Database and Expert Systems Applications

Collecting, releasing and sharing microdata about individuals is needed in some domains to support research initiatives aiming to create new valuable knowledge, by means of data mining and analysis tools. Thus, seeking individuals’ anonymity is required to guarantee their privacy prior publication. The k-anonymity by microaggregation, is a widely accepted model for data anonymization. It consists in de-associating the relationship between the identity of data subjects, i.e. individuals, and their confidential information. However, this method shows limits when dealing with real datasets. Indeed, the latter are characterized by their large number of attributes and the presence of noisy data. Thus, decreasing the information loss during the anonymization process is a compelling task to achieve. This paper aims to deal with such challenge. Doing so, we propose a microaggregation algorithm called Micro-PFSOM, based on fuzzy possibilitic clustering. The main thrust of this algorithm stands in applying an hybrid anonymization process.