Deduction and algorithmic approaches to reason about risk, privacy and security in multi-agent systems

Recently there has been an increasing interest in privacy and its control in multi-agent systems. The need to share information and the desire to keep it private are two competing concepts, in some cases even in conflict, which affect the multi-agent systems, in particular collaboration systems. The main problem that I deal with is the protection of security in multi-agent systems. I propose different approaches, which are all related to each other. The first approach is an algorithmic one, used for ensuring a multi-agent system from different attacks performed by its members. I later introduce a deductive approach, which permits one to perform risk analysis. The deductive approach is composed by a tableau system that is based on a framework for representing and reasoning about risk. This tableau system uses the causal relations between events. For improving the risk analysis I introduce a logic with its rewriting system that finds plausible courses of events (that are events related to each other by different causal relations).