Hybrid Data Publishing Based on Differential Privacy

Abstract

The advent of the information and intelligence era has led to explosive growth of data. The author proposes a hybrid data model based on differential privacy. The main content of this model is based on the study of differential privacy, processing the data through a noise mechanism, using the calculation of tuple attribute differences and noise addition, and finally constructing a mixed data model based on differential privacy through experiments. The experimental results indicate that: as the value of k increases, the clustering results tend to be optimal, verifying that clustering the original data can reduce noise addition. However, ICMD-DP anonymizes the original dataset, resulting in much higher information loss than DCKPDP and prototype algorithms.  A mixed data model based on differential privacy enables better clustering performance of the original dataset, thereby utilizing differential privacy to better protect the data.