One Concept Equals One Data Element: A Rule for Developing Data Elements - High-Performance Web Databases: Design, Development, and Deployment - page 80

Exhibit 6-6. Personnel department's melded data chunks.

EMPLOYEE FULL-TIME START DATE
EMPLOYEE PART-TIME START DATE
EMPLOYEE COOP STUDENT FIRST DAY
EMPLOYEE EXECUTIVE INITIAL STATUS DATE
EMPLOYEE DATE REINSTATED
EMPLOYEE MALE START DATE
EMPLOYEE FEMALE START DATE
EMPLOYEE HANDICAPPED START DATE
EMPLOYEE VETERAN START DATE
EMPLOYEE POSITION BEGIN DATE

rules should have the capability to be applied effectively outside any spe-
cific data modeling effort. They should also support data modeling efforts
by providing a quality check of the content of the data model. In other
words, if the data elements developed during modeling do not comply with
the data element standards or if some of the standard data elements devel-
oped do not fit with the model, a closer look at the model would be prudent.

Data element naming standards must provide guidelines in selecting
modifiers to ensure that the data element complies with the one-concept-
equals-one-data-element rule. The following example illustrates how
uncontrolled use of modifiers violates the rule and results in melded data
chunks instead of data elements.

The melded data chunks in Exhibit 6 were created to meet the needs of
an organization's personnel department. They may have been created over
a period of 10 years as the personnel information system was modified to
keep pace with the business requirements, or they may have been created
all at one time to satisfy different divisions within the personnel depart-
ment. Melded data chunks and redundant data elements are created inad-
vertently by carelessly inserting modifiers in data element names. If the
previous examples were restricted to one prime word, one modifier, and
one class word, the result — EMPLOYEE START DATE — makes it clear that
the concept is the date the employee started working for the organization.
The more modifiers a melded data chunk name has, the stronger the warn-
ing should sound that several concepts are included.

Some of those additional concepts are easier to spot than others, but they
can be uncovered during data element cleanup efforts by restricting data ele-
ment names to one modifier. Then, as large groups of related melded data
chunks are being analyzed and refined into data elements, modifiers can be
kept to a minimum. Typically during this process it becomes obvious how
the quality of the data elements is related to the quality of the data model