Research On Privacy Protection Model Recommendation And Anonymous Result Evaluation Method

With the rapid development of Internet,Internet of Things,cloud computing and other technologies,global data is experiencing explosive growth.The use of machine learning,data mining and other technologies can effectively mine the hidden information contained in data,release the value of data,help managers make decisions,or help researchers conduct in-depth research on related technologies.But at the same time,these behaviors also lead to the disclosure of personal privacy data,which is difficult for ordinary individuals to detect.For the problem of personal privacy disclosure,the state has begun to legislate to regulate the abuse of data by enterprises,research institutions and other subjects.On the one hand,data must be shared and used to release value;on the other hand,personal privacy must be effectively protected.Therefore,privacy protection technology has become a necessary technology to achieve data security sharing.Among them,the anonymization technology based on the privacy protection model can effectively protect the sensitive information in the original data and reduce the risk of data re identification by anonymizing the data in the data release stage.However,the privacy protection model itself is more complex and requires higher professional background knowledge of users.Facing this practical problem,this paper conducts the following research:(1)Research on hybrid recommendation algorithm of privacy protection model based on user satisfaction.The algorithm is aimed at both users with certain privacy protection background knowledge and users without relevant background knowledge.According to the different characteristics of different users,a privacy protection model suitable for the current dataset is recommended based on user satisfaction.This paper analyzes the needs of different users from the perspective of practical application scenarios.Combined with these requirements,an anonymous result evaluation model based on user satisfaction is established.Then,through forward process algorithm and reverse process algorithm,collaborative filtering algorithm based on data attribute characteristics and user satisfaction is applied to recommend different users.Experiments and applications show that the recommendation algorithm can greatly improve the efficiency of data anonymization and reduce the time and energy spent by users.(2)Research on privacy model bidirectional recommendation algorithm for data value optimization.Because the needs of data users and data publishers are inconsistent,data users expect the data to be as effective as possible,compared with the data publisher’s expectation that the risk of data re identification should be as low as possible.Therefore,this paper studies the constraint relationship between data re identification risk and data utility,and establishes a risk constraint model.Based on this model,from the perspective of the forward process and the reverse recommendation process,considering the scenario that the resource pool of the historical configuration scheme continues to expand,the improved k-means clustering algorithm(K-Means algorithm)is used to optimize and recommend the collaborative filtering algorithm.Experiments and applications show that the privacy model configuration recommended by this algorithm can meet the requirements of data re-identification risk while maximizing data availability and effectively preserving data value.(3)Design and implementation of data anonymous computing system.In order to meet the needs of data publishers and data users for data privacy security,data sharing,data value release and other aspects,a data anonymization computing system is designed and implemented based on anonymization technology,data anonymity result evaluation technology and privacy model recommendation technology described in this paper,taking national laws and regulations as the criterion and user needs as the premise.The system takes anonymization technology as the core,supplemented by automated recommendation method of privacy model,to provide different types of users with simple and fast operation methods and high-quality and efficient anonymization processes.Use the data anonymity result evaluation technology to evaluate the quality of data anonymization,and issue an evaluation report to provide the basis for users and privacy model automatic recommendation methods to optimize the anonymity quality.Practice shows that the system provides a clear and efficient solution to the contradiction between data security and data sharing.