Selected results and related issues of confidentiality-preserving controlled interaction execution

Abstract

Controlled Interaction Execution has been developed as a security server for inference control shielding an isolated, logic-oriented information system when interacting over the time with a client by means of messages, in particular for query and transaction processing. The control aims at preserving confidentiality in a formalized sense, intuitively and simplifying rephrased as follows: Even when having (assumed) a priori knowledge, recording the interaction history, being aware of the details of the control mechanism, and unrestrictedly rationally reasoning, the client should never be able to infer the validity of any sentence declared as a potential secret in the security server’s confidentiality policy. To enforce this goal, for each of a rich variety of specific situations a dedicated censor has been designed. As far as needed, a censor distorts a functionally expected reaction message such that suitably weakened or even believably incorrect information is communicated to the client. In this article, we consider selected results of recent and ongoing work and discuss several issues for further research and development. The topics covered range from the impact of the underlying logic, whether propositional or first-order or for non-monotonic beliefs or an abstraction from any specific one, to the kind of the interactions, whether only queries or also view publishing or updates or revisions or even procedural programs.