Spatio-temporal Trajectory Features Based Sensitive Relationship Protection

In recent years,with the rapid development of social network services and mobile smart applications,the social network data has shown an explosive growth.As a result,there is a risk of privacy leakage during data publishing and sharing,which leads to threats to the personal and property security of users.Therefore,the sensitive relationship privacy protection,as an important privacy protection issue in social networks,has received great attention of relevant state departments and has become a hot spot in the field of database research.In existing social relationship prediction and privacy protection research,a series of problems arise,such as simple inference mechanism and impractical privacy protection methods.In this paper,a privacy protection algorithm for sensitive relationships based on spatial-temporal trajectory features in social network is proposed.In the algorithm,we devise a method through swapping user's check-in data to prevent the leakage of sensitive relationships.Firstly,it elaborates on the inference of sensitive relationships and defines the related knowledge,including the link inference of sensitive relations,the calculation method of user similarity based on spatio-temporal trajectory features and the information loss function,which laid the foundation for the implementation of the privacy-sensitive sensitive relations based on spatio-temporal trajectory features and the implementation of protection technologies.Secondly,the spatio-temporal trajectory features based sensitive relationship protection algorithm is proposed,for protecting sensitive relationships in social networks.A heuristic is designed based on the inference contribution and information loss of the data swapping,through which a secure inferred social network graph with high data utility is obtained in this algorithm.Extensive experiments on real datasets and comparisons with the existing methods demonstrate the practicality and high efficiency of the algorithm are proposed in this paper and the high utility of the published social networks.