I have been performing normalisation of a dataset and was wondering if
and where I should introduce a unique identifier for a table.
Basically the table consists of 3 attributes: Item type (unique), item
description and item expert.
Whilst Item type is unique it is not a code and is made up of entries
such as: 'Small copper widget' and 'medium steel widget', not exactly
data entry proof. Therefore I want to introduce a new attribute
called item code and make it the new primary key for this table which
will consist of something much easier to input such as A1, A2, B3
(etc.) or 001, 002, 003, etc. This would dramatically reduce the
liklihood of data entry errors.
At what point of the normalisation process can I introduce this new PK
(if at all), this is for a uni assignment so it needs to be the
technically correct point if there is one.