Efficient inference control for open relational queries

Abstract

We present a control mechanism for preserving confidentiality in relational databases under open queries. This mechanism is based on a reduction of costly inference control to efficient access control that has recently been developed for closed database queries. Our approach guarantees that secrets being declared in form of a confidentiality policy are not disclosed to database users even if they utilize their a priori knowledge to draw inferences. It turns out that there is no straightforward transition from the approach for closed queries to open queries. We show, however, that hiding the confidentiality policy from database users is sufficient to preserve confidentiality. Moreover, we propose an algorithmic implementation of the control mechanism. © 2010 IFIP International Federation for Information Processing.