Database Reference
In-Depth Information
Figure 4-1.
Normal Forms
The
normalization procedure
involves decomposing relations into other relations of
repeatedly higher normal forms. The process is reversible. Moreover, normalization must
be conducted in a manner that ensures that there is no loss of information.
4.4 Functional Dependence and Non-loss
Decomposition
Before discussion of the normal forms, we need to define and clarify two fundamental
concepts:
functional dependence
and
non-loss decomposition
.
4.4.1 Functional Dependence
Given a relation, R{A, B, C, ...}, then attribute B is
functionally dependent
on attribute
A, written A → B (read as “A determines B”) if and only if (denoted
iff
from this point
onwards) each value of A in R has precisely one B-value in R at any point in time.
Attributes A and B may or may not be composite.
An alternate way to describe functional dependence (FD) is as follows: Given a value
of attribute A, one can deduce a value for attribute B since any two tuples which agree on
A must necessarily agree on B.
Example 1:
From definition of primary key, all attributes of a relation are functionally dependent on
the primary key. This is precisely what is required; in fact, an attribute (or group of attributes)
qualifies as a candidate key iff all other attributes of the entity are dependent on it.
