Research And Application On Privacy Protection For Data Mining

In recent years, with the rise of commercial enterprises, the Internet ushered in the era of the outbreak of data, people can easily obtain the information and services by network. Data mining technology can turn the data without rule into available data in the environment of big data.Then these available data can be effectively applied to various aspects. However, the data mining technology in the application process, will inevitably contact with the user’s data, which makes users worry about the security of their own data. Then, users would not want to share some useful data information with others,which limits the effect of data mining techniques. As a result, the application of privacy protection method based on data mining technology can effectively solve the security problems of data, and the security of data mining technology has become a hot issue of research.At first, this thesis analyzes the current situation and shortage of privacy protection technology for data mining. Then, it improves the privacy protection technology to make it more applicable to data mining.For the anonymous model of privacy protection technology, this thesis proposes an improved algorithm of individuation K-anonymity for multiple sensitive attributes. By setting parameters ?and l, it can restrain sensitive attribute values in equivalence class, to make a more balanced distribution of sensitive attributes and satisfy the demand of diversity, then this algorithm is applied to K-anonymity model. The result of experiment shows that this improved model can preserve the privacy of sensitive data, and it can also reduce the information hidden rate.For randomized method of privacy protection technology, this thesis combines privacy protection technology with trust evaluation model. This model uses the method of ERRPH(Extended Randomized Response with Partial Hidding) method. This method can protect privacy effectively and do not affect the judgment of trust. When making trust decisions, this method can ensure the privacy and accuracy of data. The thesis also combines with calculation based on the Bayesian theory in the trust decision. Through simulation, it proves that this model have effect of privacy protection. This model is also a more effective trust evaluation methods.For encryption method of privacy protection technology, this thesis comes up with amulti-keyword search scheme based on privacy protection. The program combines data retrieval with similarity with encryption technology, to improve the efficiency of data retrieval and protect the security of data.Finally, this thesis designs a system for data distribution, which based on theoretical models that is proposed in this thesis. The system can realize the functions of publishing data, protecting privacy and so on.