Foundations for an Access Control Model for Privacy Preservation in Multi-Relational Association Rule Mining

Oliveira, S.R.M. and Zaiane, O.R.

Recent data mining algorithms have been designed for application domains that involve several types of objects stored in multiple relations in relational databases. This fact has motivated the increasing number of successful applications of relational data mining over recent years. On the other hand, such applications have introduced a new threat to privacy and information security since from non-sensitive data one is able to infer sensitive information, including personal information, facts or even patterns that are not supposed to be disclosed. The existing access control models adopted to successfully manage the access of information in complex systems present some limitations in the context of data mining tasks. The main reason is that such models were designed to protect the access to explicit data (e.g. tables, attributes, views, etc), whereas data mining tasks deal with the discovery of implicit data (e.g. pat- terns). In this paper, we take a first step toward an access control model for ensuring privacy in relational data mining, notably in multi-relational association rules (MRAR). In this model, users associated with different mining access levels, even using the same algorithm, are allowed to mine different sets of association rules. We provide the groundwork to build our ac- cess control model over existing technologies and discuss some directions for future work.

Cite as: Oliveira, S.R.M. and Zaiane, O.R. (2002). Foundations for an Access Control Model for Privacy Preservation in Multi-Relational Association Rule Mining. In Proc. IEEE ICDM Workshop on Privacy, Security and Data Mining (PSDM 2002), Maebashi City, Japan. CRPIT, 14. Clifton, C. and Estivill-Castro, V., Eds. ACS. 19-26.

(from crpit.com) (local if available)