[BLC15a] Building an Ontology to Capitalize and Share Knowledge on Anonymization Techniques

Conférence Internationale avec comité de lecture : ECKM 2015 (16th European Conference on Knowledge Management ), September 2015, pp.122-131, Udine, Italy,

Mots clés: Domain ontology, anonymization, privacy, generalization algorithm, abstraction

Résumé: Privacy is one of the major concerns when publishing or sharing data. It refers to different forms of disclosure regarding the type of published or shared content. Identity disclosure, for instance, can occur when publishing personnel data. Privacy is guaranteed thanks to the anonymization processes applied to the data sets. Anonymization techniques include generalization, swapping, shuffling, data masking, etc. Each technique may be implemented using different algorithms. Choosing a suitable algorithm that ensures data privacy and that preserves data usefulness is a complex task that requires some knowledge on existing anonymization techniques and on their associated algorithms. We propose to build an ontology enabling the capitalization of such knowledge currently embedded in the myriad of research papers on this topic. The present paper describes, first, the ontology requirements and, second, the incremental ontology building process we have chosen. Then, it focuses on the three-step knowledge acquisition phase that we re-use at each increment. This phase leads to the elicitation of the main concepts of our ontology, of the relationships between them, and of some extensions. These results contribute to the conceptualization and formalization of the ontology. We use respectively UML class and object diagrams to describe conceptually the structure and the instances of the ontology. Moreover, we use OWL language to formalize the resulting ontology. The transition from the conceptualization to the formalization of the ontology is performed thanks to transformation rules proposed by the OMG. To demonstrate the feasibility of our knowledge acquisition phase, we apply it to the generalization technique and its nine algorithms. This anonymization technique is dedicated to tabular data. It consists in replacing data values with more general ones. Therefore, data are true but less precise. Future research will concentrate on the development of a guidance approach and an associated tool based on our ontology and helping data publishers in selecting and launching the relevant anonymization algorithms.

Equipe: isid


@inproceedings {
title="{Building an Ontology to Capitalize and Share Knowledge on Anonymization Techniques}",
author=" F. BEN FREDJ and N. Lammari and I. Comyn-Wattiau ",
booktitle="{ECKM 2015 (16th European Conference on Knowledge Management )}",
address="Udine, Italy",