Repeating Data Groups

Repeating data groups can be defined as lists, repeating elements, or internal structures inside an attribute. This structure, although common in legacy data structures, violates first normal form and must be eliminated in an RDBMS model. An RDBMS cannot handle variable-length repeating fields because it offers no ability to subscript through arrays of this type. The entity below contains a repeating data group, �children's-names.� Repeating data groups violate first normal form, which basically states that an entity is in first normal form if each of its attributes has a single meaning and not more than one value for each instance.

Repeating data groups, as shown below, present problems when defining a database to contain the actual data. For example, after designing the EMPLOYEE entity, you are faced with the questions, �How many children's names do you need to record?� �How much space should you leave in each row in the database for the names?� and �What will you do if you have more names than remaining space?�

The following sample instance table might clarify the problem:

EMPLOYEE

emp-id	emp-name	emp-address	children's-names
E1	Tom	Berkeley	Jane
E2	Don	Berkeley	Tom, Dick, Donna
E3	Bob	Princeton	-
E4	John	New York	Lisa
E5	Carol	Berkeley	-

In order to fix the design, it is necessary to somehow remove the list of children's names from the EMPLOYEE entity. One way to do this is to add a CHILD table to contain the information about employee's children, as follows:

Once that is done, you can represent the names of the children as single entries in the CHILD table. In terms of the physical record structure for employee, this can resolve some of your questions about space allocation, and prevent wasting space in the record structure for employees who have no children or, conversely, deciding how much space to allocate for employees with families.

The following tables are the sample instance tables for the EMPLOYEE-CHILD model:

EMPLOYEE

emp-id	emp-name	emp-address
E1	Tom	Berkeley
E2	Don	Berkeley
E3	Bob	Princeton
E4	Carol	Berkeley

CHILD

emp-id	child-id	child-name
E2	C1	Tom
E2	C2	Dick
E2	C3	Donna
E4	C1	Lisa

This change makes the first step toward a normalized model; conversion to first normal form. Both entities now contain only fixed-length fields, which are easy to understand and program.