Spatiotemporal Data Mining And Privacy Protection For Campus Users

Big Data on Campus includes all aspects of students'learning and life.Mining the temporal and spatial attributes of campus data can provide more information for schools to support management decision-making.However,with the release of mining results,the campus users'sensitive information hidden in massive data will be exposed.At the same time,the campus big data itself also contains other user-sensitive information that needs privacy protection.How to effectively mine the spatiotemporal data of campus users and at the same time realize the privacy protection of user information is the problem that this paper studies and solves.In view of the above problems,a data mining and privacy protection data publishing scheme for campus is designed.In this paper,a method of building friendship and mining behavior features based on bipartite network is proposed,and a bipartite network is constructed according to the features of campus spatiotemporal data.The hypothesis test method is used to verify the randomness of students'co-occurrence,and the student-friend network is obtained.Combining with the data of campus users'life and study,this paper analyses the features of students' behavior and the relationship between the features.A differential privacy data publishing method based on k-ary tree is designed,and a perfect k-ary tree is constructed according to the situation of campus user data.Different Laplacian noise is added to each node based on the level of each node.Then synthesize privacy protection data.As a result,we can obtain secure and reliable release data.The experimental data is based on the spatiotemporal data of a college student.The experimental results are analyzed from three aspects:the construction of friendship and the mining of behavior features,and the effect of privacy protection.The experimental results show that the scheme can effectively mine the relationship between friends and the association of behavior features among students,and complete data publishing under the condition of ensuring data availability and user privacy security.In the mining process,hypothesis test was used to delete 46.1%of the edges in the bipartite network,and a more real network of friends was obtained,and then the relationships among students'behaviors were obtained.The published data not only protects personal privacy and security,but also reduces the accumulation of interval query noise and improves the availability of data.The research results play an important reference role in university management and decision-making.