Nouvelle heuristique pour la protection des données sensibles lors de la fouille de motifs fréquents

Abstract

Data mining is a data analysis technique that uncovers hidden information and relationships between data. It is used to extract useful information from large amounts of unstructured data. Our study particularly focused on solving the PPDM (Privacy preserving data mining) problem in the context of mining frequent itemsets from transactional databases. The goal is to make a minimal change to the database in order not to disclose sensitive information during the process of mining frequent itemsets. In this work we are interested in the study, the implementation and the comparison of two heuristic approaches for solving the PPDM problem: the "Aggregated" approach which removes certain transactions and the new method that we propose bsd on the well-known combinatorial optimization problem of the set cover problem (SCP). The experimental study was carried out two databases: “Mushroom” and “D1”.