Towards Preference-Constrained k-Anonymisation

Abstract. In this paper, we propose a novel preference-constrained approach to k-anonymisation. In contrast to the existing works on kanonymisation which attempt to satisfy a minimum level of protection requirement as a constraint and then optimise data utility within that constraint, we allow data owners and users to specify their detailed protection and usage requirements as a set of preferences on attributes or data values, treat such preferences as constraints and solve them as a multi-objective optimisation problem. This ensures that anonymised data will be actually useful to data users in their applications and sufficiently protected for data owners. Our preliminary experiments show that our method is capable of producing anonymisations that satisfy a range of preferences and have a high level of data utility and protection.

• Type: Conference
• Year: 2009
• Where: DASFAA
• Authors: Grigorios Loukides, Achilles Tziatzios, Jianhua Shao
• Keywords: DASFAA 2009, Data Owners, Data Utility, Database, Multi-objective Optimisation Problem