The normal forms from 1NF and upwards have well known definitions in papers by Codd and others.

I have sometimes seen the term 0NF or zeroth normal form used for a data set which is not in first normal form, but the definition seem to vary. Two definitions I have seen:

  • A data set which cannot be viewed as a valid relation, e.g. no primary key, duplicates, varying number of columns.
  • A valid relation (primary key, no duplicates, unambiguous column names and so forth) which is not in first normal form due to non-simple domains on some attributes.

Is there some commonly accepted definition, or is 0NF just an imprecise term?

  • Never really looked at it as a separate concept, just the lack of any normalization according to the accepted normal forms, like lack of light = darkness. Perhaps someone else can ahem, enlighten us. Jun 9 at 0:28
  • In the literature on the subject, the term Non First Normal Form has been used for relations with attributes with repeating or complex values.
    – Renzo
    Jun 9 at 9:07
  • @Renzo: So I guess NFNF is equivalent to the second definition: Valid relations (unique, unordered etc.) which allows relation-valued attributes and therefore does not conform to 1NF.
    – JacquesB
    Jun 9 at 9:18
  • Yes, that was the terminology used. A typical situation was that of attributes having as value another relation. So the idea was to remain in the context of the relational data model (i.e. no duplicates, etc.), just seeing them as an “extension” of the classical model. And some of the first works on object-oriented data models for databases sometimes where considered as a further extension.
    – Renzo
    Jun 9 at 11:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.