A Clustering Method for Privacy Protection of Multidimensional Sensitive Attributes Based on Data Association

With the development of network information technology, data exchange and information sharing bring huge challenges to information privacy protection. This paper conducts a detailed study on the privacy and security issues of multi - dimensional attributes of sensitive data, and proposes a privacy - data protection scheme based on data correlation. First, the Mahalanobis distance is used to calculate the correlation of data attributes, ensuring that the anonymized data can resist background - knowledge attacks. In addition, regarding the selection of centroids in the clustering algorithm, a centroid - selection scheme during the clustering process is presented to improve the rationality of the algorithm. Experiments prove that the privacy - protection method for multiple sensitive attributes based on data correlation proposed in this paper has a lower loss degree than the traditional K - means algorithm. Moreover, in terms of attribute leakage, it can efficiently and systematically avoid and reduce the threat of attribute leakage.