Hybrid Refining Approach of PrOnto Ontology

Abstract

This paper presents a refinement of PrOnto ontology using a validation test based on legal experts’ annotation of privacy policies combined with an Open Knowledge Extraction (OKE) algorithm. To ensure robustness of the results while preserving an interdisciplinary approach, the integration of legal and technical knowledge has been carried out as follows. The set of privacy policies was first analysed by the legal experts to discover legal concepts and map the text into PrOnto. The mapping was then provided to computer scientists to perform the OKE analysis. Results were validated by the legal experts, who provided feedbacks and refinements (i.e. new classes and modules) of the ontology according to MeLOn methodology. Three iterations were performed on a set of (development) policies, and a final test using a new set of privacy policies. The results are 75,43% of detection of concepts in the policy texts and an increase of roughly 33% in the accuracy gain on the test set, using the new refined version of PrOnto enriched with SKOS-XL lexicon terms and definitions.