In existing implementations of usage control, policies have been specified at the implementation level by intelligent users who understand the technical details of the systems in place. However, end users who want to protect their data are not always technical experts. So they would like to specify policies in abstract terms which could somehow be translated in a format that technical systems understand. This paper describes a generic and automated policy derivation where end users can specify their usage control requirements in structured natural language sentences, from which system-understandable technical policies are derived and deployed without further human intervention.