Semantic Web languages like OWL and RDFS promise to be viable means for representing metadata describing users and resources available over the Internet. Recently, interest has been raised on the use of such languages to represent individual data items contained in Personally Identifiable Information (PII), supporting fine-grained release. To achieve this goal, the informative content of a credential must be dissected into atomic components so that users can selectively single out those to be released. In this paper, we outline methodologies for taking advantage of a distributed ontology-based framework for controlled release at both policy writing and evaluation time.