Re: What are the limits to the usefulness of normalization?

Oh dear, Dwaun!

I apologize for not following this newsgroup as much as I would like. I see that all of the previous answers you received to your question were not only incorrect but dangerously incorrect.

Normalization is limited to the logical model of the data. It provides no assistance for constructing a suitable conceptual model of the information, and it provides no assistance for arranging a physical schema. Neither does it hinder either of these tasks.

Sadly, vendors have (so far) failed to provide adequate independence between the logical model and physical layout of the data, and few practitioners know or can appreciate the distinction between the conceptual and logical levels of discourse. Subsequently, both vendors and practitioners frequently blame normalization for their own failings.

Even at the logical level, normalization has its limitations, which led Date and McGoveran to posit uniqueness of predicate as "A New Database Design Principle", for instance.

Normalization employs non-loss decomposition using projection to construct a valid logic system suitable for machine interpretation. Its primary focus is the removal of update anomalies by reducing redundancy. In this manner, normalization converts a conceptual information model into a logically addressable data model.

Ron Fagin proved that fifth normal form is both necessary and sufficient for addressing update anomalies using non-loss decomposition by projection. Other non-loss decompositions are possible. While we currently have no theory to suggest their use, it is at least conceivable that one day we might.

I would direct you to "C. J. Date, _Relational Database Writings 1991-1994_, Addison-Wesley, 1995" for the justification of unique predicates in "A New Database Design Principle".

I would also direct you to Fabian Pascal's latest book _Practical Issues in Database Management: A Reference for the Thinking Practitioner_, Addison-Wesley, 2000 (ISBN: 0201485559) as an excellent resource covering the practical implications of normalization and other issues.

Regards,
Bob

"Dwaun" <dwaunthomas_at_yahoo.com> wrote in message news:ca9a87de.0112170347.292d9d65_at_posting.google.com...
> Please help.
> Thanks,
> Dwaun
Received on Mon Dec 31 2001 - 21:39:00 CET