Databases Reference
In-Depth Information
-
when data is not of a good quality, its value is quite
simply wrong. It must not be used by the business processes;
-
when data presents a certain level of reliability, its
value can be used, on condition that the processes are able to
tolerate it.
Let us take the simple example of the email address of a
client. This address has a different degree of reliability
depending on whether it has been automatically retrieved
from the internet during a publicity campaign, bought with a
marketing file, given by a teleoperator, or directly validated
by the client himself on a secure website.
What attitude should be adopted when faced with these
different degrees of address reliability? Should the least
reliable address be substituted with the most reliable one?
(This is the same as allocating a reliability scale depending
on the way in which the address was obtained). Should all
the variations be saved and the business processes
themselves be allowed to decide upon the correct address
depending on their needs?
These approaches, and perhaps others similar to them,
are all valid and depend on the case in question. Not taking
into consideration this reality which influences a number of
data could turn out to be dangerous: data has a level of
reliability that depends in particular on its source. Our
example of the email address nicely illustrates the degrees of
reliability, but is not very representative of the risk incurred
in cases of poor data use.
Let us take a second example, more acute: a company
must manage a formal classification of its financial
operations and assets in compliance with a business
regulation, in the spirit of Basel 2. This allocation cannot
always be done in an automated manner; human
intervention is often necessary. Several players are likely to
