Along with the rapid development of data mining, the privacy preservationproblem has been put on the agenda, in order to guarantee the accuracy of data mining and reach a higher level of privacy preservation. Privacy-preservation mining gradually becomes an important research consideration in the field of distributed data mining. This paper focuses on vertically partitioned data, proposes a k-means clustering method using secure multi-party computation and homomorphic encryption technique based on relevant research in this field. The method makes multiple parties collaboratively conduct kmeans clustering, without disclosing private data to each other. It utilizes homomorphic encryption keys to provide protection for private data and combine with Secure Multi-party Computation protocol. Theoretical analysis shows that this method can achieve effective privacy protection and a low calculation overhead.
展开▼
