Data Incremental Clustering Algorithm based on Differential Privacy

In the information era, the issues of data security and availability are more prominent. Normal cluster analysis algorithms mostly rely on the clustering of static datasets, which leads to the problem of high time cost, and can not effectively protect users' privacy. In response to the above two issues, we propose an incremental clustering algorithm called ICDP that satisfies differential privacy protection. Firstly, we compare the Pearson correlation coefficient and the threshold between the input sample points, adaptively introduce noise to the clustering results based on clustering errors, and then further compress the data through the FP-Growth structure in data mining, solving the problems of information overload and inadequate protection of sensitive information in the era of big data. Finally, experimental results demonstrate that compared to existing clustering algorithms, ICDP algorithm significantly reduces the runtime while ensuring data availability and security under the same level of privacy protection.