This paper describes use cases and requirements for a privacy policy data schema. It describes problems with existing schemas in relation to these requirements (P3P 1.0, P3P 1.1 and RDFS schema for P3P). It proposes and motivates the use of an OWL schema to describe the same semantics, which fulfils all the requirements and may be used in a semantic web based privacy and identity management context. It describes the advantages which this gives to a policy evaluation engine based on such a schema and describes some of the reasoning use cases addressed in modelling the schema.

Modelling the schema using OWL appears simple at first sight, because the entire schema can be constructed with OWL-Lite predicates or using one custom predicate. However, the fact that modal logical statements must be made about data types in the schema (eg Organization x May Collect Data of type Y) makes reasoning over the typing schema challenging. The paper also looks at syntactic and semantic validation using the schema as well as extensions and modifications to the vocabulary items supported.