A Novel Secure Vector Product for Protecting the Privacy of Data in Vertically Partitioned Dataset

Abstract

AbstractThe contemporary advancements in the field of data collecting, storage and networking have resulted in data being saved and kept at many locations. Data mining methods allow hidden knowledge to be extracted from massive data sources. Data on the centralized server are not always present. Data might be obtained at individual or several places. Data are divided vertically, horizontally or at both sites and disseminated between places. The sharing of data repositories existing at many stages in the mining process leads to discomfort in data privacy. Data mining protection is a technology that prevents sensitive information from being disclosed to others during a mining operation. Simple multiparty calculations and anonymization are the most common privacy protecting technology for data mining. Secure multiparty (SMC) computing allows parties to do distributed data mining activities in separate places without disclosing more personal data. The SMC protocol employs mostly encryptions that offer high security and enhance the complexity of protocols. Before it is utilized for mining, perturbation alters the original sensitive data. Distortion of the data may diminish the mining performance accuracy to decrease the usefulness. A novel vector scalar product is suggested to save execution time and to protect privacy data privacy in the vertically partitioned dataset, retaining the induction of decision tree. A downward approach to the safe scalar product is taken. Each node is a radix or fewer party groups. The binary vector is divided as well. Each group discovers a safe sum of binary vectors with the secret sharing of Shamir. The running time of the CPU was established between sequences and when running downstream. The execution time was determined. The proposed bottom-up approach results in comparison with following sequential approach.KeywordsClassificationCryptographyDecision treePrivacy preserving data miningVector productSMC