Research On Association Rules Mining Based On Privacy Preserving

Big data era is coming and the problem of privacy protection is becoming more prominent. However, the research of privacy protection is closely related to the research of trust.Because privacy and trust are mutually dependent in many environments.Different levels of trust in reality directly lead to different levels of privacy protection.The mainly security problem is the problem of privacy protection and trust in most of the network environment.And there is a great relationship between them.So it is necessary to conduct in-depth study of trust while in the study of privacy protection.The research on trust model has been more than ten years at present.It has been more mature.However, there are two problems in the selection and calculation of recommendation information in the trust model.First, recommended information depends on the choice of the recommended chain to rely on a few simple subjective conditions of the judgment.Second, in the calculation of trust,the aggregation parameters of the direct trust and indirect trust used static fixed parameters, not according to the actual situation.In view of the above problems, this paper proposes a trust model based on recommendation chain classification.The classification method was based on honesty attribute of nodes, which conld choose an effective recommendation chain on the basis of practical experience data. The recommendation information dissemination parameters were based on the information gain, which made recommendation information be more accurate. The factor of time was also considered in this model. The ability of interaction and the one of honesty were distinguished clearly. The concept of information entropy in information theory was used in the final aggregation calculation of direct trust and recommendation trust, which conld get rid of the ambiguity of the previous subjective parameter settings. And the main parameters of the model can be modified dynamically with the interaction. A new method based on Dijkstra algorithm is proposed for the selection of recommended information.Data mining is playing an important role in the era of big data.Association rule mining as a main way of data mining is already a irreplaceable role in the era of big data.Data mining has brought great benefits to people,but the question of privacy has been not solved. At present, there are mainly the following problems in the association rules mining based on privacy protection:(1)At present, the sensitive association rule hiding algorithm can not avoid the problem of the impact of non sensitive Association Rules Mining.At the same time, the hidden sensitive association rules can lead to redundant rules, or the normal association rules can not be mined.(2)Currently,there are existing problems under the protection of privacy in the association rule mining algorithm about safety, precision and algorithm overhead.For example, in order to ensure the safety of mining and lead to data distortion, there is a big problem of mining accuracy.To solve the above problem, this paper proposes a new method to hide sensitive association rules by increasing the support of front-end itemset,redundant rules generated before are filtered by use of association mode assessment methods.According to the second problem, this paper proposes a method of association rule mining based on Paillier encryption algorithm, which is based on the distributed environment.The method separates the computation and the decryption side to ensure the security of data mining.At the same time, the algorithm does not affect the accuracy of mining based on the encryption algorithm.And the computation cost of encryption and decryption is greatly reduced by the Montgomerie algorithm.