Towards automatic translation of social network policies into controlled natural language

Abstract

On social networks, the storage, usage, and sharing of users data is usually regulated by privacy policies: natural language terms, in which specific actions are authorised, obliged, or denied, under some contextual conditions. Although guaranteeing degrees of readability and clarity, policies in natural language are not machine readable, thus preventing automatic controls on how the data are actually going to be used and processed by the entities that operate on them. In this paper, we propose an ontology-based approach for automatic translation of privacy statements, from natural language to a controlled natural one, to facilitate machine-readable processing. We provide a prototype implementation of the software-based translation tool, showing its effectiveness on a set of Facebook data policies.