Research On Mask Algorithm Of Privacy Protection Association Rules

With the rapid development of the network,the demand for data is growing,and the era of large data is coming.However,the emergence of large data also brings about data security problems.Therefore,research on data mining technology based on privacy protection has been widely studied.In this paper,the author has analyzed MASK algorithm based on privacy protection association rules and has done the following tasks:Firstly,the author has briefly introduced such related theories as data mining,privacy protection,association rules and cryptography.Secondly,the author has introduced related theoretical knowledge of MASK algorithm and cryptography;and the author has put forward two methods of improving the first step of MASK algorithm by analyzing MASK algorithm disruption data set and frequent item sets.The first method is to disrupt the data by scrambling the data set in the first step,then to restore the disrupted data set to the original data set in the second step when getting the frequent item set,and lastly to mine the association rules between attributes.The second method is to disturb the original data set by using cryptography's matrix encryption algorithm in the first step of disrupting data set,restore the disrupted data set to the original data set in the second step of obtaining frequent item sets by using decryption algorithm,and then to mine the association rules between attributes.Experiments show that the improved algorithm can effectively improve the running time of the mining process,but there is a certain disadvantage in the privacy protection of the data.Furthermore,the author has put forward two methods of improving the second step of MASK algorithm by analyzing MASK algorithm disruption data set and frequent item sets.The first method to conduct matrix block processing to one of the involved matrix inversion in the second step of obtaining frequent item set,get the frequent item set's real supportdegree,and to mine the association rules between attributes.The second method is to conduct direct mathematical inversion formula to one of the involved matrix inversion in the second step of obtaining frequent item set,get the frequent item set's real support degree,and to mine the association rules between attributes.Experiments show that the improved algorithm can minimize the running time of the mining process to the greatest extent and ensure the privacy protection of MASK algorithm.Finally,the author summarizes the improved algorithm and expects to have a larger prospect space.