Attribute disclosure risk for k-anonymity: the case of numerical data

Abstract k -Anonymity is one of the most well-known privacy models. Internal and external attacks were discussed for this model, both focusing on categorical data. These can be seen as attribute disclosure a particular attribute. Then, p -sensitivity -diversity proposed solutions these That is, way to avoid very In paper we discuss case numerical data, show that also take place. For this, use rules detect sensitive cells in tabular data protection. Our experiments -anonymity not immune sense. We have analyzed results two different algorithms achieving -anonymity. First, MDAV provide microaggregation Second, Mondrian. fact, our surprise, number detected quite significant, there are no fundamental differences between Mondrian MDAV. describe considered, obtained. define dominance rule compliant p%-rule taking into account disclosure. conclude with an analysis directions future research.