Parallel based Hiding of Sensitive Knowledge
Panteleimon Krasadakis, Vassilios S. Verykios, Evangelos Sakkopoulos · 2020
Nowadays, privacy preserving data mining is an increasingly popular field of data mining research. frequent itemset-based approaches comprise a large category of techniques where the hiding is exemplified through the Apriori algorithm. This paper extends the work on a previously presented hiding scheme that sanitizes the input database by extending it based on the ideal positive border that is computed through the application of a constraint-based frequent itemset mining algorithm. The goal of the proposed approach is to improve at large on computational aspects of the hiding methodology, and in particular to accommodate bigger datasets by customizing the hiding scheme and allowing it to run in parallel while ensuring the hiding of the sensitive knowledge in its entirety. The proposed approach is a significant acceleration factor provided to business user directly in the form of Software as a Service.