On Reasoning about Access to Knowledge

Controlling access to knowledge plays a crucial role in many multi-agent systems. In- deed, it is related to different central aspects in interactions among agents such as privacy, security, and cooperation. In this paper, we propose a framework for dealing with access to knowledge that is based on the inference process in classical propositional logic: an agent has access to every piece of knowledge that can be derived from the available knowledge using the classical inference process. We first introduce a basic problem in which an agent has to hide pieces of knowledge, and we show that this problem can be solved through the computation of maximal consistent subsets. In the same way, we also propose a coun- terpart of the previous problem in which an agent has to share pieces of knowledge, and we show that this problem can be solved through the computation of minimal inconsis- tent subsets. Then, we propose a generalization of the previous problem where an agent has to share pieces of knowledge and hide at the same time others. In this context, we introduce several concepts that allow capturing interesting aspects. Finally, we propose a weight-based approach by associating integers with the pieces of knowledge that have to be shared or hidden.