The information period has brought us the explosive growth of the data on quantity and complexity, and engenders the challenging research area: Data Mining. In many cases, data are held by different organizations and located in different locations. Considering security and sensitivity of the data, holders maybe unwilling to share their data directly. How to get across the gap between data mining and data privacy to process various researches and applications, a major research direction of data mining is called Privacy-Preserving Data Mining.The research results of the association rule mining were summarized in the thesis and were made an in-depth introduction and analysis to the current prevalent privacy-preserving association rule mining algorithm.In view of the deficiencies of MASK algorithm, in this thesis the randomized response technology and the association rule mining algorithm were integrated and a Multi-parameters Randomized Disturb algorithm was proposed, which was called MRD algorithm. When the data sets were processed with different random parameters, the original data could be disturbed and hidden, and the defects of using data diturb and data hiding strategy singly were solved,and the privacy-preserving degree of the algorithm was improved effectively. On this basis, the algorithm of generating frequent items from transformed data sets was proposed. Finally, the experiments showed that when the parameters were choosen suitably, the privacy and accuracy of MRD algorithm were both better than the original algorithm. |