Research On Privacy Preserving Distributed Clustering Algorithm

The thesis analyzes the study situation of data mining and clustering analysis, and points out that the current problems needed to be researched and solved in the field. On this basis, according to the combination of secure multi-party computation protocol and clustering analysis, the thesis proposes distributed clustering privacy preserving algorithm, and then establishes the multi-Agent model based on data mining privacy preserving. The main content is as follows:(1) Through comparing the key data mining techniques, such as association rules, classification and cluster analysis features and applications, and the study situation of privacy preserving in this algorithm, the thesis points out that the lack of data mining methods and the advantages and disadvantages of privacy preserving.(2) The thesis describes commonly used clustering algorithm, intensively comparing K-means clustering algorithm and analyzing the advantages and disadvantages of clustering algorithm. Focusing on the basic protocol of secure multi-party computation, it studies the theory of privacy preserving, and intensively analyzes and compares the possibility of the basic protocol of secure multi-party computation in data mining.(3) In view of existing features of data mining methods under the privacy preserving based on distributed clustering algorithm and combining the features of data mining and multi-Agent in distributed environments, it designs a distributed clustering mining model in the Agent implementation.(4) In summary, this thesis presents a clustering algorithm based on distributed privacy preserving. This algorithm not only realizes clustering data mining but also preserves private data of every sit, that is, achieves the data mining claim and the privacy preserving, thereby, the thesis reaches a win-win situation. Eventually, the thesis establishes a multi-Agent of data mining models using the advantages of Agent, which can further improve the performance of privacy preserving of data mining. The model can improve not only the algorithm to achieve the work, but also the efficiency of data mining operations.