An Improved Algorithm of Individuation K-Anonymity for Multiple Sensitive Attributes

Abstract

At present, most of privacy preserving approaches in data publishing are applied to single sensitive attribute. However, applying single-sensitive-attribute privacy preserving techniques directly into data with multiple sensitive attributes often causes leakage of large amount of private information. This paper focuses on the privacy preserving methods in data publishing for multiple sensitive attributes. It combines data anonymous methods based on lossy join with the idea of clustering. And it proposes an improved algorithm of individuation K-anonymity for multiple sensitive attributes— $$ MSA(\alpha ,l) $$ algorithm. By setting parameters $$ \alpha $$ and $$ l $$ , it can restrain sensitive attribute values in equivalence class, to make a more balanced distribution of sensitive attributes and satisfy the demand of diversity, then this algorithm is applied to K-anonymity model. Finally, the result of experiment shows that this improved model can preserve the privacy of sensitive data, and it can also reduce the information hidden rate.