Towards flexible K-anonymity

Abstract

Data published online nowadays needs a high level of privacy to gain confidentiality as well as to maintain the privacy laws. The focus on k-anonymity enhancements along the last decade, allows this method to be elected as the starting point of any research. In this paper we focus on the external anonymization through a new method: the « Flexible k-anonymity ». It aims to anonymize external published data, by defining a semantic ontology that distinguishes between sparse and abundant quasi-identifiers, and describes aggregation levels relations, in order to achieve adequate k-blocks. For the validation of our proposal, we apply the aforementioned anonymization method to the Comiqual dataset. Comiqual (Collaborative measurement of internet quality), is a large-scale measurement platform for assessing the internet quality access of mobile and ADSL users by collecting mobility traces and private data related to internet metric values.