Semi-Automatic Translations Of Data Privacy Policies Into Controlled Natural Languages

Natural languages (NLs) are a simple and understandable form for presenting knowledge. However, they are ambiguous and it turns out to be quite complex to process them with machines. Controlled Natural Languages (CNLs) are usually simpler versions of NLs that are obtained by restricting the grammar and vocabulary to reduce, or even eliminate, ambiguity and complexity. CNLs look informal like NLs and are easy to read and understand and can be easily be transformed into machine-readable forms. In this paper, we present NLPT for semi-automatic translation of privacy statements, from NL to a controlled natural one, to improve machine processing. To assess the performance, we experiment on a large set of Twitter data policies. Here, we consider two main aspects i) the translation of social network data privacy policy and ii) the efficiency and efficacy of the proposed system. We also perform an empirical analysis of the results and conclude that our system can be used to translate input policies effectively and efficiently.