Data mining performance on perturbed databases: important influences on classification accuracy
by Mohammad Saad Al-Ahmadi, Peter A. Rosen, Rick L. Wilson
International Journal of Information and Computer Security (IJICS), Vol. 2, No. 1, 2008

Abstract: Data perturbation via the Generalised Additive Data Perturbation (GADP) method has been shown to be an effective technique for protecting disclosure of confidential attributes in databases. GADP is a viable internal security tool that preserves the statistical relationships in a database while hiding confidential data. Unfortunately, the potential impact of GADP on the ability of data mining tools to discover knowledge in a perturbed database has not been extensively studied. This study fills this gap with a comprehensive investigation of the impact of various factors surrounding databases, data security and data mining. Results support the notion that data perturbation techniques may reduce the ability of data mining tools to accurately find knowledge, and that there are other factors that also influence tool performance. These include the underlying structure of the knowledge to be discovered, the relationship of the tool to this so-called knowledge structure, the degree of noise in the knowledge, and the relationship of the confidential attributes to the knowledge.

Online publication date: Thu, 24-Jan-2008

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Information and Computer Security (IJICS):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com