Download as pdf or txt
Download as pdf or txt
You are on page 1of 1

CHAPTER

DQAF Concepts
5
Purpose
This chapter describes the scope of the DQAF: To define a set of measures that enable basic IT stew-
ardship of data, based on objective aspects of the dimensions of quality. The dimensions defined are
completeness, timeliness, validity, consistency, and integrity. The related concepts are objects of
measurement, assessment categories, and functions of measurement (collect, calculate, compare).

The Problem the DQAF Addresses


The DQAF was originally developed to solve a problem: establishing an approach to data quality
measurement that would work across data storage systems, provide measurements meaningful to data
consumers, and help improve data quality. This solution required establishing common vocabulary
to bridge the distance between the conceptual dimensions of data quality and measurements of spe-
cific data. This chapter describes why the DQAF was created and outlines the assumptions, defini-
tions, and governing ideas of the framework, including the dimensions of quality that it measures, the
objects that it measures, and the categories of assessment it enables.
Several years ago, I was involved in a project focused on establishing a common approach to
data quality measurement. The goal of the project was to establish a systematic, reusable approach
to data quality measurement, one that produced measurement results people could understand across
systems. But at the beginning of the project, we were not even speaking the same language. People
tended to talk about data quality measurement in one of two ways: either in reference to dimensions
of data quality as described by data quality experts or in terms of specific data quality problems they
had encountered and wanted to prevent in the future.
Both approaches make sense. If you are trying to solve a problem, it’s smart to research how other
people have solved it and bring to bear their expertise on your own situation. And if you are aware of
a risk that has been costly in the past, you would be unwise not to prevent its recurrence. However,
neither of these approaches enabled an extendable approach to data quality measurement. The first
was too abstract. It is important to understand dimensions of data quality, but these do not immedi-
ately lend themselves to enabling specific measurements. So people were saying, “We need to mea-
sure data completeness” or “We need to ensure we have accurate data,” but they could not say how to
measure completeness or accuracy. The second approach was too concrete. Measurements to detect
specific, known problems are useful for managing those specific problems, but they do not help miti-
gate risks associated with other data. In some of our systems, the second approach led to a pile of

57

You might also like