Research On Privacy Protection Method Of Association Rules Based On Random Interference

Data mining is a very important method of knowledge discovery.Through it,the information and hidden rules from massive data can be found by people/public and used to create wealth.Meanwhile,some private information,government classified documents or business secrets,however,are at risk of being leaked.It is very likely to impact on people's normal life and even social harmony and stability partly.In view of the contradiction between data mining and privacy protection,experts and scholars put forward a series of privacy protection algorithms for data mining.This paper analyzed the privacy protection algorithm for association rules based on random interference.Aiming at the problems in the AOPAM,such as low operating efficiency and large consistency errors caused by the complexity of the inverse matrix of the transition probability matrix.This paper put forward the algorithm of PHTPMRC(Partially hidden transition probability matrix for recursive calls).The algorithm combines recursive call and the first row of matrix elements in the inversion of the transfer probability matrix of random interference to optimize the calculation process of frequent item-set support,and combines the Apriori algorithm to mine association rules based on the reconstruction of frequent item-set.The other algorithm named BITAPM(Bit and Partially hidden matrix)is also proposed in this paper.On the premise of using the transition probability matrix to randomly interfere with the data,this algorithm combines the method of granularity calculation It also optimizes the process of statistics and support calculation of candidate item set.In this way,the support count of item set can be easily calculated by the intersection of basic particles rather than multiple comparisons.The experimental results show that the algorithms of PHTPMRC and BITAPM not only interfere with the original data set according to the probability,but also effectively prevent the data mining workers from obtaining the privacy directly or indirectly.Moreover,the accuracy of association rule mining has been improved.In addition,the algorithms of PHTPMRC and BITAPM also reduce the time complexity and improve the operation efficiency.