Research On Privacy Preserving Algorithm Towards Data Mining

With the development of the computer processing power, database technology and Internet technology, more and more information is gathered together in digital form for storage. Data mining technology can effectively analyze the data; process information, and transform them into guidance, answers, information in decision support and some understandable knowledge. However, the phenomenon that individual or corporate private information leak has appeared frequently when applying data mining in some professional fields such as finance, communication, medicine. It becomes a pressing problem that how to prevent the leakage of privacy in the process of data mining. Therefore, the research on the technologies and algorithms towards data mining turns into one of the hot spots in data mining and information security field.In order to solve the problem and protect the privacy information from leaking, the privacy preserving algorithm towards data mining is applied on the original data set for data mining.In this thesis, the algorithms towards data mining are discussed and the focal point is the privacy preserving algorithm about frequent pattern mining. Concepts of privacy, privacy preserving and data mining are reviewed. The concepts and features of privacy preserving in data mining are analyzed. The privacy preserving algorithms towards data mining are classified in several standards and summed up according the different technologies of data processing. The criterions of evaluating the algorithms are put forward, which are taken to analyze these algorithms. On the basis of analyzing the data cleaning-based privacy preserving algorithm for frequent pattern data mining, a new algorithm for frequent patterns mining which is based on noise transaction is raised and two core problems of the algorithm are solved which are the calculation of the amount and generation of the noise transaction. Finally, an experiment is designed to verify the validity of the algorithm put forward in this thesis and to compare the time and space efficiency with data cleaning-based privacy preserving algorithm for frequent pattern data mining.The thesis has done beneficial research work on privacy preserving towards data mining.