Improved Privacy Preserving Algorithm Based on Association Rule Mining

Abstract

Listen

Translate article

Multi-parameters randomized perturbation was a privacy preserving algorithm based on association rule mining, and had high data mining accuracy. However, the exponential complexity of reconstructing frequent itemset support led to the decrease of algorithm efficiency. Aiming at the insufficient, the method of matrix block and set operations was used to improve the algorithm, and when calculating the inverse matrix of the transformation matrix, calculating all the elements of the inverse matrix changed to the first line elements. In the case of maintaining accuracy, the improved algorithm simplifies the process of calculating synthetic itemsets, eliminates the exponential complexity of reconstructing itemset support. Experimental results and analysis show that the improved algorithm has better execution efficiency.