Policy refinement: Decomposition and operationalization for dynamic domains

We describe a method for policy refinement. The refinement process involves stages of decomposition, operationalization, deployment and re-refinement, and operates on policies expressed in a logical language flexible enough to be translated into many different enforceable policy dialects. We illustrate with examples from a coalition scenario, and describe how the stages of decomposition and operationaliztion work internally, and fit together in an interleaved fashion. Domains are represented in a logical formalization of UML diagrams. Both authorization and obligation policies are supported.