that have been transformed into certain special forms (e.g., Hermite normal form and Smith normal form), and I believe this is where Codd applied the concept.

Also, it's used in logic to describe statements formulated in a standard way involving so-called literals (e.g., conjunctive normal form and disjunctive normal form), and in the theory of special functions to mean the uniquely-determined holonomic function (i.e., solution of a linear homogenous ordinary differential equation with polynomial coefficients) of lowest order up to multiplication by polynomials.

Wow, that was a mouthful, I can't believe I used to understand that stuff.

Anyone else have an opinion on this? Or Codd anecdotes to share?

-NR

Neil Raden
President
Hired Brains, Inc.
www.hiredbrains.com

	<Prev	Next>		<<First	<Prev	Next>	Last>>