An evolutionary computation-based privacy-preserving data mining model under a multithreshold constraint

An evolutionary computation-based privacy-preserving data mining model under a multithreshold constraint

Privacy-preserving data mining (PPDM) is a popular research topic in the data mining field. For individual information protection, it is vital to protect sensitive information during data mining procedures. Furthermore, it is also a serious offense to spill sensitive private knowledge. Recently, many PPDM data mining algorithms have been proposed to conceal sensitive items in a given database to disclose high-frequency items. These recent methods have already proven to be excellent in protecting confidential information and maintaining the integrity of the input database. All prior techniques, however, ignored a crucial problem in setting minimum support thresholds. If a sensitive itemset includes more items, it will cause it the become more likely to be found. Before performing mining processes, a fixed value of the minimum support threshold will be set. In this paper, a new concept of minimal support for solving this issue is proposed. In compliance with a given threshold function, the proposed approach would set a tighter threshold for an object containing several items. The results of the experiments show the performance of the traditional Greedy PPDM approach, Genetic algorithm (GA)-based PPDM approaches, and the proposed particle swarm optimization-based algorithm with the new minimal support function. The results show that the proposed method performs similarly to conventional algorithms and offers higher protection than previous methods.

article