Checking and Enhancing Information and Data Quality

Over time business changes. This implies that business processes change, the people as well as the systems. Also, a business is multi-disciplinary, which implies that there are many viewpoints for the same aspect.

This leads slowly to what is called data pollution. Can we solve data pollution? And how?

Background and Issues

There are several good reasons why to look into the data quality of the business processes. Some very often applicable reasons are:

• When a business manager wants to introduce the automatic calculation and presentation of Key Performance Indicators (KPI's).
• When a business manager decides to migrate his business process to new situation, either the migration to a new business application, or a to automate and integrate a a business process completely or even partially.
• New regulations, government or otherwise, specify when, where, and how to report about the business process and content. Not doing so, or not doing it properly might lead to fines and image damage.

When such a mentioned change is planned one of the issues is the trust in the current data quality. That has to be assessed in order that the migration to the new situation can be completed successfully. If the business manager has no, or not sufficient trust in his business data he cannot operate successfully, because he lacks the information to do so.

So establishing the data quality is an essential prerequisite for such migration.

Checking and Enhancing Data Quality Approach

In order to check and enhance the quality of data several steps are required. The main issue is how to re-engineer business requirements and retrieve the original specifications, and transform these specifications into appropriate Soll business requirements.

1. Make sure that you understand all business requirements, both process and content.
2. Reduce the semantics of the data as much as possible.
3. Create a data model of the result.
4. Check the data model, for instance:
   • Required fields;
   • Value constraints;
   • Primary keys;
   • Subset relations.
5. Enrich the data, and make sure that it meets the business requirements meta model as well as the resulting data model.

In this way you are sure that there is a trade off between costs and benefits. Automated checking of business requirements may be possible for some steps.