Privacy-Preserving DBSCAN Clustering Algorithm Based on Negative Database
Ming-Kun Zhang, Hucheng Liao · 2020
The negative database (NDB) is a new type of privacy protection and data security technology, which enhances the security of the data by storing the compressed form of the original data complement, thereby achieving the protection of privacy data. In practical applications, each record in the original database (DB) is usually transformed into a negative database to obtain negative database NDBsto achieve the protection of privacy data. Then, use the classification and clustering methods on the negative database are used to mine and analyze the privacy data. The DBSCAN clustering algorithm is a classical density-based clustering algorithm, and the Euclidean distance formula is one of the most commonly used distance measurement formulas in the clustering algorithm, and the DBSCAN algorithm is of no exception. However, the current Euclidean distance measurement of DBSCAN clustering algorithm is based on the distance measurement of plaintext data, so it is impossible to cluster the privacy data transformed into negative database. In this paper, we introduce a DBSCAN clustering algorithm based on the Euclidean distance formula on a negative database, which is used to complete clustering research while protecting privacy data. The experimental result showed that our algorithm achieved high clustering accuracy and effectively protected the security of privacy data by using irreversible negative database. Therefore, the algorithm we designed is very effective and feasible.