Hello there. After a period of banging my head against the desk, I'm hoping someone can give me some pointers on my problem.

I'll start by stating immediately that this is part of an assignment for college. I, of course, don't want the complete answer, just some direction please. Basically, am I heading in the right direction?

I have to normalize the data to 3NF and whilst I was confident of my answer initially, I find myself stuck at assigning a foreign key to the groups that contain repeating information.

I will attach a diagram of the table, and the postion that I am stuck at.

Basically, would my choice of foreign key for the 2nd repeating group be appropriate, or is there indeed only one group of attributes that repeat. (albeit, a large group)

Thanks in advance

Here's a few hints to get you on your way:

• What's the difference between the account field and the customer fields?
• Does the price relate to the order or to some sort of item?
• Is the delivery person a field or a thing?
• If data can be worked out by totalling up other data then we calculate it rather than store it again.

Ohforf

Actually you are doing well, you're moving too fast and jumping ahead (which is quite logical, most capable IT types can produce 3NF in the first attempt). I would not try to identify FKs until you reach 3NF. So you have to slow down and do one NF at a time (if that is what is required), or just produce 3NF and ensure that it is correct within itself.

The Order containing and AccountId, and the Account containing an OrderId cannot be correct; choose one or the other and proceed from there.

I think (I am not sure) the purpose of this particular exercise in your course is for you to identify the entities that are implied but not identified at 0NF. So identify them and their attributes, upon which the identified 0NF attributes are dependent. Get the repeating groups abstracted out, so they are no longer repeating.

BTW, I have no idea why in 2009, colleges still teach this in character/text form and not in visual form. The human mind is much better at dealing with pictures, and the end result (the data model ) is a diagram, not table definition scripts. If you are visual, just draw the tables out.

How are you with this understanding?

A Simple Guide to Five Normal Forms in Relational Database Theory

Thank you for you replies. More food for thought

To respond to something Derek said - The question I was given was to "Identify the attributes for a bottom up model. Normalise the database to 3NF." There was a paper record of a clients business (a newsagents) and it was from there that the attributes should be deriven.

Ohforf
Brett's link is one of the better "cheat sheets" out there on the web (and there are more bad ones than good). However, all of them, even the good ones, use horrendous examples, it is hard for a newcomer to learn from them. Kent's cheat sheet is incomplete, but that matter for you because it is "fine" up to 3NF.

If you are doing bottom up, then start properly (your starting point is incorrect). List every paper record, and every field, explicit or implicit. You will have twice the fields you have listed in 0NF. Forget about columns.

(And of course my post 3 does not apply.)

Thanks to everyone who replied.

After several re-thinks, the course work has been submitted and I'm awaiting grading.

- Update -

Just to say thanks for you help. Finally got the graded assignment returned and was awarded a very healthy 'B'

After the post mortem with my lecturer, I found that if I had paid more attention to detail, I would have received an 'A'.

Always check and re-check your work people!

Thanks for coming back to let us know how things turned out! Even though I wasn't involved in the original discussion, it is good to know that you ended up with an answer you could live with and better still that you learned some new things in the process.

-PatP

Even better to know that us seasoned professionals working together on a college assignment are quite capable of achieving a 'B'