Identifying and representing the security semantics of an application

The author approaches database security from the semantic level. He identifies the need to classify outputs from multilevel secure database systems at a level which accurately reflects the contents. Specifically, he addresses the question of what really makes information classified, that is, the security semantics of an application. A multidimensional taxonomy of generic secrecy constraints is presented with examples that illustrate application-specific security semantics. Using labels to represent security semantics is shown to be ambiguous and therefore ineffective. Representing security semantics external to the database is proposed and several approaches are discussed. Finally, the use of a semantic data model, on top of a logic-based representation, is proposed to explicitly represent the security semantics of an application