A dynamic method for handling the inference problem in multilevel secure databases

The inference problem exits when users can infer sensitive data classified at higher security levels from the knowledge of data at their level by performing inference. Inference problems greatly compromise database security, especially in multilevel secure (MLS) databases where both users and data are classified into different security levels. This paper presents an approach of dynamic control over the inference problem after all inference channels have been identified in a multilevel database. A set of key schemes are used for this purpose. We prove that these schemes are more efficient, in both space and time complexity, than previously proposed approaches.