Research On Privacy Protection For Group Recommendation

With the popularization and development of recommendation systems,group recommendation has gradually become a research hotspot in the recommendation field.The group recommendation system needs to collect a large amount of user history data for recommendation,but these data may contain user sensitive information,and there is a risk of leaking user privacy.In addition,due to the unique group user communication and preference sharing features unique to the group recommendation system,this process may present a risk of leaking user privacy.Therefore,how to ensure that user privacy is not leaked into a privacy protection research problem for group recommendation in the process of implementing group recommendation and user preference sharing and user communication.In view of the above problems,this thesis has conducted in-depth research on the privacy protection methods for group recommendation.The main work is as follows:Firstly,a personalized privacy protection framework based on trusted clients for group recommendation is proposed,aiming at the existing group recommendation system fails to meet the personalized privacy protection needs of users.Based on this framework,a privacy protection method of group sensitive preference is proposed.This method collects the historical data and privacy preference requirements of users in the group on the trusted client.For users with privacy protection needs,similar users in the group are found by using the similarity of user sensitive topics.Through random collaborative disturbance of the first k users,personalized privacy protection of users in the group is realized.Simulation and comparison experiments show that the personalized privacy protection method proposed in this thesis can meet the privacy needs of different users and has better performance.Secondly,in view of the possibility of leaking user privacy in the process of preference sharing and user communication in the group recommendation system,a privacy protection method for secure communication of group users is proposed.In this method,the user nodes in the group are used as buffer nodes.When the k-anonymity condition is satisfied,the buffer nodes will disturb the communication data to protect the privacy of users.Simulation and comparison experiments show that the privacy protection method for group user communication proposed in this thesis can realize data sharing of group users and ensure that the privacy of group users will not be leaked due to theincrease of published data.Finally,based on the above theories and methods,this thesis constructs a prototype system of user privacy protection for group recommendation and gives an application demonstration.Prototype system based on group found,user communication and group preference share,recommendation and user privacy protection and other functional requirements analysis and design,to demonstrate the application according to the traditional software development process,to verify the proposed in this thesis the feasibility of the method and theory,shows the group oriented user privacy protection method recommended by group tourism scenarios in the actual application effect.