A new algorithm to preserve sensitive frequents itemsets (APSFI)‎ in horizontal or vertical database

Abstract EN

This research aimed to preserve on privacy of sensitive information from adversaries.

We propose an Algorithm to Preserve Sensitive Frequents Itemsets (APSFI) with two ramifications to hides sensitive frequents itemsets in horizontal or vertical databases which minimize the number of database scanning processes during hiding operation.

The main approach to hide sensitive frequent itemsets is to reduce the support of each given frequents sensitive 1-itemsets to be insensitive and convert another insensitive to be sensitive in the same transaction to avoid the change of database size and transaction's nature to avoid adversaries' doubt.

The experiments of APSFI showed very encouraging results ; it excluded 91 % of database scan operations in vertical databases and 41 % in horizontal layout databases in comparison with the well-known FHSFI algorithm.

The experiments depict the APSFI tolerance for database size scalability, and its linear outperformance, from execution time aspect, in contrast with FHSFI.